This is an archive of the discontinued LLVM Phabricator instance.

Differential D37388

[X86] Allow cross-lane permutations for sub targets supporting AVX2.
ClosedPublic

Authored by jbhateja on Sep 1 2017, 11:27 AM.

Details

Reviewers
aymanmus
craig.topper
RKSimon
spatel
guyblank
modocache
Commits
rG2c139f77c773: [X86] Allow cross-lane permutations for sub targets supporting AVX2.
rL312608: [X86] Allow cross-lane permutations for sub targets supporting AVX2.
Summary

Most instructions in AVX work “in-lane”, that is, each source element is applied only to other
elements of the same lane, thus a cross lane permutation is costly and needs more than one instrution.
AVX2 includes instructions to perform any-to-any permutation of words over a 256-bit register
and vectorized table lookup.

This should also Fix PR34369

Diff Detail

Event Timeline

jbhateja created this revision.Sep 1 2017, 11:27 AM
Harbormaster completed remote builds in B9845: Diff 113560.Sep 1 2017, 11:27 AM
jbhateja added reviewers: aymanmus, craig.topper, RKSimon, spatel, guyblank.Sep 1 2017, 11:48 AM
jbhateja added a subscriber: llvm-commits.
jbhateja added a comment.Sep 3 2017, 12:08 AM

Ping reviewers.

guyblank edited edge metadata.Sep 3 2017, 1:06 AM

Hi Jatin,
Thanks for working on this.
You didn't add llvm-commits as a subscriber upon review creation, so this won't show up on the mailing list. Can you open a new review?

jbhateja added a comment.Sep 3 2017, 3:06 AM
In D37388#859841, @guyblank wrote:

Hi Jatin,
Thanks for working on this.
You didn't add llvm-commits as a subscriber upon review creation, so this won't show up on the mailing list. Can you open a new review?

Hi Guy,

llvm-commits is already on subscriber's list for this differential revision. I did not not get your point in opeining a new review.

Thanks

guyblank added a comment.Sep 3 2017, 4:22 AM
In D37388#859847, @jbhateja wrote:
In D37388#859841, @guyblank wrote:

Hi Jatin,
Thanks for working on this.
You didn't add llvm-commits as a subscriber upon review creation, so this won't show up on the mailing list. Can you open a new review?

Hi Guy,

llvm-commits is already on subscriber's list for this differential revision. I did not not get your point in opeining a new review.

Thanks

If you want the llvm-commits mailing list to be notified when you create a review, you need to add it as a subscriber when creating the review, not afterwards.
When you add it after the initial creation, there will be no mail notification to the list.
So in your case, the first message the list got was the 'ping'.
If llvm-commits was added upon creation, the first message would have been:
jatin created this revision.
<the description of the revision>

Which I think is useful for people browsing the list.

jbhateja added a comment.Sep 3 2017, 4:29 AM

Oh Ok , thanks I shall take care in future.

@reviewers , kindly pour in your comment.

RKSimon edited edge metadata.Sep 4 2017, 4:30 AM

I've added the test case for PR34369 at rL312481 - please can you rebase?

lib/Target/X86/X86ISelLowering.cpp
12079

Update this comment to explain the AVX2 path

jbhateja updated this revision to Diff 113782.Sep 4 2017, 11:27 AM
Herald added a reviewer: modocache. · View Herald TranscriptSep 4 2017, 11:27 AM
jbhateja updated this revision to Diff 113783.Sep 4 2017, 11:34 AM
  • Reverting lit script change.
aymanmus edited edge metadata.Sep 5 2017, 1:16 AM

The patch shows impressive results.
LGTM, unless @RKSimon have other comments.

RKSimon accepted this revision.Sep 5 2017, 4:15 AM

LGTM with a few minors to fix before commit

lib/Target/X86/X86ISelLowering.cpp
12071

clang-format ? (80-column limit)

12082

half of the lanes are not

This revision is now accepted and ready to land.Sep 5 2017, 4:15 AM
jbhateja updated this revision to Diff 113949.Sep 5 2017, 7:53 PM
  • Formatting changes
Harbormaster completed remote builds in B9926: Diff 113949.Sep 5 2017, 7:53 PM
Closed by commit rL312608: [X86] Allow cross-lane permutations for sub targets supporting AVX2. (authored by jbhateja). · Explain WhySep 5 2017, 8:01 PM
This revision was automatically updated to reflect the committed changes.

Revision Contents

Diff 113560

lib/Target/X86/X86ISelLowering.cpp

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 12,061 Lines • ▼ Show 20 Lines
/// from a permuted copy of the vector. This lowering strategy results in four /// from a permuted copy of the vector. This lowering strategy results in four
/// instructions in the worst case for a single-input cross lane shuffle which /// instructions in the worst case for a single-input cross lane shuffle which
/// is lower than any other fully general cross-lane shuffle strategy I'm aware /// is lower than any other fully general cross-lane shuffle strategy I'm aware
/// of. Special cases for each particular shuffle pattern should be handled /// of. Special cases for each particular shuffle pattern should be handled
/// prior to trying this lowering. /// prior to trying this lowering.
static SDValue lowerVectorShuffleAsLanePermuteAndBlend(const SDLoc &DL, MVT VT, static SDValue lowerVectorShuffleAsLanePermuteAndBlend(const SDLoc &DL, MVT VT,
SDValue V1, SDValue V2, SDValue V1, SDValue V2,
ArrayRef<int> Mask, ArrayRef<int> Mask,
SelectionDAG &DAG) { SelectionDAG &DAG,
const X86Subtarget &Subtarget) {
RKSimonUnsubmitted
Not Done
ReplyInline Actions

clang-format ? (80-column limit)

RKSimon: clang-format ? (80-column limit)
// FIXME: This should probably be generalized for 512-bit vectors as well. // FIXME: This should probably be generalized for 512-bit vectors as well.
assert(VT.is256BitVector() && "Only for 256-bit vector shuffles!"); assert(VT.is256BitVector() && "Only for 256-bit vector shuffles!");
int Size = Mask.size(); int Size = Mask.size();
int LaneSize = Size / 2; int LaneSize = Size / 2;
// If there are only inputs from one 128-bit lane, splitting will in fact be // If there are only inputs from one 128-bit lane, splitting will in fact be
// less expensive. The flags track whether the given lane contains an element // less expensive. The flags track whether the given lane contains an element
// that crosses to another lane. // that crosses to another lane.
RKSimonUnsubmitted
Not Done
ReplyInline Actions

Update this comment to explain the AVX2 path

RKSimon: Update this comment to explain the AVX2 path
if (!Subtarget.hasAVX2()) {
bool LaneCrossing[2] = {false, false}; bool LaneCrossing[2] = {false, false};
for (int i = 0; i < Size; ++i) for (int i = 0; i < Size; ++i)
RKSimonUnsubmitted
Not Done
ReplyInline Actions

half of the lanes are not

RKSimon: half of the lanes are not
if (Mask[i] >= 0 && (Mask[i] % Size) / LaneSize != i / LaneSize) if (Mask[i] >= 0 && (Mask[i] % Size) / LaneSize != i / LaneSize)
LaneCrossing[(Mask[i] % Size) / LaneSize] = true; LaneCrossing[(Mask[i] % Size) / LaneSize] = true;
if (!LaneCrossing[0] || !LaneCrossing[1]) if (!LaneCrossing[0] || !LaneCrossing[1])
return splitAndLowerVectorShuffle(DL, VT, V1, V2, Mask, DAG); return splitAndLowerVectorShuffle(DL, VT, V1, V2, Mask, DAG);
} else {
bool LaneUsed[2] = {false, false};
for (int i = 0; i < Size; ++i)
if (Mask[i] >= 0)
LaneUsed[(Mask[i] / LaneSize)] = true;
if (!LaneUsed[0] || !LaneUsed[1])
return splitAndLowerVectorShuffle(DL, VT, V1, V2, Mask, DAG);
}
assert(V2.isUndef() && assert(V2.isUndef() &&
"This last part of this routine only works on single input shuffles"); "This last part of this routine only works on single input shuffles");
SmallVector<int, 32> FlippedBlendMask(Size); SmallVector<int, 32> FlippedBlendMask(Size);
for (int i = 0; i < Size; ++i) for (int i = 0; i < Size; ++i)
FlippedBlendMask[i] = FlippedBlendMask[i] =
Mask[i] < 0 ? -1 : (((Mask[i] % Size) / LaneSize == i / LaneSize) Mask[i] < 0 ? -1 : (((Mask[i] % Size) / LaneSize == i / LaneSize)
▲ Show 20 Lines • Show All 598 Lines • ▼ Show 20 Lines if (V2.isUndef()) {
// Try to create an in-lane repeating shuffle mask and then shuffle the // Try to create an in-lane repeating shuffle mask and then shuffle the
// the results into the target lanes. // the results into the target lanes.
if (SDValue V = lowerShuffleAsRepeatedMaskAndLanePermute( if (SDValue V = lowerShuffleAsRepeatedMaskAndLanePermute(
DL, MVT::v4f64, V1, V2, Mask, Subtarget, DAG)) DL, MVT::v4f64, V1, V2, Mask, Subtarget, DAG))
return V; return V;
// Otherwise, fall back. // Otherwise, fall back.
return lowerVectorShuffleAsLanePermuteAndBlend(DL, MVT::v4f64, V1, V2, Mask, return lowerVectorShuffleAsLanePermuteAndBlend(DL, MVT::v4f64, V1, V2, Mask,
DAG); DAG, Subtarget);
} }
// Use dedicated unpack instructions for masks that match their pattern. // Use dedicated unpack instructions for masks that match their pattern.
if (SDValue V = if (SDValue V =
lowerVectorShuffleWithUNPCK(DL, MVT::v4f64, Mask, V1, V2, DAG)) lowerVectorShuffleWithUNPCK(DL, MVT::v4f64, Mask, V1, V2, DAG))
return V; return V;
if (SDValue Blend = lowerVectorShuffleAsBlend(DL, MVT::v4f64, V1, V2, Mask, if (SDValue Blend = lowerVectorShuffleAsBlend(DL, MVT::v4f64, V1, V2, Mask,
▲ Show 20 Lines • Show All 186 Lines • ▼ Show 20 Lines if (V2.isUndef()) {
if (!is128BitLaneCrossingShuffleMask(MVT::v8f32, Mask)) if (!is128BitLaneCrossingShuffleMask(MVT::v8f32, Mask))
return DAG.getNode(X86ISD::VPERMILPV, DL, MVT::v8f32, V1, VPermMask); return DAG.getNode(X86ISD::VPERMILPV, DL, MVT::v8f32, V1, VPermMask);
if (Subtarget.hasAVX2()) if (Subtarget.hasAVX2())
return DAG.getNode(X86ISD::VPERMV, DL, MVT::v8f32, VPermMask, V1); return DAG.getNode(X86ISD::VPERMV, DL, MVT::v8f32, VPermMask, V1);
// Otherwise, fall back. // Otherwise, fall back.
return lowerVectorShuffleAsLanePermuteAndBlend(DL, MVT::v8f32, V1, V2, Mask, return lowerVectorShuffleAsLanePermuteAndBlend(DL, MVT::v8f32, V1, V2, Mask,
DAG); DAG, Subtarget);
} }
// Try to simplify this by merging 128-bit lanes to enable a lane-based // Try to simplify this by merging 128-bit lanes to enable a lane-based
// shuffle. // shuffle.
if (SDValue Result = lowerVectorShuffleByMerging128BitLanes( if (SDValue Result = lowerVectorShuffleByMerging128BitLanes(
DL, MVT::v8f32, V1, V2, Mask, Subtarget, DAG)) DL, MVT::v8f32, V1, V2, Mask, Subtarget, DAG))
return Result; return Result;
// If we have VLX support, we can use VEXPAND. // If we have VLX support, we can use VEXPAND.
▲ Show 20 Lines • Show All 184 Lines • ▼ Show 20 Lines if (SDValue V = lowerShuffleAsRepeatedMaskAndLanePermute(
DL, MVT::v16i16, V1, V2, Mask, Subtarget, DAG)) DL, MVT::v16i16, V1, V2, Mask, Subtarget, DAG))
return V; return V;
if (V2.isUndef()) { if (V2.isUndef()) {
// There are no generalized cross-lane shuffle operations available on i16 // There are no generalized cross-lane shuffle operations available on i16
// element types. // element types.
if (is128BitLaneCrossingShuffleMask(MVT::v16i16, Mask)) if (is128BitLaneCrossingShuffleMask(MVT::v16i16, Mask))
return lowerVectorShuffleAsLanePermuteAndBlend(DL, MVT::v16i16, V1, V2, return lowerVectorShuffleAsLanePermuteAndBlend(DL, MVT::v16i16, V1, V2,
Mask, DAG); Mask, DAG, Subtarget);
SmallVector<int, 8> RepeatedMask; SmallVector<int, 8> RepeatedMask;
if (is128BitLaneRepeatedShuffleMask(MVT::v16i16, Mask, RepeatedMask)) { if (is128BitLaneRepeatedShuffleMask(MVT::v16i16, Mask, RepeatedMask)) {
// As this is a single-input shuffle, the repeated mask should be // As this is a single-input shuffle, the repeated mask should be
// a strictly valid v8i16 mask that we can pass through to the v8i16 // a strictly valid v8i16 mask that we can pass through to the v8i16
// lowering to handle even the v16 case. // lowering to handle even the v16 case.
return lowerV8I16GeneralSingleInputVectorShuffle( return lowerV8I16GeneralSingleInputVectorShuffle(
DL, MVT::v16i16, V1, RepeatedMask, Subtarget, DAG); DL, MVT::v16i16, V1, RepeatedMask, Subtarget, DAG);
▲ Show 20 Lines • Show All 68 Lines • ▼ Show 20 Linesstatic SDValue lowerV32I8VectorShuffle(const SDLoc &DL, ArrayRef<int> Mask,
if (SDValue V = lowerShuffleAsRepeatedMaskAndLanePermute( if (SDValue V = lowerShuffleAsRepeatedMaskAndLanePermute(
DL, MVT::v32i8, V1, V2, Mask, Subtarget, DAG)) DL, MVT::v32i8, V1, V2, Mask, Subtarget, DAG))
return V; return V;
// There are no generalized cross-lane shuffle operations available on i8 // There are no generalized cross-lane shuffle operations available on i8
// element types. // element types.
if (V2.isUndef() && is128BitLaneCrossingShuffleMask(MVT::v32i8, Mask)) if (V2.isUndef() && is128BitLaneCrossingShuffleMask(MVT::v32i8, Mask))
return lowerVectorShuffleAsLanePermuteAndBlend(DL, MVT::v32i8, V1, V2, Mask, return lowerVectorShuffleAsLanePermuteAndBlend(DL, MVT::v32i8, V1, V2, Mask,
DAG); DAG, Subtarget);
if (SDValue PSHUFB = lowerVectorShuffleWithPSHUFB( if (SDValue PSHUFB = lowerVectorShuffleWithPSHUFB(
DL, MVT::v32i8, Mask, V1, V2, Zeroable, Subtarget, DAG)) DL, MVT::v32i8, Mask, V1, V2, Zeroable, Subtarget, DAG))
return PSHUFB; return PSHUFB;
// Try to simplify this by merging 128-bit lanes to enable a lane-based // Try to simplify this by merging 128-bit lanes to enable a lane-based
// shuffle. // shuffle.
if (SDValue Result = lowerVectorShuffleByMerging128BitLanes( if (SDValue Result = lowerVectorShuffleByMerging128BitLanes(
▲ Show 20 Lines • Show All 23,686 Lines • Show Last 20 Lines

test/CodeGen/X86/vector-shuffle-256-v16.ll

Show First 20 Lines • Show All 1,837 Lines • ▼ Show 20 Lines
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] ; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,0,2,4,5,6,7] ; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,0,2,4,5,6,7]
; AVX1-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,4,7] ; AVX1-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,4,7]
; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,0,1,1] ; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_00_01_00_01_02_03_02_11_08_09_08_09_10_11_10_11: ; AVX2-LABEL: shuffle_v16i16_00_01_00_01_02_03_02_11_08_09_08_09_10_11_10_11:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] ; AVX2-NEXT: vpbroadcastq %xmm1, %xmm1
; AVX2OR512VL-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,0,2,4,5,6,7] ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[0,0,1,1,4,4,5,5]
; AVX2OR512VL-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,4,7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2OR512VL-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,0,1,1] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0 ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: retq ;
; AVX512VL-LABEL: shuffle_v16i16_00_01_00_01_02_03_02_11_08_09_08_09_10_11_10_11:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [0,1,0,1,2,3,2,11,8,9,8,9,10,11,10,11]
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 1, i32 0, i32 1, i32 2, i32 3, i32 2, i32 11, i32 8, i32 9, i32 8, i32 9, i32 10, i32 11, i32 10, i32 11> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 1, i32 0, i32 1, i32 2, i32 3, i32 2, i32 11, i32 8, i32 9, i32 8, i32 9, i32 10, i32 11, i32 10, i32 11>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_06_07_04_05_02_03_00_09_14_15_12_13_10_11_08_09(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_06_07_04_05_02_03_00_09_14_15_12_13_10_11_08_09(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_06_07_04_05_02_03_00_09_14_15_12_13_10_11_08_09: ; AVX1-LABEL: shuffle_v16i16_06_07_04_05_02_03_00_09_14_15_12_13_10_11_08_09:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3,4,5,6,7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3,4,5,6,7]
; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[3,2,1,0] ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[3,2,1,0]
; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[3,2,1,0] ; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[3,2,1,0]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_06_07_04_05_02_03_00_09_14_15_12_13_10_11_08_09: ; AVX2-LABEL: shuffle_v16i16_06_07_04_05_02_03_00_09_14_15_12_13_10_11_08_09:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3,4,5,6,7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2OR512VL-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[3,2,1,0] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[3,2,1,0] ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0 ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: retq ;
; AVX512VL-LABEL: shuffle_v16i16_06_07_04_05_02_03_00_09_14_15_12_13_10_11_08_09:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [6,7,4,5,2,3,0,9,14,15,12,13,10,11,8,9]
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 6, i32 7, i32 4, i32 5, i32 2, i32 3, i32 0, i32 9, i32 14, i32 15, i32 12, i32 13, i32 10, i32 11, i32 8, i32 9> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 6, i32 7, i32 4, i32 5, i32 2, i32 3, i32 0, i32 9, i32 14, i32 15, i32 12, i32 13, i32 10, i32 11, i32 8, i32 9>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_04_05_06_07_16_17_18_27_12_13_14_15_24_25_26_27(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_04_05_06_07_16_17_18_27_12_13_14_15_24_25_26_27(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_04_05_06_07_16_17_18_27_12_13_14_15_24_25_26_27: ; AVX1-LABEL: shuffle_v16i16_04_05_06_07_16_17_18_27_12_13_14_15_24_25_26_27:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm3 ; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm3
; AVX1-NEXT: vpalignr {{.*#+}} xmm2 = xmm2[8,9,10,11,12,13,14,15],xmm3[0,1,2,3,4,5,6,7] ; AVX1-NEXT: vpalignr {{.*#+}} xmm2 = xmm2[8,9,10,11,12,13,14,15],xmm3[0,1,2,3,4,5,6,7]
; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3] ; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3]
; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,4,5,4,5,6,7,0,1,4,5,8,9,14,15] ; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,4,5,4,5,6,7,0,1,4,5,8,9,14,15]
; AVX1-NEXT: vpunpckhqdq {{.*#+}} xmm0 = xmm0[1],xmm1[1] ; AVX1-NEXT: vpunpckhqdq {{.*#+}} xmm0 = xmm0[1],xmm1[1]
; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2-LABEL: shuffle_v16i16_04_05_06_07_16_17_18_27_12_13_14_15_24_25_26_27: ; AVX2-LABEL: shuffle_v16i16_04_05_06_07_16_17_18_27_12_13_14_15_24_25_26_27:
; AVX2: # BB#0: ; AVX2: # BB#0:
; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3],ymm1[4,5],ymm0[6,7] ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3],ymm1[4,5],ymm0[6,7]
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[3],xmm0[4,5,6,7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[2,3,0,1] ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
; AVX2-NEXT: retq ; AVX2-NEXT: retq
; ;
; AVX512VL-LABEL: shuffle_v16i16_04_05_06_07_16_17_18_27_12_13_14_15_24_25_26_27: ; AVX512VL-LABEL: shuffle_v16i16_04_05_06_07_16_17_18_27_12_13_14_15_24_25_26_27:
; AVX512VL: # BB#0: ; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [4,5,6,7,16,17,18,27,12,13,14,15,24,25,26,27] ; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [4,5,6,7,16,17,18,27,12,13,14,15,24,25,26,27]
; AVX512VL-NEXT: vpermt2w %ymm1, %ymm2, %ymm0 ; AVX512VL-NEXT: vpermt2w %ymm1, %ymm2, %ymm0
; AVX512VL-NEXT: retq ; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 27, i32 12, i32 13, i32 14, i32 15, i32 24, i32 25, i32 26, i32 27> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 27, i32 12, i32 13, i32 14, i32 15, i32 24, i32 25, i32 26, i32 27>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_00_00_00_00_00_00_00_08_08_08_08_08_08_08_08_08(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_00_00_00_00_00_00_00_08_08_08_08_08_08_08_08_08(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_00_00_00_00_00_00_00_08_08_08_08_08_08_08_08_08: ; AVX1-LABEL: shuffle_v16i16_00_00_00_00_00_00_00_08_08_08_08_08_08_08_08_08:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] ; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,2,3] ; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,2,3]
; AVX1-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,0,0,0,4,5,6,7] ; AVX1-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,0,0,0,4,5,6,7]
; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,0,1,1] ; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_00_00_00_00_00_00_00_08_08_08_08_08_08_08_08_08: ; AVX2-LABEL: shuffle_v16i16_00_00_00_00_00_00_00_08_08_08_08_08_08_08_08_08:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] ; AVX2-NEXT: vpbroadcastw %xmm1, %xmm1
; AVX2OR512VL-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,2,3] ; AVX2-NEXT: vpshuflw {{.*#+}} ymm0 = ymm0[0,0,0,0,4,5,6,7,8,8,8,8,12,13,14,15]
; AVX2OR512VL-NEXT: vpbroadcastw %xmm1, %xmm1 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[0,0,1,1,4,4,5,5]
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0 ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2OR512VL-NEXT: retq ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2-NEXT: retq
;
; AVX512VL-LABEL: shuffle_v16i16_00_00_00_00_00_00_00_08_08_08_08_08_08_08_08_08:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [0,0,0,0,0,0,0,8,8,8,8,8,8,8,8,8]
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_00_00_00_00_04_04_04_12_08_08_08_08_12_12_12_12(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_00_00_00_00_04_04_04_12_08_08_08_08_12_12_12_12(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_00_00_00_00_04_04_04_12_08_08_08_08_12_12_12_12: ; AVX1-LABEL: shuffle_v16i16_00_00_00_00_04_04_04_12_08_08_08_08_12_12_12_12:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpsllq $48, %xmm1, %xmm2 ; AVX1-NEXT: vpsllq $48, %xmm1, %xmm2
; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,0,0,0,4,5,6,7] ; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,0,0,0,4,5,6,7]
; AVX1-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,4,7] ; AVX1-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,4,7]
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
; AVX1-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,0,0,0,4,5,6,7] ; AVX1-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,0,0,0,4,5,6,7]
; AVX1-NEXT: vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,4,4,4] ; AVX1-NEXT: vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,4,4,4]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_00_00_00_00_04_04_04_12_08_08_08_08_12_12_12_12: ; AVX2-LABEL: shuffle_v16i16_00_00_00_00_04_04_04_12_08_08_08_08_12_12_12_12:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vpsllq $48, %xmm1, %xmm2 ; AVX2-NEXT: vpsllq $48, %xmm1, %xmm1
; AVX2OR512VL-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,0,0,0,4,5,6,7] ; AVX2-NEXT: vpshuflw {{.*#+}} ymm0 = ymm0[0,0,0,0,4,5,6,7,8,8,8,8,12,13,14,15]
; AVX2OR512VL-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,4,7] ; AVX2-NEXT: vpshufhw {{.*#+}} ymm0 = ymm0[0,1,2,3,4,4,4,4,8,9,10,11,12,12,12,12]
; AVX2OR512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2OR512VL-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,0,0,0,4,5,6,7] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,4,4,4] ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0 ;
; AVX2OR512VL-NEXT: retq ; AVX512VL-LABEL: shuffle_v16i16_00_00_00_00_04_04_04_12_08_08_08_08_12_12_12_12:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [0,0,0,0,4,4,4,12,8,8,8,8,12,12,12,12]
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 0, i32 0, i32 0, i32 4, i32 4, i32 4, i32 12, i32 8, i32 8, i32 8, i32 8, i32 12, i32 12, i32 12, i32 12> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 0, i32 0, i32 0, i32 4, i32 4, i32 4, i32 12, i32 8, i32 8, i32 8, i32 8, i32 12, i32 12, i32 12, i32 12>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_uu_00_uu_01_uu_02_uu_11_uu_08_uu_09_uu_10_uu_11(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_uu_00_uu_01_uu_02_uu_11_uu_08_uu_09_uu_10_uu_11(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_uu_00_uu_01_uu_02_uu_11_uu_08_uu_09_uu_10_uu_11: ; AVX1-LABEL: shuffle_v16i16_uu_00_uu_01_uu_02_uu_11_uu_08_uu_09_uu_10_uu_11:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm2 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] ; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm2 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] ; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,0,2,2,4,5,6,7] ; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,0,2,2,4,5,6,7]
; AVX1-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,6,7] ; AVX1-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,6,7]
; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_uu_00_uu_01_uu_02_uu_11_uu_08_uu_09_uu_10_uu_11: ; AVX2-LABEL: shuffle_v16i16_uu_00_uu_01_uu_02_uu_11_uu_08_uu_09_uu_10_uu_11:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vpunpcklwd {{.*#+}} xmm2 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = <255,255,255,255,255,255,0,0,u,u,u,u,u,u,u,u,255,255,255,255,255,255,255,255,u,u,u,u,u,u,u,u>
; AVX2OR512VL-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,0,2,2,4,5,6,7] ; AVX2-NEXT: vpunpcklwd {{.*#+}} ymm0 = ymm0[0,0,1,1,2,2,3,3,8,8,9,9,10,10,11,11]
; AVX2OR512VL-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,6,7] ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm2, %ymm0, %ymm0 ;
; AVX2OR512VL-NEXT: retq ; AVX512VL-LABEL: shuffle_v16i16_uu_00_uu_01_uu_02_uu_11_uu_08_uu_09_uu_10_uu_11:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = <u,0,u,1,u,2,u,11,u,8,u,9,u,10,u,11>
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 undef, i32 0, i32 undef, i32 1, i32 undef, i32 2, i32 undef, i32 11, i32 undef, i32 8, i32 undef, i32 9, i32 undef, i32 10, i32 undef, i32 11> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 undef, i32 0, i32 undef, i32 1, i32 undef, i32 2, i32 undef, i32 11, i32 undef, i32 8, i32 undef, i32 9, i32 undef, i32 10, i32 undef, i32 11>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_uu_04_uu_05_uu_06_uu_15_uu_12_uu_13_uu_14_uu_15(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_uu_04_uu_05_uu_06_uu_15_uu_12_uu_13_uu_14_uu_15(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_uu_04_uu_05_uu_06_uu_15_uu_12_uu_13_uu_14_uu_15: ; AVX1-LABEL: shuffle_v16i16_uu_04_uu_05_uu_06_uu_15_uu_12_uu_13_uu_14_uu_15:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm2 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7] ; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm2 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm0 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7] ; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm0 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,0,2,2,4,5,6,7] ; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,0,2,2,4,5,6,7]
; AVX1-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,6,7] ; AVX1-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,6,7]
; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_uu_04_uu_05_uu_06_uu_15_uu_12_uu_13_uu_14_uu_15: ; AVX2-LABEL: shuffle_v16i16_uu_04_uu_05_uu_06_uu_15_uu_12_uu_13_uu_14_uu_15:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vpunpckhwd {{.*#+}} xmm2 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7] ; AVX2-NEXT: vpunpckhwd {{.*#+}} ymm0 = ymm0[4,4,5,5,6,6,7,7,12,12,13,13,14,14,15,15]
; AVX2OR512VL-NEXT: vpunpckhwd {{.*#+}} xmm0 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7] ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3],ymm0[4,5,6,7]
; AVX2OR512VL-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,0,2,2,4,5,6,7] ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,6,7] ;
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm2, %ymm0, %ymm0 ; AVX512VL-LABEL: shuffle_v16i16_uu_04_uu_05_uu_06_uu_15_uu_12_uu_13_uu_14_uu_15:
; AVX2OR512VL-NEXT: retq ; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = <u,4,u,5,u,6,u,15,u,12,u,13,u,14,u,15>
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 undef, i32 4, i32 undef, i32 5, i32 undef, i32 6, i32 undef, i32 15, i32 undef, i32 12, i32 undef, i32 13, i32 undef, i32 14, i32 undef, i32 15> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 undef, i32 4, i32 undef, i32 5, i32 undef, i32 6, i32 undef, i32 15, i32 undef, i32 12, i32 undef, i32 13, i32 undef, i32 14, i32 undef, i32 15>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_03_01_02_00_06_07_04_13_11_09_10_08_14_15_12_13(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_03_01_02_00_06_07_04_13_11_09_10_08_14_15_12_13(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_03_01_02_00_06_07_04_13_11_09_10_08_14_15_12_13: ; AVX1-LABEL: shuffle_v16i16_03_01_02_00_06_07_04_13_11_09_10_08_14_15_12_13:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4],xmm1[5],xmm0[6,7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4],xmm1[5],xmm0[6,7]
; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[3,1,2,0,4,5,6,7] ; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[3,1,2,0,4,5,6,7]
; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,1,3,2] ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,1,3,2]
; AVX1-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[3,1,2,0,4,5,6,7] ; AVX1-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[3,1,2,0,4,5,6,7]
; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,1,3,2] ; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,1,3,2]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_03_01_02_00_06_07_04_13_11_09_10_08_14_15_12_13: ; AVX2-LABEL: shuffle_v16i16_03_01_02_00_06_07_04_13_11_09_10_08_14_15_12_13:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4],xmm1[5],xmm0[6,7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2OR512VL-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[3,1,2,0,4,5,6,7] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,1,3,2] ; AVX2-NEXT: vpshuflw {{.*#+}} ymm0 = ymm0[3,1,2,0,4,5,6,7,11,9,10,8,12,13,14,15]
; AVX2OR512VL-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[3,1,2,0,4,5,6,7] ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[0,1,3,2,4,5,7,6]
; AVX2OR512VL-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,1,3,2] ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0 ;
; AVX2OR512VL-NEXT: retq ; AVX512VL-LABEL: shuffle_v16i16_03_01_02_00_06_07_04_13_11_09_10_08_14_15_12_13:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [3,1,2,0,6,7,4,13,11,9,10,8,14,15,12,13]
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 3, i32 1, i32 2, i32 0, i32 6, i32 7, i32 4, i32 13, i32 11, i32 9, i32 10, i32 8, i32 14, i32 15, i32 12, i32 13> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 3, i32 1, i32 2, i32 0, i32 6, i32 7, i32 4, i32 13, i32 11, i32 9, i32 10, i32 8, i32 14, i32 15, i32 12, i32 13>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_04_04_04_04_00_00_00_08_12_12_12_12_08_08_08_08(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_04_04_04_04_00_00_00_08_12_12_12_12_08_08_08_08(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_04_04_04_04_00_00_00_08_12_12_12_12_08_08_08_08: ; AVX1-LABEL: shuffle_v16i16_04_04_04_04_00_00_00_08_12_12_12_12_08_08_08_08:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpslldq {{.*#+}} xmm2 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm1[0,1] ; AVX1-NEXT: vpslldq {{.*#+}} xmm2 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm1[0,1]
; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8,9,8,9,8,9,8,9,0,1,0,1,0,1,14,15] ; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8,9,8,9,8,9,8,9,0,1,0,1,0,1,14,15]
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[8,9,8,9,8,9,8,9,0,1,0,1,0,1,0,1] ; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[8,9,8,9,8,9,8,9,0,1,0,1,0,1,0,1]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_04_04_04_04_00_00_00_08_12_12_12_12_08_08_08_08: ; AVX2-LABEL: shuffle_v16i16_04_04_04_04_00_00_00_08_12_12_12_12_08_08_08_08:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vpbroadcastw %xmm1, %xmm2 ; AVX2-NEXT: vpbroadcastw %xmm1, %xmm1
; AVX2OR512VL-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8,9,8,9,8,9,8,9,0,1,0,1,0,1,14,15] ; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[8,9,8,9,8,9,8,9,0,1,0,1,0,1,0,1,24,25,24,25,24,25,24,25,16,17,16,17,16,17,16,17]
; AVX2OR512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2OR512VL-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[8,9,8,9,8,9,8,9,0,1,0,1,0,1,0,1] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0 ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: retq ;
; AVX512VL-LABEL: shuffle_v16i16_04_04_04_04_00_00_00_08_12_12_12_12_08_08_08_08:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [4,4,4,4,0,0,0,8,12,12,12,12,8,8,8,8]
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 4, i32 4, i32 4, i32 4, i32 0, i32 0, i32 0, i32 8, i32 12, i32 12, i32 12, i32 12, i32 8, i32 8, i32 8, i32 8> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 4, i32 4, i32 4, i32 4, i32 0, i32 0, i32 0, i32 8, i32 12, i32 12, i32 12, i32 12, i32 8, i32 8, i32 8, i32 8>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_02_03_00_01_06_07_04_13_10_11_08_09_14_15_12_13(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_02_03_00_01_06_07_04_13_10_11_08_09_14_15_12_13(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_02_03_00_01_06_07_04_13_10_11_08_09_14_15_12_13: ; AVX1-LABEL: shuffle_v16i16_02_03_00_01_06_07_04_13_10_11_08_09_14_15_12_13:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4],xmm1[5],xmm0[6,7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4],xmm1[5],xmm0[6,7]
; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,0,3,2] ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,0,3,2]
; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[1,0,3,2] ; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[1,0,3,2]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_02_03_00_01_06_07_04_13_10_11_08_09_14_15_12_13: ; AVX2-LABEL: shuffle_v16i16_02_03_00_01_06_07_04_13_10_11_08_09_14_15_12_13:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4],xmm1[5],xmm0[6,7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2OR512VL-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,0,3,2] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[1,0,3,2] ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[1,0,3,2,5,4,7,6]
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0 ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: retq ;
; AVX512VL-LABEL: shuffle_v16i16_02_03_00_01_06_07_04_13_10_11_08_09_14_15_12_13:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [2,3,0,1,6,7,4,13,10,11,8,9,14,15,12,13]
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 2, i32 3, i32 0, i32 1, i32 6, i32 7, i32 4, i32 13, i32 10, i32 11, i32 8, i32 9, i32 14, i32 15, i32 12, i32 13> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 2, i32 3, i32 0, i32 1, i32 6, i32 7, i32 4, i32 13, i32 10, i32 11, i32 8, i32 9, i32 14, i32 15, i32 12, i32 13>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_02_03_00_02_06_07_04_13_10_11_08_10_14_15_12_13(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_02_03_00_02_06_07_04_13_10_11_08_10_14_15_12_13(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_02_03_00_02_06_07_04_13_10_11_08_10_14_15_12_13: ; AVX1-LABEL: shuffle_v16i16_02_03_00_02_06_07_04_13_10_11_08_10_14_15_12_13:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4],xmm1[5],xmm0[6,7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4],xmm1[5],xmm0[6,7]
; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[2,3,0,2,4,5,6,7] ; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[2,3,0,2,4,5,6,7]
; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,1,3,2] ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,1,3,2]
; AVX1-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[2,3,0,2,4,5,6,7] ; AVX1-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[2,3,0,2,4,5,6,7]
; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,1,3,2] ; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,1,3,2]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_02_03_00_02_06_07_04_13_10_11_08_10_14_15_12_13: ; AVX2-LABEL: shuffle_v16i16_02_03_00_02_06_07_04_13_10_11_08_10_14_15_12_13:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4],xmm1[5],xmm0[6,7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = <255,255,u,u,255,255,255,255,255,255,0,0,255,255,255,255,255,255,u,u,255,255,255,255,255,255,255,255,255,255,255,255>
; AVX2OR512VL-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[2,3,0,2,4,5,6,7] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,1,3,2] ; AVX2-NEXT: vpshuflw {{.*#+}} ymm0 = ymm0[2,3,0,2,4,5,6,7,10,11,8,10,12,13,14,15]
; AVX2OR512VL-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[2,3,0,2,4,5,6,7] ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[0,1,3,2,4,5,7,6]
; AVX2OR512VL-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,1,3,2] ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0 ;
; AVX2OR512VL-NEXT: retq ; AVX512VL-LABEL: shuffle_v16i16_02_03_00_02_06_07_04_13_10_11_08_10_14_15_12_13:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [2,3,0,2,6,7,4,13,10,11,8,10,14,15,12,13]
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 2, i32 3, i32 0, i32 2, i32 6, i32 7, i32 4, i32 13, i32 10, i32 11, i32 8, i32 10, i32 14, i32 15, i32 12, i32 13> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 2, i32 3, i32 0, i32 2, i32 6, i32 7, i32 4, i32 13, i32 10, i32 11, i32 8, i32 10, i32 14, i32 15, i32 12, i32 13>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_02_03_00_01_06_07_04_15_10_11_08_09_14_15_12_15(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_02_03_00_01_06_07_04_15_10_11_08_09_14_15_12_15(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_02_03_00_01_06_07_04_15_10_11_08_09_14_15_12_15: ; AVX1-LABEL: shuffle_v16i16_02_03_00_01_06_07_04_15_10_11_08_09_14_15_12_15:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,0,3,2] ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,0,3,2]
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm1[7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm1[7]
; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[1,0,2,3] ; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[1,0,2,3]
; AVX1-NEXT: vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,6,7,4,7] ; AVX1-NEXT: vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,6,7,4,7]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_02_03_00_01_06_07_04_15_10_11_08_09_14_15_12_15: ; AVX2-LABEL: shuffle_v16i16_02_03_00_01_06_07_04_15_10_11_08_09_14_15_12_15:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,0,3,2] ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[1,0,2,3,5,4,6,7]
; AVX2OR512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm1[7] ; AVX2-NEXT: vpshufhw {{.*#+}} ymm0 = ymm0[0,1,2,3,6,7,4,7,8,9,10,11,14,15,12,15]
; AVX2OR512VL-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[1,0,2,3] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2OR512VL-NEXT: vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,6,7,4,7] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0 ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: retq ;
; AVX512VL-LABEL: shuffle_v16i16_02_03_00_01_06_07_04_15_10_11_08_09_14_15_12_15:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [2,3,0,1,6,7,4,15,10,11,8,9,14,15,12,15]
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 2, i32 3, i32 0, i32 1, i32 6, i32 7, i32 4, i32 15, i32 10, i32 11, i32 8, i32 9, i32 14, i32 15, i32 12, i32 15> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 2, i32 3, i32 0, i32 1, i32 6, i32 7, i32 4, i32 15, i32 10, i32 11, i32 8, i32 9, i32 14, i32 15, i32 12, i32 15>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_07_05_06_04_03_01_02_08_15_13_14_12_11_09_10_08(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_07_05_06_04_03_01_02_08_15_13_14_12_11_09_10_08(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_07_05_06_04_03_01_02_08_15_13_14_12_11_09_10_08: ; AVX1-LABEL: shuffle_v16i16_07_05_06_04_03_01_02_08_15_13_14_12_11_09_10_08:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vmovdqa {{.*#+}} xmm2 = [14,15,10,11,12,13,8,9,6,7,2,3,4,5,0,1] ; AVX1-NEXT: vmovdqa {{.*#+}} xmm2 = [14,15,10,11,12,13,8,9,6,7,2,3,4,5,0,1]
; AVX1-NEXT: vpshufb %xmm2, %xmm1, %xmm3 ; AVX1-NEXT: vpshufb %xmm2, %xmm1, %xmm3
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm1[0],xmm0[1,2,3,4,5,6,7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm1[0],xmm0[1,2,3,4,5,6,7]
; AVX1-NEXT: vpshufb %xmm2, %xmm0, %xmm0 ; AVX1-NEXT: vpshufb %xmm2, %xmm0, %xmm0
; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_07_05_06_04_03_01_02_08_15_13_14_12_11_09_10_08: ; AVX2-LABEL: shuffle_v16i16_07_05_06_04_03_01_02_08_15_13_14_12_11_09_10_08:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vmovdqa {{.*#+}} xmm2 = [14,15,10,11,12,13,8,9,6,7,2,3,4,5,0,1] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2OR512VL-NEXT: vpshufb %xmm2, %xmm1, %xmm3 ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm1[0],xmm0[1,2,3,4,5,6,7] ; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[14,15,10,11,12,13,8,9,6,7,2,3,4,5,0,1,30,31,26,27,28,29,24,25,22,23,18,19,20,21,16,17]
; AVX2OR512VL-NEXT: vpshufb %xmm2, %xmm0, %xmm0 ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm3, %ymm0, %ymm0 ;
; AVX2OR512VL-NEXT: retq ; AVX512VL-LABEL: shuffle_v16i16_07_05_06_04_03_01_02_08_15_13_14_12_11_09_10_08:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [7,5,6,4,3,1,2,8,15,13,14,12,11,9,10,8]
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 7, i32 5, i32 6, i32 4, i32 3, i32 1, i32 2, i32 8, i32 15, i32 13, i32 14, i32 12, i32 11, i32 9, i32 10, i32 8> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 7, i32 5, i32 6, i32 4, i32 3, i32 1, i32 2, i32 8, i32 15, i32 13, i32 14, i32 12, i32 11, i32 9, i32 10, i32 8>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_01_00_05_04_05_04_01_08_09_08_13_12_13_12_09_08(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_01_00_05_04_05_04_01_08_09_08_13_12_13_12_09_08(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_01_00_05_04_05_04_01_08_09_08_13_12_13_12_09_08: ; AVX1-LABEL: shuffle_v16i16_01_00_05_04_05_04_01_08_09_08_13_12_13_12_09_08:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpslldq {{.*#+}} xmm2 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm1[0,1] ; AVX1-NEXT: vpslldq {{.*#+}} xmm2 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm1[0,1]
; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[2,3,0,1,10,11,8,9,10,11,8,9,2,3,2,3] ; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[2,3,0,1,10,11,8,9,10,11,8,9,2,3,2,3]
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[2,3,0,1,10,11,8,9,10,11,8,9,2,3,0,1] ; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[2,3,0,1,10,11,8,9,10,11,8,9,2,3,0,1]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_01_00_05_04_05_04_01_08_09_08_13_12_13_12_09_08: ; AVX2-LABEL: shuffle_v16i16_01_00_05_04_05_04_01_08_09_08_13_12_13_12_09_08:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vpbroadcastw %xmm1, %xmm2 ; AVX2-NEXT: vpbroadcastw %xmm1, %xmm1
; AVX2OR512VL-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[2,3,0,1,10,11,8,9,10,11,8,9,2,3,2,3] ; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[2,3,0,1,10,11,8,9,10,11,8,9,2,3,0,1,18,19,16,17,26,27,24,25,26,27,24,25,18,19,16,17]
; AVX2OR512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2OR512VL-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[2,3,0,1,10,11,8,9,10,11,8,9,2,3,0,1] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0 ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: retq ;
; AVX512VL-LABEL: shuffle_v16i16_01_00_05_04_05_04_01_08_09_08_13_12_13_12_09_08:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [1,0,5,4,5,4,1,8,9,8,13,12,13,12,9,8]
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 1, i32 0, i32 5, i32 4, i32 5, i32 4, i32 1, i32 8, i32 9, i32 8, i32 13, i32 12, i32 13, i32 12, i32 9, i32 8> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 1, i32 0, i32 5, i32 4, i32 5, i32 4, i32 1, i32 8, i32 9, i32 8, i32 13, i32 12, i32 13, i32 12, i32 9, i32 8>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_05_04_01_00_05_04_01_08_13_12_09_08_13_12_09_08(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_05_04_01_00_05_04_01_08_13_12_09_08_13_12_09_08(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_05_04_01_00_05_04_01_08_13_12_09_08_13_12_09_08: ; AVX1-LABEL: shuffle_v16i16_05_04_01_00_05_04_01_08_13_12_09_08_13_12_09_08:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpslldq {{.*#+}} xmm2 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm1[0,1] ; AVX1-NEXT: vpslldq {{.*#+}} xmm2 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm1[0,1]
; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[10,11,8,9,2,3,0,1,10,11,8,9,2,3,2,3] ; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[10,11,8,9,2,3,0,1,10,11,8,9,2,3,2,3]
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[10,11,8,9,2,3,0,1,10,11,8,9,2,3,0,1] ; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[10,11,8,9,2,3,0,1,10,11,8,9,2,3,0,1]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_05_04_01_00_05_04_01_08_13_12_09_08_13_12_09_08: ; AVX2-LABEL: shuffle_v16i16_05_04_01_00_05_04_01_08_13_12_09_08_13_12_09_08:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vpbroadcastw %xmm1, %xmm2 ; AVX2-NEXT: vpbroadcastw %xmm1, %xmm1
; AVX2OR512VL-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[10,11,8,9,2,3,0,1,10,11,8,9,2,3,2,3] ; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[10,11,8,9,2,3,0,1,10,11,8,9,2,3,0,1,26,27,24,25,18,19,16,17,26,27,24,25,18,19,16,17]
; AVX2OR512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2OR512VL-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[10,11,8,9,2,3,0,1,10,11,8,9,2,3,0,1] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0 ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: retq ;
; AVX512VL-LABEL: shuffle_v16i16_05_04_01_00_05_04_01_08_13_12_09_08_13_12_09_08:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [5,4,1,0,5,4,1,8,13,12,9,8,13,12,9,8]
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 5, i32 4, i32 1, i32 0, i32 5, i32 4, i32 1, i32 8, i32 13, i32 12, i32 9, i32 8, i32 13, i32 12, i32 9, i32 8> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 5, i32 4, i32 1, i32 0, i32 5, i32 4, i32 1, i32 8, i32 13, i32 12, i32 9, i32 8, i32 13, i32 12, i32 9, i32 8>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_05_04_01_00_01_00_05_12_13_12_09_08_09_08_13_12(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_05_04_01_00_01_00_05_12_13_12_09_08_09_08_13_12(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_05_04_01_00_01_00_05_12_13_12_09_08_09_08_13_12: ; AVX1-LABEL: shuffle_v16i16_05_04_01_00_01_00_05_12_13_12_09_08_09_08_13_12:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpsllq $48, %xmm1, %xmm2 ; AVX1-NEXT: vpsllq $48, %xmm1, %xmm2
; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[10,11,8,9,2,3,0,1,2,3,0,1,10,11,2,3] ; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[10,11,8,9,2,3,0,1,2,3,0,1,10,11,2,3]
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[10,11,8,9,2,3,0,1,2,3,0,1,10,11,8,9] ; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[10,11,8,9,2,3,0,1,2,3,0,1,10,11,8,9]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_05_04_01_00_01_00_05_12_13_12_09_08_09_08_13_12: ; AVX2-LABEL: shuffle_v16i16_05_04_01_00_01_00_05_12_13_12_09_08_09_08_13_12:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vpsllq $48, %xmm1, %xmm2 ; AVX2-NEXT: vpsllq $48, %xmm1, %xmm1
; AVX2OR512VL-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[10,11,8,9,2,3,0,1,2,3,0,1,10,11,2,3] ; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[10,11,8,9,2,3,0,1,2,3,0,1,10,11,8,9,26,27,24,25,18,19,16,17,18,19,16,17,26,27,24,25]
; AVX2OR512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2OR512VL-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[10,11,8,9,2,3,0,1,2,3,0,1,10,11,8,9] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0 ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: retq ;
; AVX512VL-LABEL: shuffle_v16i16_05_04_01_00_01_00_05_12_13_12_09_08_09_08_13_12:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [5,4,1,0,1,0,5,12,13,12,9,8,9,8,13,12]
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 5, i32 4, i32 1, i32 0, i32 1, i32 0, i32 5, i32 12, i32 13, i32 12, i32 9, i32 8, i32 9, i32 8, i32 13, i32 12> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 5, i32 4, i32 1, i32 0, i32 1, i32 0, i32 5, i32 12, i32 13, i32 12, i32 9, i32 8, i32 9, i32 8, i32 13, i32 12>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_00_04_04_00_00_04_04_08_08_12_12_08_08_12_12_08(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_00_04_04_00_00_04_04_08_08_12_12_08_08_12_12_08(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_00_04_04_00_00_04_04_08_08_12_12_08_08_12_12_08: ; AVX1-LABEL: shuffle_v16i16_00_04_04_00_00_04_04_08_08_12_12_08_08_12_12_08:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpslldq {{.*#+}} xmm2 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm1[0,1] ; AVX1-NEXT: vpslldq {{.*#+}} xmm2 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm1[0,1]
; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,0,1,8,9,8,9,2,3] ; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,0,1,8,9,8,9,2,3]
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,8,9,8,9,0,1,0,1,8,9,8,9,0,1] ; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,8,9,8,9,0,1,0,1,8,9,8,9,0,1]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_00_04_04_00_00_04_04_08_08_12_12_08_08_12_12_08: ; AVX2-LABEL: shuffle_v16i16_00_04_04_00_00_04_04_08_08_12_12_08_08_12_12_08:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vpbroadcastw %xmm1, %xmm2 ; AVX2-NEXT: vpbroadcastw %xmm1, %xmm1
; AVX2OR512VL-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,0,1,8,9,8,9,2,3] ; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,8,9,8,9,0,1,0,1,8,9,8,9,0,1,16,17,24,25,24,25,16,17,16,17,24,25,24,25,16,17]
; AVX2OR512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2OR512VL-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,8,9,8,9,0,1,0,1,8,9,8,9,0,1] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0 ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: retq ;
; AVX512VL-LABEL: shuffle_v16i16_00_04_04_00_00_04_04_08_08_12_12_08_08_12_12_08:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [0,4,4,0,0,4,4,8,8,12,12,8,8,12,12,8]
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 4, i32 4, i32 0, i32 0, i32 4, i32 4, i32 8, i32 8, i32 12, i32 12, i32 8, i32 8, i32 12, i32 12, i32 8> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 4, i32 4, i32 0, i32 0, i32 4, i32 4, i32 8, i32 8, i32 12, i32 12, i32 8, i32 8, i32 12, i32 12, i32 8>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_04_00_00_04_04_00_00_12_12_08_08_12_12_08_08_12(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_04_00_00_04_04_00_00_12_12_08_08_12_12_08_08_12(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_04_00_00_04_04_00_00_12_12_08_08_12_12_08_08_12: ; AVX1-LABEL: shuffle_v16i16_04_00_00_04_04_00_00_12_12_08_08_12_12_08_08_12:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpsllq $48, %xmm1, %xmm2 ; AVX1-NEXT: vpsllq $48, %xmm1, %xmm2
; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8,9,0,1,0,1,8,9,8,9,0,1,0,1,2,3] ; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8,9,0,1,0,1,8,9,8,9,0,1,0,1,2,3]
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[8,9,0,1,0,1,8,9,8,9,0,1,0,1,8,9] ; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[8,9,0,1,0,1,8,9,8,9,0,1,0,1,8,9]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_04_00_00_04_04_00_00_12_12_08_08_12_12_08_08_12: ; AVX2-LABEL: shuffle_v16i16_04_00_00_04_04_00_00_12_12_08_08_12_12_08_08_12:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vpsllq $48, %xmm1, %xmm2 ; AVX2-NEXT: vpsllq $48, %xmm1, %xmm1
; AVX2OR512VL-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8,9,0,1,0,1,8,9,8,9,0,1,0,1,2,3] ; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[8,9,0,1,0,1,8,9,8,9,0,1,0,1,8,9,24,25,16,17,16,17,24,25,24,25,16,17,16,17,24,25]
; AVX2OR512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2OR512VL-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[8,9,0,1,0,1,8,9,8,9,0,1,0,1,8,9] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0 ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: retq ;
; AVX512VL-LABEL: shuffle_v16i16_04_00_00_04_04_00_00_12_12_08_08_12_12_08_08_12:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [4,0,0,4,4,0,0,12,12,8,8,12,12,8,8,12]
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 4, i32 0, i32 0, i32 4, i32 4, i32 0, i32 0, i32 12, i32 12, i32 8, i32 8, i32 12, i32 12, i32 8, i32 8, i32 12> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 4, i32 0, i32 0, i32 4, i32 4, i32 0, i32 0, i32 12, i32 12, i32 8, i32 8, i32 12, i32 12, i32 8, i32 8, i32 12>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_02_06_04_00_05_01_07_11_10_14_12_08_13_09_15_11(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_02_06_04_00_05_01_07_11_10_14_12_08_13_09_15_11(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_02_06_04_00_05_01_07_11_10_14_12_08_13_09_15_11: ; AVX1-LABEL: shuffle_v16i16_02_06_04_00_05_01_07_11_10_14_12_08_13_09_15_11:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vmovdqa {{.*#+}} xmm2 = [4,5,12,13,8,9,0,1,10,11,2,3,14,15,6,7] ; AVX1-NEXT: vmovdqa {{.*#+}} xmm2 = [4,5,12,13,8,9,0,1,10,11,2,3,14,15,6,7]
; AVX1-NEXT: vpshufb %xmm2, %xmm1, %xmm3 ; AVX1-NEXT: vpshufb %xmm2, %xmm1, %xmm3
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[3],xmm0[4,5,6,7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[3],xmm0[4,5,6,7]
; AVX1-NEXT: vpshufb %xmm2, %xmm0, %xmm0 ; AVX1-NEXT: vpshufb %xmm2, %xmm0, %xmm0
; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_02_06_04_00_05_01_07_11_10_14_12_08_13_09_15_11: ; AVX2-LABEL: shuffle_v16i16_02_06_04_00_05_01_07_11_10_14_12_08_13_09_15_11:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vmovdqa {{.*#+}} xmm2 = [4,5,12,13,8,9,0,1,10,11,2,3,14,15,6,7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2OR512VL-NEXT: vpshufb %xmm2, %xmm1, %xmm3 ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[3],xmm0[4,5,6,7] ; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[4,5,12,13,8,9,0,1,10,11,2,3,14,15,6,7,20,21,28,29,24,25,16,17,26,27,18,19,30,31,22,23]
; AVX2OR512VL-NEXT: vpshufb %xmm2, %xmm0, %xmm0 ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm3, %ymm0, %ymm0 ;
; AVX2OR512VL-NEXT: retq ; AVX512VL-LABEL: shuffle_v16i16_02_06_04_00_05_01_07_11_10_14_12_08_13_09_15_11:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [2,6,4,0,5,1,7,11,10,14,12,8,13,9,15,11]
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 2, i32 6, i32 4, i32 0, i32 5, i32 1, i32 7, i32 11, i32 10, i32 14, i32 12, i32 8, i32 13, i32 9, i32 15, i32 11> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 2, i32 6, i32 4, i32 0, i32 5, i32 1, i32 7, i32 11, i32 10, i32 14, i32 12, i32 8, i32 13, i32 9, i32 15, i32 11>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_02_00_06_04_05_01_07_11_10_08_14_12_13_09_15_11(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_02_00_06_04_05_01_07_11_10_08_14_12_13_09_15_11(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_02_00_06_04_05_01_07_11_10_08_14_12_13_09_15_11: ; AVX1-LABEL: shuffle_v16i16_02_00_06_04_05_01_07_11_10_08_14_12_13_09_15_11:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vmovdqa {{.*#+}} xmm2 = [4,5,0,1,12,13,8,9,10,11,2,3,14,15,6,7] ; AVX1-NEXT: vmovdqa {{.*#+}} xmm2 = [4,5,0,1,12,13,8,9,10,11,2,3,14,15,6,7]
; AVX1-NEXT: vpshufb %xmm2, %xmm1, %xmm3 ; AVX1-NEXT: vpshufb %xmm2, %xmm1, %xmm3
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[3],xmm0[4,5,6,7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[3],xmm0[4,5,6,7]
; AVX1-NEXT: vpshufb %xmm2, %xmm0, %xmm0 ; AVX1-NEXT: vpshufb %xmm2, %xmm0, %xmm0
; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_02_00_06_04_05_01_07_11_10_08_14_12_13_09_15_11: ; AVX2-LABEL: shuffle_v16i16_02_00_06_04_05_01_07_11_10_08_14_12_13_09_15_11:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vmovdqa {{.*#+}} xmm2 = [4,5,0,1,12,13,8,9,10,11,2,3,14,15,6,7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2OR512VL-NEXT: vpshufb %xmm2, %xmm1, %xmm3 ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[3],xmm0[4,5,6,7] ; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[4,5,0,1,12,13,8,9,10,11,2,3,14,15,6,7,20,21,16,17,28,29,24,25,26,27,18,19,30,31,22,23]
; AVX2OR512VL-NEXT: vpshufb %xmm2, %xmm0, %xmm0 ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm3, %ymm0, %ymm0 ;
; AVX2OR512VL-NEXT: retq ; AVX512VL-LABEL: shuffle_v16i16_02_00_06_04_05_01_07_11_10_08_14_12_13_09_15_11:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [2,0,6,4,5,1,7,11,10,8,14,12,13,9,15,11]
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 2, i32 0, i32 6, i32 4, i32 5, i32 1, i32 7, i32 11, i32 10, i32 8, i32 14, i32 12, i32 13, i32 9, i32 15, i32 11> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 2, i32 0, i32 6, i32 4, i32 5, i32 1, i32 7, i32 11, i32 10, i32 8, i32 14, i32 12, i32 13, i32 9, i32 15, i32 11>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_02_06_04_00_01_03_07_13_10_14_12_08_09_11_15_13(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_02_06_04_00_01_03_07_13_10_14_12_08_09_11_15_13(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_02_06_04_00_01_03_07_13_10_14_12_08_09_11_15_13: ; AVX1-LABEL: shuffle_v16i16_02_06_04_00_01_03_07_13_10_14_12_08_09_11_15_13:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vmovdqa {{.*#+}} xmm2 = [4,5,12,13,8,9,0,1,2,3,6,7,14,15,10,11] ; AVX1-NEXT: vmovdqa {{.*#+}} xmm2 = [4,5,12,13,8,9,0,1,2,3,6,7,14,15,10,11]
; AVX1-NEXT: vpshufb %xmm2, %xmm1, %xmm3 ; AVX1-NEXT: vpshufb %xmm2, %xmm1, %xmm3
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4],xmm1[5],xmm0[6,7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4],xmm1[5],xmm0[6,7]
; AVX1-NEXT: vpshufb %xmm2, %xmm0, %xmm0 ; AVX1-NEXT: vpshufb %xmm2, %xmm0, %xmm0
; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_02_06_04_00_01_03_07_13_10_14_12_08_09_11_15_13: ; AVX2-LABEL: shuffle_v16i16_02_06_04_00_01_03_07_13_10_14_12_08_09_11_15_13:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vmovdqa {{.*#+}} xmm2 = [4,5,12,13,8,9,0,1,2,3,6,7,14,15,10,11] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2OR512VL-NEXT: vpshufb %xmm2, %xmm1, %xmm3 ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4],xmm1[5],xmm0[6,7] ; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[4,5,12,13,8,9,0,1,2,3,6,7,14,15,10,11,20,21,28,29,24,25,16,17,18,19,22,23,30,31,26,27]
; AVX2OR512VL-NEXT: vpshufb %xmm2, %xmm0, %xmm0 ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm3, %ymm0, %ymm0 ;
; AVX2OR512VL-NEXT: retq ; AVX512VL-LABEL: shuffle_v16i16_02_06_04_00_01_03_07_13_10_14_12_08_09_11_15_13:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [2,6,4,0,1,3,7,13,10,14,12,8,9,11,15,13]
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 2, i32 6, i32 4, i32 0, i32 1, i32 3, i32 7, i32 13, i32 10, i32 14, i32 12, i32 8, i32 9, i32 11, i32 15, i32 13> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 2, i32 6, i32 4, i32 0, i32 1, i32 3, i32 7, i32 13, i32 10, i32 14, i32 12, i32 8, i32 9, i32 11, i32 15, i32 13>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_06_06_07_05_01_06_04_11_14_14_15_13_09_14_12_11(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_06_06_07_05_01_06_04_11_14_14_15_13_09_14_12_11(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_06_06_07_05_01_06_04_11_14_14_15_13_09_14_12_11: ; AVX1-LABEL: shuffle_v16i16_06_06_07_05_01_06_04_11_14_14_15_13_09_14_12_11:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vmovdqa {{.*#+}} xmm2 = [12,13,12,13,14,15,10,11,2,3,12,13,8,9,6,7] ; AVX1-NEXT: vmovdqa {{.*#+}} xmm2 = [12,13,12,13,14,15,10,11,2,3,12,13,8,9,6,7]
; AVX1-NEXT: vpshufb %xmm2, %xmm1, %xmm3 ; AVX1-NEXT: vpshufb %xmm2, %xmm1, %xmm3
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
; AVX1-NEXT: vpshufb %xmm2, %xmm0, %xmm0 ; AVX1-NEXT: vpshufb %xmm2, %xmm0, %xmm0
; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_06_06_07_05_01_06_04_11_14_14_15_13_09_14_12_11: ; AVX2-LABEL: shuffle_v16i16_06_06_07_05_01_06_04_11_14_14_15_13_09_14_12_11:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vmovdqa {{.*#+}} xmm2 = [12,13,12,13,14,15,10,11,2,3,12,13,8,9,6,7] ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3,4,5,6,7]
; AVX2OR512VL-NEXT: vpshufb %xmm2, %xmm1, %xmm3 ; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[12,13,12,13,14,15,10,11,2,3,12,13,8,9,6,7,28,29,28,29,30,31,26,27,18,19,28,29,24,25,22,23]
; AVX2OR512VL-NEXT: vpblendd {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3] ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: vpshufb %xmm2, %xmm0, %xmm0 ;
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm3, %ymm0, %ymm0 ; AVX512VL-LABEL: shuffle_v16i16_06_06_07_05_01_06_04_11_14_14_15_13_09_14_12_11:
; AVX2OR512VL-NEXT: retq ; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [6,6,7,5,1,6,4,11,14,14,15,13,9,14,12,11]
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 6, i32 6, i32 7, i32 5, i32 1, i32 6, i32 4, i32 11, i32 14, i32 14, i32 15, i32 13, i32 9, i32 14, i32 12, i32 11> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 6, i32 6, i32 7, i32 5, i32 1, i32 6, i32 4, i32 11, i32 14, i32 14, i32 15, i32 13, i32 9, i32 14, i32 12, i32 11>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_00_00_04_04_04_04_04_12_08_08_12_12_12_12_12_12(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_00_00_04_04_04_04_04_12_08_08_12_12_12_12_12_12(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_00_00_04_04_04_04_04_12_08_08_12_12_12_12_12_12: ; AVX1-LABEL: shuffle_v16i16_00_00_04_04_04_04_04_12_08_08_12_12_12_12_12_12:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpsllq $48, %xmm1, %xmm2 ; AVX1-NEXT: vpsllq $48, %xmm1, %xmm2
; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,8,9,8,9,8,9,8,9,8,9,14,15] ; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,8,9,8,9,8,9,8,9,8,9,14,15]
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,0,1,8,9,8,9,8,9,8,9,8,9,8,9] ; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,0,1,8,9,8,9,8,9,8,9,8,9,8,9]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_00_00_04_04_04_04_04_12_08_08_12_12_12_12_12_12: ; AVX2-LABEL: shuffle_v16i16_00_00_04_04_04_04_04_12_08_08_12_12_12_12_12_12:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vpsllq $48, %xmm1, %xmm2 ; AVX2-NEXT: vpsllq $48, %xmm1, %xmm1
; AVX2OR512VL-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,8,9,8,9,8,9,8,9,8,9,14,15] ; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,0,1,8,9,8,9,8,9,8,9,8,9,8,9,16,17,16,17,24,25,24,25,24,25,24,25,24,25,24,25]
; AVX2OR512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2OR512VL-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,0,1,8,9,8,9,8,9,8,9,8,9,8,9] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0 ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: retq ;
; AVX512VL-LABEL: shuffle_v16i16_00_00_04_04_04_04_04_12_08_08_12_12_12_12_12_12:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [0,0,4,4,4,4,4,12,8,8,12,12,12,12,12,12]
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 0, i32 4, i32 4, i32 4, i32 4, i32 4, i32 12, i32 8, i32 8, i32 12, i32 12, i32 12, i32 12, i32 12, i32 12> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 0, i32 4, i32 4, i32 4, i32 4, i32 4, i32 12, i32 8, i32 8, i32 12, i32 12, i32 12, i32 12, i32 12, i32 12>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_04_04_00_00_04_04_04_12_12_12_08_08_12_12_12_12(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_04_04_00_00_04_04_04_12_12_12_08_08_12_12_12_12(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_04_04_00_00_04_04_04_12_12_12_08_08_12_12_12_12: ; AVX1-LABEL: shuffle_v16i16_04_04_00_00_04_04_04_12_12_12_08_08_12_12_12_12:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpsllq $48, %xmm1, %xmm2 ; AVX1-NEXT: vpsllq $48, %xmm1, %xmm2
; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8,9,8,9,0,1,0,1,8,9,8,9,8,9,14,15] ; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8,9,8,9,0,1,0,1,8,9,8,9,8,9,14,15]
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[8,9,8,9,0,1,0,1,8,9,8,9,8,9,8,9] ; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[8,9,8,9,0,1,0,1,8,9,8,9,8,9,8,9]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_04_04_00_00_04_04_04_12_12_12_08_08_12_12_12_12: ; AVX2-LABEL: shuffle_v16i16_04_04_00_00_04_04_04_12_12_12_08_08_12_12_12_12:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vpsllq $48, %xmm1, %xmm2 ; AVX2-NEXT: vpsllq $48, %xmm1, %xmm1
; AVX2OR512VL-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8,9,8,9,0,1,0,1,8,9,8,9,8,9,14,15] ; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[8,9,8,9,0,1,0,1,8,9,8,9,8,9,8,9,24,25,24,25,16,17,16,17,24,25,24,25,24,25,24,25]
; AVX2OR512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2OR512VL-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[8,9,8,9,0,1,0,1,8,9,8,9,8,9,8,9] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0 ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: retq ;
; AVX512VL-LABEL: shuffle_v16i16_04_04_00_00_04_04_04_12_12_12_08_08_12_12_12_12:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [4,4,0,0,4,4,4,12,12,12,8,8,12,12,12,12]
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 4, i32 4, i32 0, i32 0, i32 4, i32 4, i32 4, i32 12, i32 12, i32 12, i32 8, i32 8, i32 12, i32 12, i32 12, i32 12> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 4, i32 4, i32 0, i32 0, i32 4, i32 4, i32 4, i32 12, i32 12, i32 12, i32 8, i32 8, i32 12, i32 12, i32 12, i32 12>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_00_04_04_00_04_04_04_12_08_12_12_08_12_12_12_12(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_00_04_04_00_04_04_04_12_08_12_12_08_12_12_12_12(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_00_04_04_00_04_04_04_12_08_12_12_08_12_12_12_12: ; AVX1-LABEL: shuffle_v16i16_00_04_04_00_04_04_04_12_08_12_12_08_12_12_12_12:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpsllq $48, %xmm1, %xmm2 ; AVX1-NEXT: vpsllq $48, %xmm1, %xmm2
; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,8,9,8,9,8,9,14,15] ; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,8,9,8,9,8,9,14,15]
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,8,9,8,9,0,1,8,9,8,9,8,9,8,9] ; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,8,9,8,9,0,1,8,9,8,9,8,9,8,9]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_00_04_04_00_04_04_04_12_08_12_12_08_12_12_12_12: ; AVX2-LABEL: shuffle_v16i16_00_04_04_00_04_04_04_12_08_12_12_08_12_12_12_12:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vpsllq $48, %xmm1, %xmm2 ; AVX2-NEXT: vpsllq $48, %xmm1, %xmm1
; AVX2OR512VL-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,8,9,8,9,8,9,14,15] ; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,8,9,8,9,0,1,8,9,8,9,8,9,8,9,16,17,24,25,24,25,16,17,24,25,24,25,24,25,24,25]
; AVX2OR512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2OR512VL-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,8,9,8,9,0,1,8,9,8,9,8,9,8,9] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0 ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: retq ;
; AVX512VL-LABEL: shuffle_v16i16_00_04_04_00_04_04_04_12_08_12_12_08_12_12_12_12:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [0,4,4,0,4,4,4,12,8,12,12,8,12,12,12,12]
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 4, i32 4, i32 0, i32 4, i32 4, i32 4, i32 12, i32 8, i32 12, i32 12, i32 8, i32 12, i32 12, i32 12, i32 12> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 4, i32 4, i32 0, i32 4, i32 4, i32 4, i32 12, i32 8, i32 12, i32 12, i32 8, i32 12, i32 12, i32 12, i32 12>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_00_04_04_00_00_00_00_08_08_12_12_08_08_08_08_08(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_00_04_04_00_00_00_00_08_08_12_12_08_08_08_08_08(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_00_04_04_00_00_00_00_08_08_12_12_08_08_08_08_08: ; AVX1-LABEL: shuffle_v16i16_00_04_04_00_00_00_00_08_08_12_12_08_08_08_08_08:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpslldq {{.*#+}} xmm2 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm1[0,1] ; AVX1-NEXT: vpslldq {{.*#+}} xmm2 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm1[0,1]
; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,0,1,0,1,0,1,14,15] ; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,0,1,0,1,0,1,14,15]
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,8,9,8,9,0,1,0,1,0,1,0,1,0,1] ; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,8,9,8,9,0,1,0,1,0,1,0,1,0,1]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_00_04_04_00_00_00_00_08_08_12_12_08_08_08_08_08: ; AVX2-LABEL: shuffle_v16i16_00_04_04_00_00_00_00_08_08_12_12_08_08_08_08_08:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vpbroadcastw %xmm1, %xmm2 ; AVX2-NEXT: vpbroadcastw %xmm1, %xmm1
; AVX2OR512VL-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,0,1,0,1,0,1,14,15] ; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,8,9,8,9,0,1,0,1,0,1,0,1,0,1,16,17,24,25,24,25,16,17,16,17,16,17,16,17,16,17]
; AVX2OR512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2OR512VL-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,8,9,8,9,0,1,0,1,0,1,0,1,0,1] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0 ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: retq ;
; AVX512VL-LABEL: shuffle_v16i16_00_04_04_00_00_00_00_08_08_12_12_08_08_08_08_08:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [0,4,4,0,0,0,0,8,8,12,12,8,8,8,8,8]
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 4, i32 4, i32 0, i32 0, i32 0, i32 0, i32 8, i32 8, i32 12, i32 12, i32 8, i32 8, i32 8, i32 8, i32 8> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 4, i32 4, i32 0, i32 0, i32 0, i32 0, i32 8, i32 8, i32 12, i32 12, i32 8, i32 8, i32 8, i32 8, i32 8>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_00_04_04_00_04_05_06_15_08_12_12_08_12_13_14_15(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_00_04_04_00_04_05_06_15_08_12_12_08_12_13_14_15(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_00_04_04_00_04_05_06_15_08_12_12_08_12_13_14_15: ; AVX1-LABEL: shuffle_v16i16_00_04_04_00_04_05_06_15_08_12_12_08_12_13_14_15:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3] ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,0,4,5,6,7] ; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,0,4,5,6,7]
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm1[7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm1[7]
; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3] ; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
; AVX1-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,2,2,0,4,5,6,7] ; AVX1-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,2,2,0,4,5,6,7]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_00_04_04_00_04_05_06_15_08_12_12_08_12_13_14_15: ; AVX2-LABEL: shuffle_v16i16_00_04_04_00_04_05_06_15_08_12_12_08_12_13_14_15:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3] ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
; AVX2OR512VL-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,0,4,5,6,7] ; AVX2-NEXT: vpshuflw {{.*#+}} ymm0 = ymm0[0,2,2,0,4,5,6,7,8,10,10,8,12,13,14,15]
; AVX2OR512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm1[7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2OR512VL-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,2,2,0,4,5,6,7] ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0 ;
; AVX2OR512VL-NEXT: retq ; AVX512VL-LABEL: shuffle_v16i16_00_04_04_00_04_05_06_15_08_12_12_08_12_13_14_15:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [0,4,4,0,4,5,6,15,8,12,12,8,12,13,14,15]
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 4, i32 4, i32 0, i32 4, i32 5, i32 6, i32 15, i32 8, i32 12, i32 12, i32 8, i32 12, i32 13, i32 14, i32 15> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 4, i32 4, i32 0, i32 4, i32 5, i32 6, i32 15, i32 8, i32 12, i32 12, i32 8, i32 12, i32 13, i32 14, i32 15>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_00_uu_04_04_04_04_04_12_08_uu_12_12_12_12_12_12(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_00_uu_04_04_04_04_04_12_08_uu_12_12_12_12_12_12(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_00_uu_04_04_04_04_04_12_08_uu_12_12_12_12_12_12: ; AVX1-LABEL: shuffle_v16i16_00_uu_04_04_04_04_04_12_08_uu_12_12_12_12_12_12:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpsllq $48, %xmm1, %xmm2 ; AVX1-NEXT: vpsllq $48, %xmm1, %xmm2
; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,8,9,8,9,8,9,8,9,8,9,14,15] ; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,8,9,8,9,8,9,8,9,8,9,14,15]
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,2,3,8,9,8,9,8,9,8,9,8,9,8,9] ; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,2,3,8,9,8,9,8,9,8,9,8,9,8,9]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_00_uu_04_04_04_04_04_12_08_uu_12_12_12_12_12_12: ; AVX2-LABEL: shuffle_v16i16_00_uu_04_04_04_04_04_12_08_uu_12_12_12_12_12_12:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vpsllq $48, %xmm1, %xmm2 ; AVX2-NEXT: vpsllq $48, %xmm1, %xmm1
; AVX2OR512VL-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,8,9,8,9,8,9,8,9,8,9,14,15] ; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,2,3,8,9,8,9,8,9,8,9,8,9,8,9,16,17,18,19,24,25,24,25,24,25,24,25,24,25,24,25]
; AVX2OR512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = <255,255,u,u,255,255,255,255,255,255,255,255,255,255,0,0,255,255,u,u,255,255,255,255,255,255,255,255,255,255,255,255>
; AVX2OR512VL-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,2,3,8,9,8,9,8,9,8,9,8,9,8,9] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0 ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: retq ;
; AVX512VL-LABEL: shuffle_v16i16_00_uu_04_04_04_04_04_12_08_uu_12_12_12_12_12_12:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = <0,u,4,4,4,4,4,12,8,u,12,12,12,12,12,12>
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 undef, i32 4, i32 4, i32 4, i32 4, i32 4, i32 12, i32 8, i32 undef, i32 12, i32 12, i32 12, i32 12, i32 12, i32 12> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 undef, i32 4, i32 4, i32 4, i32 4, i32 4, i32 12, i32 8, i32 undef, i32 12, i32 12, i32 12, i32 12, i32 12, i32 12>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_04_04_uu_00_04_04_04_12_12_12_uu_08_12_12_12_12(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_04_04_uu_00_04_04_04_12_12_12_uu_08_12_12_12_12(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_04_04_uu_00_04_04_04_12_12_12_uu_08_12_12_12_12: ; AVX1-LABEL: shuffle_v16i16_04_04_uu_00_04_04_04_12_12_12_uu_08_12_12_12_12:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpsllq $48, %xmm1, %xmm2 ; AVX1-NEXT: vpsllq $48, %xmm1, %xmm2
; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8,9,8,9,8,9,0,1,8,9,8,9,8,9,14,15] ; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8,9,8,9,8,9,0,1,8,9,8,9,8,9,14,15]
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[8,9,8,9,8,9,0,1,8,9,8,9,8,9,8,9] ; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[8,9,8,9,8,9,0,1,8,9,8,9,8,9,8,9]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_04_04_uu_00_04_04_04_12_12_12_uu_08_12_12_12_12: ; AVX2-LABEL: shuffle_v16i16_04_04_uu_00_04_04_04_12_12_12_uu_08_12_12_12_12:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vpsllq $48, %xmm1, %xmm2 ; AVX2-NEXT: vpsllq $48, %xmm1, %xmm1
; AVX2OR512VL-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8,9,8,9,8,9,0,1,8,9,8,9,8,9,14,15] ; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[8,9,8,9,8,9,0,1,8,9,8,9,8,9,8,9,24,25,24,25,24,25,16,17,24,25,24,25,24,25,24,25]
; AVX2OR512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = <255,255,255,255,u,u,255,255,255,255,255,255,255,255,0,0,255,255,255,255,u,u,255,255,255,255,255,255,255,255,255,255>
; AVX2OR512VL-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[8,9,8,9,8,9,0,1,8,9,8,9,8,9,8,9] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0 ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: retq ;
; AVX512VL-LABEL: shuffle_v16i16_04_04_uu_00_04_04_04_12_12_12_uu_08_12_12_12_12:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = <4,4,u,0,4,4,4,12,12,12,u,8,12,12,12,12>
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 4, i32 4, i32 undef, i32 0, i32 4, i32 4, i32 4, i32 12, i32 12, i32 12, i32 undef, i32 8, i32 12, i32 12, i32 12, i32 12> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 4, i32 4, i32 undef, i32 0, i32 4, i32 4, i32 4, i32 12, i32 12, i32 12, i32 undef, i32 8, i32 12, i32 12, i32 12, i32 12>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_uu_04_04_00_04_04_04_12_uu_12_12_08_12_12_12_12(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_uu_04_04_00_04_04_04_12_uu_12_12_08_12_12_12_12(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_uu_04_04_00_04_04_04_12_uu_12_12_08_12_12_12_12: ; AVX1-LABEL: shuffle_v16i16_uu_04_04_00_04_04_04_12_uu_12_12_08_12_12_12_12:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpsllq $48, %xmm1, %xmm2 ; AVX1-NEXT: vpsllq $48, %xmm1, %xmm2
; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,8,9,8,9,8,9,14,15] ; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,8,9,8,9,8,9,14,15]
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,8,9,8,9,0,1,8,9,8,9,8,9,8,9] ; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,8,9,8,9,0,1,8,9,8,9,8,9,8,9]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_uu_04_04_00_04_04_04_12_uu_12_12_08_12_12_12_12: ; AVX2-LABEL: shuffle_v16i16_uu_04_04_00_04_04_04_12_uu_12_12_08_12_12_12_12:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vpsllq $48, %xmm1, %xmm2 ; AVX2-NEXT: vpsllq $48, %xmm1, %xmm1
; AVX2OR512VL-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,8,9,8,9,8,9,14,15] ; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,8,9,8,9,0,1,8,9,8,9,8,9,8,9,16,17,24,25,24,25,16,17,24,25,24,25,24,25,24,25]
; AVX2OR512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = <u,u,255,255,255,255,255,255,255,255,255,255,255,255,0,0,u,u,255,255,255,255,255,255,255,255,255,255,255,255,255,255>
; AVX2OR512VL-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,8,9,8,9,0,1,8,9,8,9,8,9,8,9] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0 ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: retq ;
; AVX512VL-LABEL: shuffle_v16i16_uu_04_04_00_04_04_04_12_uu_12_12_08_12_12_12_12:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = <u,4,4,0,4,4,4,12,u,12,12,8,12,12,12,12>
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 undef, i32 4, i32 4, i32 0, i32 4, i32 4, i32 4, i32 12, i32 undef, i32 12, i32 12, i32 8, i32 12, i32 12, i32 12, i32 12> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 undef, i32 4, i32 4, i32 0, i32 4, i32 4, i32 4, i32 12, i32 undef, i32 12, i32 12, i32 8, i32 12, i32 12, i32 12, i32 12>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_00_01_02_07_uu_uu_uu_uu_08_09_10_15_uu_uu_uu_uu(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_00_01_02_07_uu_uu_uu_uu_08_09_10_15_uu_uu_uu_uu(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_00_01_02_07_uu_uu_uu_uu_08_09_10_15_uu_uu_uu_uu: ; AVX1-LABEL: shuffle_v16i16_00_01_02_07_uu_uu_uu_uu_08_09_10_15_uu_uu_uu_uu:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
Show All 25 Lines
; AVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm1[0,1,0,1] ; AVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm1[0,1,0,1]
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[12,13,6,7,4,5,6,7,8,9,10,11,12,13,6,7] ; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[12,13,6,7,4,5,6,7,8,9,10,11,12,13,6,7]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2-LABEL: shuffle_v16i16_uu_uu_uu_uu_04_05_06_11_uu_uu_uu_uu_12_13_14_11: ; AVX2-LABEL: shuffle_v16i16_uu_uu_uu_uu_04_05_06_11_uu_uu_uu_uu_12_13_14_11:
; AVX2: # BB#0: ; AVX2: # BB#0:
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,1,2,3]
; AVX2-NEXT: vpbroadcastq %xmm1, %xmm2 ; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[12,13,6,7,4,5,6,7,8,9,10,11,12,13,6,7,28,29,22,23,20,21,22,23,24,25,26,27,28,29,22,23]
; AVX2-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
; AVX2-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[12,13,6,7,4,5,6,7,8,9,10,11,12,13,6,7]
; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
; AVX2-NEXT: retq ; AVX2-NEXT: retq
; ;
; AVX512VL-LABEL: shuffle_v16i16_uu_uu_uu_uu_04_05_06_11_uu_uu_uu_uu_12_13_14_11: ; AVX512VL-LABEL: shuffle_v16i16_uu_uu_uu_uu_04_05_06_11_uu_uu_uu_uu_12_13_14_11:
; AVX512VL: # BB#0: ; AVX512VL: # BB#0:
; AVX512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = <u,u,u,u,4,5,6,11,u,u,u,u,12,13,14,11>
; AVX512VL-NEXT: vpbroadcastq %xmm1, %xmm2 ; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
; AVX512VL-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[3,1,2,0]
; AVX512VL-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,3,2,3,4,5,6,7]
; AVX512VL-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,1,2,0]
; AVX512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
; AVX512VL-NEXT: retq ; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 undef, i32 undef, i32 undef, i32 undef, i32 4, i32 5, i32 6, i32 11, i32 undef, i32 undef, i32 undef, i32 undef, i32 12, i32 13, i32 14, i32 11> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 undef, i32 undef, i32 undef, i32 undef, i32 4, i32 5, i32 6, i32 11, i32 undef, i32 undef, i32 undef, i32 undef, i32 12, i32 13, i32 14, i32 11>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_04_05_06_03_uu_uu_uu_uu_12_13_14_11_uu_uu_uu_uu(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_04_05_06_03_uu_uu_uu_uu_12_13_14_11_uu_uu_uu_uu(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_04_05_06_03_uu_uu_uu_uu_12_13_14_11_uu_uu_uu_uu: ; AVX1-LABEL: shuffle_v16i16_04_05_06_03_uu_uu_uu_uu_12_13_14_11_uu_uu_uu_uu:
; AVX1: # BB#0: ; AVX1: # BB#0:
▲ Show 20 Lines • Show All 44 Lines • ▼ Show 20 Lines
; AVX1-NEXT: vpshufb %xmm2, %xmm1, %xmm3 ; AVX1-NEXT: vpshufb %xmm2, %xmm1, %xmm3
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[3],xmm0[4,5,6,7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[3],xmm0[4,5,6,7]
; AVX1-NEXT: vpshufb %xmm2, %xmm0, %xmm0 ; AVX1-NEXT: vpshufb %xmm2, %xmm0, %xmm0
; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2-LABEL: shuffle_v16i16_00_01_02_07_04_05_06_11_08_09_10_15_12_13_14_11: ; AVX2-LABEL: shuffle_v16i16_00_01_02_07_04_05_06_11_08_09_10_15_12_13_14_11:
; AVX2: # BB#0: ; AVX2: # BB#0:
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2-NEXT: vmovdqa {{.*#+}} xmm2 = [0,1,2,3,4,5,14,15,8,9,10,11,12,13,6,7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2-NEXT: vpshufb %xmm2, %xmm1, %xmm3 ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[3],xmm0[4,5,6,7] ; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5,14,15,8,9,10,11,12,13,6,7,16,17,18,19,20,21,30,31,24,25,26,27,28,29,22,23]
; AVX2-NEXT: vpshufb %xmm2, %xmm0, %xmm0
; AVX2-NEXT: vinserti128 $1, %xmm3, %ymm0, %ymm0
; AVX2-NEXT: retq ; AVX2-NEXT: retq
; ;
; AVX512VL-LABEL: shuffle_v16i16_00_01_02_07_04_05_06_11_08_09_10_15_12_13_14_11: ; AVX512VL-LABEL: shuffle_v16i16_00_01_02_07_04_05_06_11_08_09_10_15_12_13_14_11:
; AVX512VL: # BB#0: ; AVX512VL: # BB#0:
; AVX512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [0,1,2,7,4,5,6,11,8,9,10,15,12,13,14,11]
; AVX512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[3],xmm0[4,5,6,7] ; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
; AVX512VL-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,5,4,7]
; AVX512VL-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,3,1,2]
; AVX512VL-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,2,1,3]
; AVX512VL-NEXT: vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,6,5,4,7]
; AVX512VL-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,3,1,2]
; AVX512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
; AVX512VL-NEXT: retq ; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 1, i32 2, i32 7, i32 4, i32 5, i32 6, i32 11, i32 8, i32 9, i32 10, i32 15, i32 12, i32 13, i32 14, i32 11> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 1, i32 2, i32 7, i32 4, i32 5, i32 6, i32 11, i32 8, i32 9, i32 10, i32 15, i32 12, i32 13, i32 14, i32 11>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_04_05_06_03_00_01_02_15_12_13_14_11_08_09_10_15(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_04_05_06_03_00_01_02_15_12_13_14_11_08_09_10_15(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_04_05_06_03_00_01_02_15_12_13_14_11_08_09_10_15: ; AVX1-LABEL: shuffle_v16i16_04_05_06_03_00_01_02_15_12_13_14_11_08_09_10_15:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8,9,10,11,12,13,6,7,0,1,2,3,4,5,2,3] ; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8,9,10,11,12,13,6,7,0,1,2,3,4,5,2,3]
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm1[7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm1[7]
; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[8,9,10,11,12,13,6,7,0,1,2,3,4,5,14,15] ; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[8,9,10,11,12,13,6,7,0,1,2,3,4,5,14,15]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2-LABEL: shuffle_v16i16_04_05_06_03_00_01_02_15_12_13_14_11_08_09_10_15: ; AVX2-LABEL: shuffle_v16i16_04_05_06_03_00_01_02_15_12_13_14_11_08_09_10_15:
; AVX2: # BB#0: ; AVX2: # BB#0:
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8,9,10,11,12,13,6,7,0,1,2,3,4,5,2,3] ; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[8,9,10,11,12,13,6,7,0,1,2,3,4,5,14,15,24,25,26,27,28,29,22,23,16,17,18,19,20,21,30,31]
; AVX2-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm1[7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[8,9,10,11,12,13,6,7,0,1,2,3,4,5,14,15] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
; AVX2-NEXT: retq ; AVX2-NEXT: retq
; ;
; AVX512VL-LABEL: shuffle_v16i16_04_05_06_03_00_01_02_15_12_13_14_11_08_09_10_15: ; AVX512VL-LABEL: shuffle_v16i16_04_05_06_03_00_01_02_15_12_13_14_11_08_09_10_15:
; AVX512VL: # BB#0: ; AVX512VL: # BB#0:
; AVX512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [4,5,6,3,0,1,2,15,12,13,14,11,8,9,10,15]
; AVX512VL-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8,9,10,11,12,13,6,7,0,1,2,3,4,5,2,3] ; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm1[7]
; AVX512VL-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[3,1,2,0]
; AVX512VL-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,3,2,1,4,5,6,7]
; AVX512VL-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[2,0,3,1]
; AVX512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
; AVX512VL-NEXT: retq ; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 4, i32 5, i32 6, i32 3, i32 0, i32 1, i32 2, i32 15, i32 12, i32 13, i32 14, i32 11, i32 8, i32 9, i32 10, i32 15> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 4, i32 5, i32 6, i32 3, i32 0, i32 1, i32 2, i32 15, i32 12, i32 13, i32 14, i32 11, i32 8, i32 9, i32 10, i32 15>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_03_07_01_00_02_07_03_13_11_15_09_08_10_15_11_13(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_03_07_01_00_02_07_03_13_11_15_09_08_10_15_11_13(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_03_07_01_00_02_07_03_13_11_15_09_08_10_15_11_13: ; AVX1-LABEL: shuffle_v16i16_03_07_01_00_02_07_03_13_11_15_09_08_10_15_11_13:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vmovdqa {{.*#+}} xmm2 = [6,7,14,15,2,3,0,1,4,5,14,15,6,7,10,11] ; AVX1-NEXT: vmovdqa {{.*#+}} xmm2 = [6,7,14,15,2,3,0,1,4,5,14,15,6,7,10,11]
; AVX1-NEXT: vpshufb %xmm2, %xmm1, %xmm3 ; AVX1-NEXT: vpshufb %xmm2, %xmm1, %xmm3
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5],xmm0[6,7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5],xmm0[6,7]
; AVX1-NEXT: vpshufb %xmm2, %xmm0, %xmm0 ; AVX1-NEXT: vpshufb %xmm2, %xmm0, %xmm0
; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_03_07_01_00_02_07_03_13_11_15_09_08_10_15_11_13: ; AVX2-LABEL: shuffle_v16i16_03_07_01_00_02_07_03_13_11_15_09_08_10_15_11_13:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vmovdqa {{.*#+}} xmm2 = [6,7,14,15,2,3,0,1,4,5,14,15,6,7,10,11] ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2],ymm0[3,4,5,6,7]
; AVX2OR512VL-NEXT: vpshufb %xmm2, %xmm1, %xmm3 ; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[6,7,14,15,2,3,0,1,4,5,14,15,6,7,10,11,22,23,30,31,18,19,16,17,20,21,30,31,22,23,26,27]
; AVX2OR512VL-NEXT: vpblendd {{.*#+}} xmm0 = xmm0[0,1],xmm1[2],xmm0[3] ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: vpshufb %xmm2, %xmm0, %xmm0 ;
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm3, %ymm0, %ymm0 ; AVX512VL-LABEL: shuffle_v16i16_03_07_01_00_02_07_03_13_11_15_09_08_10_15_11_13:
; AVX2OR512VL-NEXT: retq ; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [3,7,1,0,2,7,3,13,11,15,9,8,10,15,11,13]
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 3, i32 7, i32 1, i32 0, i32 2, i32 7, i32 3, i32 13, i32 11, i32 15, i32 9, i32 8, i32 10, i32 15, i32 11, i32 13> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 3, i32 7, i32 1, i32 0, i32 2, i32 7, i32 3, i32 13, i32 11, i32 15, i32 9, i32 8, i32 10, i32 15, i32 11, i32 13>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_00_16_01_17_02_18_03_27_08_24_09_25_10_26_11_27(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_00_16_01_17_02_18_03_27_08_24_09_25_10_26_11_27(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_00_16_01_17_02_18_03_27_08_24_09_25_10_26_11_27: ; AVX1-LABEL: shuffle_v16i16_00_16_01_17_02_18_03_27_08_24_09_25_10_26_11_27:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm2 ; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm2
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm3 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm3
; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1],xmm3[2],xmm2[2],xmm3[3],xmm2[3] ; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1],xmm3[2],xmm2[2],xmm3[3],xmm2[3]
; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3] ; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3]
; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,4,5,8,9,14,15,14,15,8,9,12,13,14,15] ; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,4,5,8,9,14,15,14,15,8,9,12,13,14,15]
; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] ; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2-LABEL: shuffle_v16i16_00_16_01_17_02_18_03_27_08_24_09_25_10_26_11_27: ; AVX2-LABEL: shuffle_v16i16_00_16_01_17_02_18_03_27_08_24_09_25_10_26_11_27:
; AVX2: # BB#0: ; AVX2: # BB#0:
; AVX2-NEXT: vextracti128 $1, %ymm1, %xmm2 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm2 = ymm1[2,3,0,1]
; AVX2-NEXT: vpunpcklwd {{.*#+}} xmm3 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm3 = <255,255,255,255,255,255,0,0,u,u,u,u,u,u,u,u,255,255,255,255,255,255,255,255,u,u,u,u,u,u,u,u>
; AVX2-NEXT: vpunpcklwd {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3] ; AVX2-NEXT: vpblendvb %ymm3, %ymm1, %ymm2, %ymm1
; AVX2-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,0,2,2,4,5,6,7] ; AVX2-NEXT: vpunpcklwd {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[8],ymm1[8],ymm0[9],ymm1[9],ymm0[10],ymm1[10],ymm0[11],ymm1[11]
; AVX2-NEXT: vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,4,6,7]
; AVX2-NEXT: vinserti128 $1, %xmm3, %ymm1, %ymm1
; AVX2-NEXT: vpunpcklwd {{.*#+}} ymm0 = ymm0[0,0,1,1,2,2,3,3,8,8,9,9,10,10,11,11]
; AVX2-NEXT: vpblendw {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7],ymm0[8],ymm1[9],ymm0[10],ymm1[11],ymm0[12],ymm1[13],ymm0[14],ymm1[15]
; AVX2-NEXT: retq ; AVX2-NEXT: retq
; ;
; AVX512VL-LABEL: shuffle_v16i16_00_16_01_17_02_18_03_27_08_24_09_25_10_26_11_27: ; AVX512VL-LABEL: shuffle_v16i16_00_16_01_17_02_18_03_27_08_24_09_25_10_26_11_27:
; AVX512VL: # BB#0: ; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [0,16,1,17,2,18,3,27,8,24,9,25,10,26,11,27] ; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [0,16,1,17,2,18,3,27,8,24,9,25,10,26,11,27]
; AVX512VL-NEXT: vpermt2w %ymm1, %ymm2, %ymm0 ; AVX512VL-NEXT: vpermt2w %ymm1, %ymm2, %ymm0
; AVX512VL-NEXT: retq ; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 16, i32 1, i32 17, i32 2, i32 18, i32 3, i32 27, i32 8, i32 24, i32 9, i32 25, i32 10, i32 26, i32 11, i32 27> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 16, i32 1, i32 17, i32 2, i32 18, i32 3, i32 27, i32 8, i32 24, i32 9, i32 25, i32 10, i32 26, i32 11, i32 27>
Show All 11 Lines
; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,4,5,8,9,14,15,14,15,8,9,12,13,14,15] ; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,4,5,8,9,14,15,14,15,8,9,12,13,14,15]
; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] ; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2-LABEL: shuffle_v16i16_00_20_01_21_02_22_03_31_08_28_09_29_10_30_11_31: ; AVX2-LABEL: shuffle_v16i16_00_20_01_21_02_22_03_31_08_28_09_29_10_30_11_31:
; AVX2: # BB#0: ; AVX2: # BB#0:
; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7] ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
; AVX2-NEXT: vmovdqa {{.*#+}} xmm1 = [0,1,8,9,2,3,10,11,4,5,12,13,6,7,14,15] ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2-NEXT: vpshufb %xmm1, %xmm0, %xmm2 ; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,8,9,2,3,10,11,4,5,12,13,6,7,14,15,16,17,24,25,18,19,26,27,20,21,28,29,22,23,30,31]
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm0 ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2-NEXT: vpblendw {{.*#+}} xmm2 = xmm2[0,1,2,3,4,5,6],xmm0[7] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2-NEXT: vpshufb %xmm1, %xmm0, %xmm0
; AVX2-NEXT: vinserti128 $1, %xmm0, %ymm2, %ymm0
; AVX2-NEXT: retq ; AVX2-NEXT: retq
; ;
; AVX512VL-LABEL: shuffle_v16i16_00_20_01_21_02_22_03_31_08_28_09_29_10_30_11_31: ; AVX512VL-LABEL: shuffle_v16i16_00_20_01_21_02_22_03_31_08_28_09_29_10_30_11_31:
; AVX512VL: # BB#0: ; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [0,20,1,21,2,22,3,31,8,28,9,29,10,30,11,31] ; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [0,20,1,21,2,22,3,31,8,28,9,29,10,30,11,31]
; AVX512VL-NEXT: vpermt2w %ymm1, %ymm2, %ymm0 ; AVX512VL-NEXT: vpermt2w %ymm1, %ymm2, %ymm0
; AVX512VL-NEXT: retq ; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 20, i32 1, i32 21, i32 2, i32 22, i32 3, i32 31, i32 8, i32 28, i32 9, i32 29, i32 10, i32 30, i32 11, i32 31> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 20, i32 1, i32 21, i32 2, i32 22, i32 3, i32 31, i32 8, i32 28, i32 9, i32 29, i32 10, i32 30, i32 11, i32 31>
Show All 9 Lines
; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm1 = xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7] ; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm1 = xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[4,5,0,1,4,5,4,5,0,1,4,5,8,9,14,15] ; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[4,5,0,1,4,5,4,5,0,1,4,5,8,9,14,15]
; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm0 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7] ; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm0 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2-LABEL: shuffle_v16i16_04_20_05_21_06_22_07_31_12_28_13_29_14_30_15_31: ; AVX2-LABEL: shuffle_v16i16_04_20_05_21_06_22_07_31_12_28_13_29_14_30_15_31:
; AVX2: # BB#0: ; AVX2: # BB#0:
; AVX2-NEXT: vextracti128 $1, %ymm1, %xmm2 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm2 = ymm1[2,3,0,1]
; AVX2-NEXT: vpunpckhwd {{.*#+}} xmm3 = xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7] ; AVX2-NEXT: vpunpckhwd {{.*#+}} ymm1 = ymm0[4],ymm1[4],ymm0[5],ymm1[5],ymm0[6],ymm1[6],ymm0[7],ymm1[7],ymm0[12],ymm1[12],ymm0[13],ymm1[13],ymm0[14],ymm1[14],ymm0[15],ymm1[15]
; AVX2-NEXT: vpunpckhwd {{.*#+}} xmm1 = xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7] ; AVX2-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1,2],ymm2[3],ymm1[4,5,6,7]
; AVX2-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,0,2,2,4,5,6,7]
; AVX2-NEXT: vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,4,6,7]
; AVX2-NEXT: vinserti128 $1, %xmm3, %ymm1, %ymm1
; AVX2-NEXT: vpunpckhwd {{.*#+}} ymm0 = ymm0[4,4,5,5,6,6,7,7,12,12,13,13,14,14,15,15] ; AVX2-NEXT: vpunpckhwd {{.*#+}} ymm0 = ymm0[4,4,5,5,6,6,7,7,12,12,13,13,14,14,15,15]
; AVX2-NEXT: vpblendw {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7],ymm0[8],ymm1[9],ymm0[10],ymm1[11],ymm0[12],ymm1[13],ymm0[14],ymm1[15] ; AVX2-NEXT: vpblendw {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7],ymm0[8],ymm1[9],ymm0[10],ymm1[11],ymm0[12],ymm1[13],ymm0[14],ymm1[15]
; AVX2-NEXT: retq ; AVX2-NEXT: retq
; ;
; AVX512VL-LABEL: shuffle_v16i16_04_20_05_21_06_22_07_31_12_28_13_29_14_30_15_31: ; AVX512VL-LABEL: shuffle_v16i16_04_20_05_21_06_22_07_31_12_28_13_29_14_30_15_31:
; AVX512VL: # BB#0: ; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [4,20,5,21,6,22,7,31,12,28,13,29,14,30,15,31] ; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [4,20,5,21,6,22,7,31,12,28,13,29,14,30,15,31]
; AVX512VL-NEXT: vpermt2w %ymm1, %ymm2, %ymm0 ; AVX512VL-NEXT: vpermt2w %ymm1, %ymm2, %ymm0
Show All 13 Lines
; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[4,5,0,1,4,5,4,5,0,1,4,5,8,9,14,15] ; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[4,5,0,1,4,5,4,5,0,1,4,5,8,9,14,15]
; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm0 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7] ; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm0 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2-LABEL: shuffle_v16i16_04_16_05_17_06_18_07_27_12_24_13_25_14_26_15_27: ; AVX2-LABEL: shuffle_v16i16_04_16_05_17_06_18_07_27_12_24_13_25_14_26_15_27:
; AVX2: # BB#0: ; AVX2: # BB#0:
; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3],ymm1[4,5],ymm0[6,7] ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3],ymm1[4,5],ymm0[6,7]
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[3],xmm0[4,5,6,7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2-NEXT: vmovdqa {{.*#+}} xmm2 = [8,9,0,1,10,11,2,3,12,13,4,5,14,15,6,7] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2-NEXT: vpshufb %xmm2, %xmm0, %xmm0 ; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[8,9,0,1,10,11,2,3,12,13,4,5,14,15,6,7,24,25,16,17,26,27,18,19,28,29,20,21,30,31,22,23]
; AVX2-NEXT: vpshufb %xmm2, %xmm1, %xmm1
; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
; AVX2-NEXT: retq ; AVX2-NEXT: retq
; ;
; AVX512VL-LABEL: shuffle_v16i16_04_16_05_17_06_18_07_27_12_24_13_25_14_26_15_27: ; AVX512VL-LABEL: shuffle_v16i16_04_16_05_17_06_18_07_27_12_24_13_25_14_26_15_27:
; AVX512VL: # BB#0: ; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [4,16,5,17,6,18,7,27,12,24,13,25,14,26,15,27] ; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [4,16,5,17,6,18,7,27,12,24,13,25,14,26,15,27]
; AVX512VL-NEXT: vpermt2w %ymm1, %ymm2, %ymm0 ; AVX512VL-NEXT: vpermt2w %ymm1, %ymm2, %ymm0
; AVX512VL-NEXT: retq ; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 4, i32 16, i32 5, i32 17, i32 6, i32 18, i32 7, i32 27, i32 12, i32 24, i32 13, i32 25, i32 14, i32 26, i32 15, i32 27> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 4, i32 16, i32 5, i32 17, i32 6, i32 18, i32 7, i32 27, i32 12, i32 24, i32 13, i32 25, i32 14, i32 26, i32 15, i32 27>
Show All 14 Lines
; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[2,3,6,7,10,11,14,15,14,15,10,11,12,13,14,15] ; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[2,3,6,7,10,11,14,15,14,15,10,11,12,13,14,15]
; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,3,2,3] ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,3,2,3]
; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] ; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2-LABEL: shuffle_v16i16_00_16_01_17_06_22_07_31_08_24_09_25_14_30_15_31: ; AVX2-LABEL: shuffle_v16i16_00_16_01_17_06_22_07_31_08_24_09_25_14_30_15_31:
; AVX2: # BB#0: ; AVX2: # BB#0:
; AVX2-NEXT: vextracti128 $1, %ymm1, %xmm2 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm2 = ymm1[2,3,0,1]
; AVX2-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,0,2,1,4,5,6,7] ; AVX2-NEXT: vpshuflw {{.*#+}} ymm1 = ymm1[0,0,2,1,4,5,6,7,8,8,10,9,12,13,14,15]
; AVX2-NEXT: vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,6,6,7] ; AVX2-NEXT: vpshufhw {{.*#+}} ymm1 = ymm1[0,1,2,3,4,6,6,7,8,9,10,11,12,14,14,15]
; AVX2-NEXT: vpblendd {{.*#+}} xmm1 = xmm1[0,1,2],xmm2[3] ; AVX2-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1,2],ymm2[3],ymm1[4,5,6,7]
; AVX2-NEXT: vpshuflw {{.*#+}} xmm2 = xmm2[0,0,2,1,4,5,6,7]
; AVX2-NEXT: vpshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,4,6,6,7]
; AVX2-NEXT: vinserti128 $1, %xmm2, %ymm1, %ymm1
; AVX2-NEXT: vpshuflw {{.*#+}} ymm0 = ymm0[0,1,1,3,4,5,6,7,8,9,9,11,12,13,14,15] ; AVX2-NEXT: vpshuflw {{.*#+}} ymm0 = ymm0[0,1,1,3,4,5,6,7,8,9,9,11,12,13,14,15]
; AVX2-NEXT: vpshufhw {{.*#+}} ymm0 = ymm0[0,1,2,3,6,5,7,7,8,9,10,11,14,13,15,15] ; AVX2-NEXT: vpshufhw {{.*#+}} ymm0 = ymm0[0,1,2,3,6,5,7,7,8,9,10,11,14,13,15,15]
; AVX2-NEXT: vpblendw {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7],ymm0[8],ymm1[9],ymm0[10],ymm1[11],ymm0[12],ymm1[13],ymm0[14],ymm1[15] ; AVX2-NEXT: vpblendw {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7],ymm0[8],ymm1[9],ymm0[10],ymm1[11],ymm0[12],ymm1[13],ymm0[14],ymm1[15]
; AVX2-NEXT: retq ; AVX2-NEXT: retq
; ;
; AVX512VL-LABEL: shuffle_v16i16_00_16_01_17_06_22_07_31_08_24_09_25_14_30_15_31: ; AVX512VL-LABEL: shuffle_v16i16_00_16_01_17_06_22_07_31_08_24_09_25_14_30_15_31:
; AVX512VL: # BB#0: ; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [0,16,1,17,6,22,7,31,8,24,9,25,14,30,15,31] ; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [0,16,1,17,6,22,7,31,8,24,9,25,14,30,15,31]
Show All 15 Lines
; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[8,9,10,11,0,1,2,3,2,3,0,1,12,13,2,3] ; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[8,9,10,11,0,1,2,3,2,3,0,1,12,13,2,3]
; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,3,2,3] ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,3,2,3]
; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] ; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2-LABEL: shuffle_v16i16_00_20_01_21_06_16_07_25_08_28_09_29_14_24_15_25: ; AVX2-LABEL: shuffle_v16i16_00_20_01_21_06_16_07_25_08_28_09_29_14_24_15_25:
; AVX2: # BB#0: ; AVX2: # BB#0:
; AVX2-NEXT: vextracti128 $1, %ymm1, %xmm2 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm2 = ymm1[2,3,0,1]
; AVX2-NEXT: vmovdqa {{.*#+}} xmm3 = [8,9,8,9,4,5,10,11,0,1,0,1,12,13,2,3] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm3 = <255,255,0,0,u,u,u,u,255,255,255,255,u,u,u,u,255,255,255,255,u,u,u,u,255,255,255,255,u,u,u,u>
; AVX2-NEXT: vpshufb %xmm3, %xmm2, %xmm4 ; AVX2-NEXT: vpblendvb %ymm3, %ymm1, %ymm2, %ymm1
; AVX2-NEXT: vpblendw {{.*#+}} xmm1 = xmm1[0],xmm2[1],xmm1[2,3,4,5,6,7] ; AVX2-NEXT: vpshufb {{.*#+}} ymm1 = ymm1[8,9,8,9,4,5,10,11,0,1,0,1,12,13,2,3,24,25,24,25,20,21,26,27,16,17,16,17,28,29,18,19]
; AVX2-NEXT: vpshufb %xmm3, %xmm1, %xmm1
; AVX2-NEXT: vinserti128 $1, %xmm4, %ymm1, %ymm1
; AVX2-NEXT: vpshuflw {{.*#+}} ymm0 = ymm0[0,1,1,3,4,5,6,7,8,9,9,11,12,13,14,15] ; AVX2-NEXT: vpshuflw {{.*#+}} ymm0 = ymm0[0,1,1,3,4,5,6,7,8,9,9,11,12,13,14,15]
; AVX2-NEXT: vpshufhw {{.*#+}} ymm0 = ymm0[0,1,2,3,6,5,7,7,8,9,10,11,14,13,15,15] ; AVX2-NEXT: vpshufhw {{.*#+}} ymm0 = ymm0[0,1,2,3,6,5,7,7,8,9,10,11,14,13,15,15]
; AVX2-NEXT: vpblendw {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7],ymm0[8],ymm1[9],ymm0[10],ymm1[11],ymm0[12],ymm1[13],ymm0[14],ymm1[15] ; AVX2-NEXT: vpblendw {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7],ymm0[8],ymm1[9],ymm0[10],ymm1[11],ymm0[12],ymm1[13],ymm0[14],ymm1[15]
; AVX2-NEXT: retq ; AVX2-NEXT: retq
; ;
; AVX512VL-LABEL: shuffle_v16i16_00_20_01_21_06_16_07_25_08_28_09_29_14_24_15_25: ; AVX512VL-LABEL: shuffle_v16i16_00_20_01_21_06_16_07_25_08_28_09_29_14_24_15_25:
; AVX512VL: # BB#0: ; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [0,20,1,21,6,16,7,25,8,28,9,29,14,24,15,25] ; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [0,20,1,21,6,16,7,25,8,28,9,29,14,24,15,25]
Show All 15 Lines
; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3] ; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[2,0,3,1,4,5,6,7] ; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[2,0,3,1,4,5,6,7]
; AVX1-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,4,7,5] ; AVX1-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,4,7,5]
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2-LABEL: shuffle_v16i16_01_00_17_16_03_02_19_26_09_08_25_24_11_10_27_26: ; AVX2-LABEL: shuffle_v16i16_01_00_17_16_03_02_19_26_09_08_25_24_11_10_27_26:
; AVX2: # BB#0: ; AVX2: # BB#0:
; AVX2-NEXT: vextracti128 $1, %ymm1, %xmm2 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm2 = ymm1[2,3,0,1]
; AVX2-NEXT: vpunpcklwd {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm3 = <255,255,255,255,0,0,255,255,u,u,u,u,u,u,u,u,255,255,255,255,255,255,255,255,u,u,u,u,u,u,u,u>
; AVX2-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,1,2,0,4,5,6,7] ; AVX2-NEXT: vpblendvb %ymm3, %ymm1, %ymm2, %ymm1
; AVX2-NEXT: vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,5,6,5] ; AVX2-NEXT: vpshufb {{.*#+}} ymm1 = ymm1[0,1,2,3,2,3,0,1,8,9,10,11,6,7,4,5,16,17,18,19,18,19,16,17,24,25,26,27,22,23,20,21]
; AVX2-NEXT: vpshufb {{.*#+}} xmm2 = xmm2[0,1,2,3,2,3,0,1,8,9,10,11,6,7,4,5]
; AVX2-NEXT: vinserti128 $1, %xmm2, %ymm1, %ymm1
; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[2,3,0,1,4,5,6,7,6,7,4,5,4,5,6,7,18,19,16,17,20,21,22,23,22,23,20,21,20,21,22,23] ; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[2,3,0,1,4,5,6,7,6,7,4,5,4,5,6,7,18,19,16,17,20,21,22,23,22,23,20,21,20,21,22,23]
; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7] ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7]
; AVX2-NEXT: retq ; AVX2-NEXT: retq
; ;
; AVX512VL-LABEL: shuffle_v16i16_01_00_17_16_03_02_19_26_09_08_25_24_11_10_27_26: ; AVX512VL-LABEL: shuffle_v16i16_01_00_17_16_03_02_19_26_09_08_25_24_11_10_27_26:
; AVX512VL: # BB#0: ; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [1,0,17,16,3,2,19,26,9,8,25,24,11,10,27,26] ; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [1,0,17,16,3,2,19,26,9,8,25,24,11,10,27,26]
; AVX512VL-NEXT: vpermt2w %ymm1, %ymm2, %ymm0 ; AVX512VL-NEXT: vpermt2w %ymm1, %ymm2, %ymm0
Show All 11 Lines
; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3] ; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,14,15,14,15,8,9,12,13,14,15] ; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,14,15,14,15,8,9,12,13,14,15]
; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3] ; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2-LABEL: shuffle_v16i16_16_00_17_01_18_02_19_11_24_08_25_09_26_10_27_11: ; AVX2-LABEL: shuffle_v16i16_16_00_17_01_18_02_19_11_24_08_25_09_26_10_27_11:
; AVX2: # BB#0: ; AVX2: # BB#0:
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm2 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm2 = ymm0[2,3,0,1]
; AVX2-NEXT: vpunpcklwd {{.*#+}} xmm3 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm3 = <255,255,255,255,255,255,0,0,u,u,u,u,u,u,u,u,255,255,255,255,255,255,255,255,u,u,u,u,u,u,u,u>
; AVX2-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3] ; AVX2-NEXT: vpblendvb %ymm3, %ymm0, %ymm2, %ymm0
; AVX2-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,0,2,2,4,5,6,7] ; AVX2-NEXT: vpunpcklwd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[1],ymm0[1],ymm1[2],ymm0[2],ymm1[3],ymm0[3],ymm1[8],ymm0[8],ymm1[9],ymm0[9],ymm1[10],ymm0[10],ymm1[11],ymm0[11]
; AVX2-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,6,7]
; AVX2-NEXT: vinserti128 $1, %xmm3, %ymm0, %ymm0
; AVX2-NEXT: vpunpcklwd {{.*#+}} ymm1 = ymm1[0],ymm0[0],ymm1[1],ymm0[1],ymm1[2],ymm0[2],ymm1[3],ymm0[3],ymm1[8],ymm0[8],ymm1[9],ymm0[9],ymm1[10],ymm0[10],ymm1[11],ymm0[11]
; AVX2-NEXT: vpblendw {{.*#+}} ymm0 = ymm1[0],ymm0[1],ymm1[2],ymm0[3],ymm1[4],ymm0[5],ymm1[6],ymm0[7],ymm1[8],ymm0[9],ymm1[10],ymm0[11],ymm1[12],ymm0[13],ymm1[14],ymm0[15]
; AVX2-NEXT: retq ; AVX2-NEXT: retq
; ;
; AVX512VL-LABEL: shuffle_v16i16_16_00_17_01_18_02_19_11_24_08_25_09_26_10_27_11: ; AVX512VL-LABEL: shuffle_v16i16_16_00_17_01_18_02_19_11_24_08_25_09_26_10_27_11:
; AVX512VL: # BB#0: ; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [0,16,1,17,2,18,3,27,8,24,9,25,10,26,11,27] ; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [0,16,1,17,2,18,3,27,8,24,9,25,10,26,11,27]
; AVX512VL-NEXT: vpermi2w %ymm0, %ymm1, %ymm2 ; AVX512VL-NEXT: vpermi2w %ymm0, %ymm1, %ymm2
; AVX512VL-NEXT: vmovdqa %ymm2, %ymm0 ; AVX512VL-NEXT: vmovdqa %ymm2, %ymm0
; AVX512VL-NEXT: retq ; AVX512VL-NEXT: retq
Show All 10 Lines
; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm0 = xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7] ; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm0 = xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[4,5,0,1,4,5,4,5,0,1,4,5,8,9,14,15] ; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[4,5,0,1,4,5,4,5,0,1,4,5,8,9,14,15]
; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm0 = xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7] ; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm0 = xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2-LABEL: shuffle_v16i16_20_04_21_05_22_06_23_15_28_12_29_13_30_14_31_15: ; AVX2-LABEL: shuffle_v16i16_20_04_21_05_22_06_23_15_28_12_29_13_30_14_31_15:
; AVX2: # BB#0: ; AVX2: # BB#0:
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm2 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm2 = ymm0[2,3,0,1]
; AVX2-NEXT: vpunpckhwd {{.*#+}} xmm3 = xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7] ; AVX2-NEXT: vpunpckhwd {{.*#+}} ymm0 = ymm0[4,4,5,5,6,6,7,7,12,12,13,13,14,14,15,15]
; AVX2-NEXT: vpunpckhwd {{.*#+}} xmm0 = xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7] ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2],ymm2[3],ymm0[4,5,6,7]
; AVX2-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,0,2,2,4,5,6,7]
; AVX2-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,6,7]
; AVX2-NEXT: vinserti128 $1, %xmm3, %ymm0, %ymm0
; AVX2-NEXT: vpunpckhwd {{.*#+}} ymm1 = ymm1[4],ymm0[4],ymm1[5],ymm0[5],ymm1[6],ymm0[6],ymm1[7],ymm0[7],ymm1[12],ymm0[12],ymm1[13],ymm0[13],ymm1[14],ymm0[14],ymm1[15],ymm0[15] ; AVX2-NEXT: vpunpckhwd {{.*#+}} ymm1 = ymm1[4],ymm0[4],ymm1[5],ymm0[5],ymm1[6],ymm0[6],ymm1[7],ymm0[7],ymm1[12],ymm0[12],ymm1[13],ymm0[13],ymm1[14],ymm0[14],ymm1[15],ymm0[15]
; AVX2-NEXT: vpblendw {{.*#+}} ymm0 = ymm1[0],ymm0[1],ymm1[2],ymm0[3],ymm1[4],ymm0[5],ymm1[6],ymm0[7],ymm1[8],ymm0[9],ymm1[10],ymm0[11],ymm1[12],ymm0[13],ymm1[14],ymm0[15] ; AVX2-NEXT: vpblendw {{.*#+}} ymm0 = ymm1[0],ymm0[1],ymm1[2],ymm0[3],ymm1[4],ymm0[5],ymm1[6],ymm0[7],ymm1[8],ymm0[9],ymm1[10],ymm0[11],ymm1[12],ymm0[13],ymm1[14],ymm0[15]
; AVX2-NEXT: retq ; AVX2-NEXT: retq
; ;
; AVX512VL-LABEL: shuffle_v16i16_20_04_21_05_22_06_23_15_28_12_29_13_30_14_31_15: ; AVX512VL-LABEL: shuffle_v16i16_20_04_21_05_22_06_23_15_28_12_29_13_30_14_31_15:
; AVX512VL: # BB#0: ; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [4,20,5,21,6,22,7,31,12,28,13,29,14,30,15,31] ; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [4,20,5,21,6,22,7,31,12,28,13,29,14,30,15,31]
; AVX512VL-NEXT: vpermi2w %ymm0, %ymm1, %ymm2 ; AVX512VL-NEXT: vpermi2w %ymm0, %ymm1, %ymm2
Show All 17 Lines
; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,1,3,4,5,6,7] ; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,1,3,4,5,6,7]
; AVX1-NEXT: vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0] ; AVX1-NEXT: vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2-LABEL: shuffle_v16i16_00_02_01_03_20_22_21_31_08_10_09_11_28_30_29_31: ; AVX2-LABEL: shuffle_v16i16_00_02_01_03_20_22_21_31_08_10_09_11_28_30_29_31:
; AVX2: # BB#0: ; AVX2: # BB#0:
; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7] ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,1,3,4,5,6,7] ; AVX2-NEXT: vpshuflw {{.*#+}} ymm0 = ymm0[0,2,1,3,4,5,6,7,8,10,9,11,12,13,14,15]
; AVX2-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,5,7] ; AVX2-NEXT: vpshufhw {{.*#+}} ymm0 = ymm0[0,1,2,3,4,6,5,7,8,9,10,11,12,14,13,15]
; AVX2-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm1[7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,2,1,3,4,5,6,7] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2-NEXT: vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,6,5,7]
; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
; AVX2-NEXT: retq ; AVX2-NEXT: retq
; ;
; AVX512VL-LABEL: shuffle_v16i16_00_02_01_03_20_22_21_31_08_10_09_11_28_30_29_31: ; AVX512VL-LABEL: shuffle_v16i16_00_02_01_03_20_22_21_31_08_10_09_11_28_30_29_31:
; AVX512VL: # BB#0: ; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [0,2,1,3,20,22,21,31,8,10,9,11,28,30,29,31] ; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [0,2,1,3,20,22,21,31,8,10,9,11,28,30,29,31]
; AVX512VL-NEXT: vpermt2w %ymm1, %ymm2, %ymm0 ; AVX512VL-NEXT: vpermt2w %ymm1, %ymm2, %ymm0
; AVX512VL-NEXT: retq ; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 2, i32 1, i32 3, i32 20, i32 22, i32 21, i32 31, i32 8, i32 10, i32 9, i32 11, i32 28, i32 30, i32 29, i32 31> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 2, i32 1, i32 3, i32 20, i32 22, i32 21, i32 31, i32 8, i32 10, i32 9, i32 11, i32 28, i32 30, i32 29, i32 31>
▲ Show 20 Lines • Show All 165 Lines • ▼ Show 20 Lines
; AVX1-NEXT: vpblendw {{.*#+}} xmm1 = xmm2[0,1,2,3],xmm1[4,5,6,7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm1 = xmm2[0,1,2,3],xmm1[4,5,6,7]
; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,2,3,4,5,10,11,8,9,10,11,12,13,6,7] ; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,2,3,4,5,10,11,8,9,10,11,12,13,6,7]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2-LABEL: shuffle_v16i16_00_01_02_21_20_21_22_11_08_09_10_29_28_29_30_11: ; AVX2-LABEL: shuffle_v16i16_00_01_02_21_20_21_22_11_08_09_10_29_28_29_30_11:
; AVX2: # BB#0: ; AVX2: # BB#0:
; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7] ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[3],xmm0[4,5,6,7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = <255,255,255,255,255,255,0,0,255,255,255,255,255,255,u,u,255,255,255,255,255,255,255,255,255,255,255,255,255,255,u,u>
; AVX2-NEXT: vmovdqa {{.*#+}} xmm2 = [0,1,2,3,4,5,10,11,8,9,10,11,12,13,6,7] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2-NEXT: vpshufb %xmm2, %xmm0, %xmm0 ; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5,10,11,8,9,10,11,12,13,6,7,16,17,18,19,20,21,26,27,24,25,26,27,28,29,22,23]
; AVX2-NEXT: vpshufb %xmm2, %xmm1, %xmm1
; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
; AVX2-NEXT: retq ; AVX2-NEXT: retq
; ;
; AVX512VL-LABEL: shuffle_v16i16_00_01_02_21_20_21_22_11_08_09_10_29_28_29_30_11: ; AVX512VL-LABEL: shuffle_v16i16_00_01_02_21_20_21_22_11_08_09_10_29_28_29_30_11:
; AVX512VL: # BB#0: ; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [0,1,2,21,20,21,22,11,8,9,10,29,28,29,30,11] ; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [0,1,2,21,20,21,22,11,8,9,10,29,28,29,30,11]
; AVX512VL-NEXT: vpermt2w %ymm1, %ymm2, %ymm0 ; AVX512VL-NEXT: vpermt2w %ymm1, %ymm2, %ymm0
; AVX512VL-NEXT: retq ; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 1, i32 2, i32 21, i32 20, i32 21, i32 22, i32 11, i32 8, i32 9, i32 10, i32 29, i32 28, i32 29, i32 30, i32 11> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 1, i32 2, i32 21, i32 20, i32 21, i32 22, i32 11, i32 8, i32 9, i32 10, i32 29, i32 28, i32 29, i32 30, i32 11>
▲ Show 20 Lines • Show All 100 Lines • ▼ Show 20 Lines
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm3[4,5,6,7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm3[4,5,6,7]
; AVX1-NEXT: vpalignr {{.*#+}} xmm0 = xmm1[10,11,12,13,14,15],xmm0[0,1,2,3,4,5,6,7,8,9] ; AVX1-NEXT: vpalignr {{.*#+}} xmm0 = xmm1[10,11,12,13,14,15],xmm0[0,1,2,3,4,5,6,7,8,9]
; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2-LABEL: shuffle_v16i16_21_22_23_00_01_02_03_12_29_30_31_08_09_10_11_12: ; AVX2-LABEL: shuffle_v16i16_21_22_23_00_01_02_03_12_29_30_31_08_09_10_11_12:
; AVX2: # BB#0: ; AVX2: # BB#0:
; AVX2-NEXT: vpblendw {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm1[5,6,7],ymm0[8,9,10,11,12],ymm1[13,14,15] ; AVX2-NEXT: vpblendw {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm1[5,6,7],ymm0[8,9,10,11,12],ymm1[13,14,15]
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4],xmm0[5,6,7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2-NEXT: vpalignr {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15,0,1,2,3,4,5,6,7,8,9] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2-NEXT: vpalignr {{.*#+}} xmm1 = xmm1[10,11,12,13,14,15,0,1,2,3,4,5,6,7,8,9] ; AVX2-NEXT: vpalignr {{.*#+}} ymm0 = ymm0[10,11,12,13,14,15,0,1,2,3,4,5,6,7,8,9,26,27,28,29,30,31,16,17,18,19,20,21,22,23,24,25]
; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
; AVX2-NEXT: retq ; AVX2-NEXT: retq
; ;
; AVX512VL-LABEL: shuffle_v16i16_21_22_23_00_01_02_03_12_29_30_31_08_09_10_11_12: ; AVX512VL-LABEL: shuffle_v16i16_21_22_23_00_01_02_03_12_29_30_31_08_09_10_11_12:
; AVX512VL: # BB#0: ; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [21,22,23,0,1,2,3,12,29,30,31,8,9,10,11,12] ; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [21,22,23,0,1,2,3,12,29,30,31,8,9,10,11,12]
; AVX512VL-NEXT: vpermt2w %ymm1, %ymm2, %ymm0 ; AVX512VL-NEXT: vpermt2w %ymm1, %ymm2, %ymm0
; AVX512VL-NEXT: retq ; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 21, i32 22, i32 23, i32 0, i32 1, i32 2, i32 3, i32 12, i32 29, i32 30, i32 31, i32 8, i32 9, i32 10, i32 11, i32 12> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 21, i32 22, i32 23, i32 0, i32 1, i32 2, i32 3, i32 12, i32 29, i32 30, i32 31, i32 8, i32 9, i32 10, i32 11, i32 12>
Show All 23 Lines
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4],xmm0[5,6,7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4],xmm0[5,6,7]
; AVX1-NEXT: vpalignr {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15,0,1,2,3,4,5,6,7,8,9] ; AVX1-NEXT: vpalignr {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15,0,1,2,3,4,5,6,7,8,9]
; AVX1-NEXT: vpalignr {{.*#+}} xmm1 = xmm1[10,11,12,13,14,15,0,1,2,3,4,5,6,7,8,9] ; AVX1-NEXT: vpalignr {{.*#+}} xmm1 = xmm1[10,11,12,13,14,15,0,1,2,3,4,5,6,7,8,9]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_05_06_07_00_01_02_03_12_13_14_15_08_09_10_11_12: ; AVX2-LABEL: shuffle_v16i16_05_06_07_00_01_02_03_12_13_14_15_08_09_10_11_12:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4],xmm0[5,6,7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2OR512VL-NEXT: vpalignr {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15,0,1,2,3,4,5,6,7,8,9] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vpalignr {{.*#+}} xmm1 = xmm1[10,11,12,13,14,15,0,1,2,3,4,5,6,7,8,9] ; AVX2-NEXT: vpalignr {{.*#+}} ymm0 = ymm0[10,11,12,13,14,15,0,1,2,3,4,5,6,7,8,9,26,27,28,29,30,31,16,17,18,19,20,21,22,23,24,25]
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0 ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: retq ;
; AVX512VL-LABEL: shuffle_v16i16_05_06_07_00_01_02_03_12_13_14_15_08_09_10_11_12:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [5,6,7,0,1,2,3,12,13,14,15,8,9,10,11,12]
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 12, i32 13, i32 14, i32 15, i32 8, i32 9, i32 10, i32 11, i32 12> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 12, i32 13, i32 14, i32 15, i32 8, i32 9, i32 10, i32 11, i32 12>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_uu_06_uu_uu_01_02_03_uu_uu_14_uu_uu_09_10_11_uu(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_uu_06_uu_uu_01_02_03_uu_uu_14_uu_uu_09_10_11_uu(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_uu_06_uu_uu_01_02_03_uu_uu_14_uu_uu_09_10_11_uu: ; AVX1-LABEL: shuffle_v16i16_uu_06_uu_uu_01_02_03_uu_uu_14_uu_uu_09_10_11_uu:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vpalignr {{.*#+}} xmm1 = xmm0[10,11,12,13,14,15,0,1,2,3,4,5,6,7,8,9] ; AVX1-NEXT: vpalignr {{.*#+}} xmm1 = xmm0[10,11,12,13,14,15,0,1,2,3,4,5,6,7,8,9]
Show All 38 Lines
; AVX1-NEXT: vpsrldq {{.*#+}} xmm1 = xmm1[6,7,8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero ; AVX1-NEXT: vpsrldq {{.*#+}} xmm1 = xmm1[6,7,8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm1[0,1,2,3,4],xmm0[5,6,7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm1[0,1,2,3,4],xmm0[5,6,7]
; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2-LABEL: shuffle_v16i16_19_20_21_22_23_00_01_10_27_28_29_30_31_08_09_10: ; AVX2-LABEL: shuffle_v16i16_19_20_21_22_23_00_01_10_27_28_29_30_31_08_09_10:
; AVX2: # BB#0: ; AVX2: # BB#0:
; AVX2-NEXT: vpblendw {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3,4,5,6,7],ymm0[8,9,10],ymm1[11,12,13,14,15] ; AVX2-NEXT: vpblendw {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3,4,5,6,7],ymm0[8,9,10],ymm1[11,12,13,14,15]
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2],xmm0[3,4,5,6,7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2-NEXT: vpalignr {{.*#+}} xmm0 = xmm0[6,7,8,9,10,11,12,13,14,15,0,1,2,3,4,5] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2-NEXT: vpalignr {{.*#+}} xmm1 = xmm1[6,7,8,9,10,11,12,13,14,15,0,1,2,3,4,5] ; AVX2-NEXT: vpalignr {{.*#+}} ymm0 = ymm0[6,7,8,9,10,11,12,13,14,15,0,1,2,3,4,5,22,23,24,25,26,27,28,29,30,31,16,17,18,19,20,21]
; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
; AVX2-NEXT: retq ; AVX2-NEXT: retq
; ;
; AVX512VL-LABEL: shuffle_v16i16_19_20_21_22_23_00_01_10_27_28_29_30_31_08_09_10: ; AVX512VL-LABEL: shuffle_v16i16_19_20_21_22_23_00_01_10_27_28_29_30_31_08_09_10:
; AVX512VL: # BB#0: ; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [3,4,5,6,7,16,17,26,11,12,13,14,15,24,25,26] ; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [3,4,5,6,7,16,17,26,11,12,13,14,15,24,25,26]
; AVX512VL-NEXT: vpermi2w %ymm0, %ymm1, %ymm2 ; AVX512VL-NEXT: vpermi2w %ymm0, %ymm1, %ymm2
; AVX512VL-NEXT: vmovdqa %ymm2, %ymm0 ; AVX512VL-NEXT: vmovdqa %ymm2, %ymm0
; AVX512VL-NEXT: retq ; AVX512VL-NEXT: retq
Show All 24 Lines
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2],xmm0[3,4,5,6,7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2],xmm0[3,4,5,6,7]
; AVX1-NEXT: vpalignr {{.*#+}} xmm0 = xmm0[6,7,8,9,10,11,12,13,14,15,0,1,2,3,4,5] ; AVX1-NEXT: vpalignr {{.*#+}} xmm0 = xmm0[6,7,8,9,10,11,12,13,14,15,0,1,2,3,4,5]
; AVX1-NEXT: vpalignr {{.*#+}} xmm1 = xmm1[6,7,8,9,10,11,12,13,14,15,0,1,2,3,4,5] ; AVX1-NEXT: vpalignr {{.*#+}} xmm1 = xmm1[6,7,8,9,10,11,12,13,14,15,0,1,2,3,4,5]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2OR512VL-LABEL: shuffle_v16i16_03_04_05_06_07_00_01_10_11_12_13_14_15_08_09_10: ; AVX2-LABEL: shuffle_v16i16_03_04_05_06_07_00_01_10_11_12_13_14_15_08_09_10:
; AVX2OR512VL: # BB#0: ; AVX2: # BB#0:
; AVX2OR512VL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2OR512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2],xmm0[3,4,5,6,7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2OR512VL-NEXT: vpalignr {{.*#+}} xmm0 = xmm0[6,7,8,9,10,11,12,13,14,15,0,1,2,3,4,5] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2OR512VL-NEXT: vpalignr {{.*#+}} xmm1 = xmm1[6,7,8,9,10,11,12,13,14,15,0,1,2,3,4,5] ; AVX2-NEXT: vpalignr {{.*#+}} ymm0 = ymm0[6,7,8,9,10,11,12,13,14,15,0,1,2,3,4,5,22,23,24,25,26,27,28,29,30,31,16,17,18,19,20,21]
; AVX2OR512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0 ; AVX2-NEXT: retq
; AVX2OR512VL-NEXT: retq ;
; AVX512VL-LABEL: shuffle_v16i16_03_04_05_06_07_00_01_10_11_12_13_14_15_08_09_10:
; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm1 = [3,4,5,6,7,0,1,10,11,12,13,14,15,8,9,10]
; AVX512VL-NEXT: vpermw %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 8, i32 9, i32 10> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 8, i32 9, i32 10>
ret <16 x i16> %shuffle ret <16 x i16> %shuffle
} }
define <16 x i16> @shuffle_v16i16_uu_04_05_06_uu_uu_01_uu_uu_12_13_14_uu_uu_09_uu(<16 x i16> %a, <16 x i16> %b) { define <16 x i16> @shuffle_v16i16_uu_04_05_06_uu_uu_01_uu_uu_12_13_14_uu_uu_09_uu(<16 x i16> %a, <16 x i16> %b) {
; AVX1-LABEL: shuffle_v16i16_uu_04_05_06_uu_uu_01_uu_uu_12_13_14_uu_uu_09_uu: ; AVX1-LABEL: shuffle_v16i16_uu_04_05_06_uu_uu_01_uu_uu_12_13_14_uu_uu_09_uu:
; AVX1: # BB#0: ; AVX1: # BB#0:
; AVX1-NEXT: vpalignr {{.*#+}} xmm1 = xmm0[6,7,8,9,10,11,12,13,14,15,0,1,2,3,4,5] ; AVX1-NEXT: vpalignr {{.*#+}} xmm1 = xmm0[6,7,8,9,10,11,12,13,14,15,0,1,2,3,4,5]
Show All 38 Lines
; AVX1-NEXT: vpsrldq {{.*#+}} xmm0 = xmm0[6,7,8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero ; AVX1-NEXT: vpsrldq {{.*#+}} xmm0 = xmm0[6,7,8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4],xmm1[5,6,7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4],xmm1[5,6,7]
; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2-LABEL: shuffle_v16i16_03_04_05_06_07_16_17_26_11_12_13_14_15_24_25_26: ; AVX2-LABEL: shuffle_v16i16_03_04_05_06_07_16_17_26_11_12_13_14_15_24_25_26:
; AVX2: # BB#0: ; AVX2: # BB#0:
; AVX2-NEXT: vpblendw {{.*#+}} ymm0 = ymm1[0,1,2],ymm0[3,4,5,6,7],ymm1[8,9,10],ymm0[11,12,13,14,15] ; AVX2-NEXT: vpblendw {{.*#+}} ymm0 = ymm1[0,1,2],ymm0[3,4,5,6,7],ymm1[8,9,10],ymm0[11,12,13,14,15]
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2],xmm0[3,4,5,6,7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2-NEXT: vpalignr {{.*#+}} xmm0 = xmm0[6,7,8,9,10,11,12,13,14,15,0,1,2,3,4,5] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2-NEXT: vpalignr {{.*#+}} xmm1 = xmm1[6,7,8,9,10,11,12,13,14,15,0,1,2,3,4,5] ; AVX2-NEXT: vpalignr {{.*#+}} ymm0 = ymm0[6,7,8,9,10,11,12,13,14,15,0,1,2,3,4,5,22,23,24,25,26,27,28,29,30,31,16,17,18,19,20,21]
; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
; AVX2-NEXT: retq ; AVX2-NEXT: retq
; ;
; AVX512VL-LABEL: shuffle_v16i16_03_04_05_06_07_16_17_26_11_12_13_14_15_24_25_26: ; AVX512VL-LABEL: shuffle_v16i16_03_04_05_06_07_16_17_26_11_12_13_14_15_24_25_26:
; AVX512VL: # BB#0: ; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [3,4,5,6,7,16,17,26,11,12,13,14,15,24,25,26] ; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [3,4,5,6,7,16,17,26,11,12,13,14,15,24,25,26]
; AVX512VL-NEXT: vpermt2w %ymm1, %ymm2, %ymm0 ; AVX512VL-NEXT: vpermt2w %ymm1, %ymm2, %ymm0
; AVX512VL-NEXT: retq ; AVX512VL-NEXT: retq
%shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 26, i32 11, i32 12, i32 13, i32 14, i32 15, i32 24, i32 25, i32 26> %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 26, i32 11, i32 12, i32 13, i32 14, i32 15, i32 24, i32 25, i32 26>
Show All 27 Lines
; AVX1-NEXT: vpblendw {{.*#+}} xmm1 = xmm1[0,1,2,3],xmm3[4,5,6,7] ; AVX1-NEXT: vpblendw {{.*#+}} xmm1 = xmm1[0,1,2,3],xmm3[4,5,6,7]
; AVX1-NEXT: vpalignr {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],xmm1[0,1,2,3,4,5,6,7,8,9] ; AVX1-NEXT: vpalignr {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],xmm1[0,1,2,3,4,5,6,7,8,9]
; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2-LABEL: shuffle_v16i16_05_06_07_16_17_18_19_28_13_14_15_24_25_26_27_28: ; AVX2-LABEL: shuffle_v16i16_05_06_07_16_17_18_19_28_13_14_15_24_25_26_27_28:
; AVX2: # BB#0: ; AVX2: # BB#0:
; AVX2-NEXT: vpblendw {{.*#+}} ymm0 = ymm1[0,1,2,3,4],ymm0[5,6,7],ymm1[8,9,10,11,12],ymm0[13,14,15] ; AVX2-NEXT: vpblendw {{.*#+}} ymm0 = ymm1[0,1,2,3,4],ymm0[5,6,7],ymm1[8,9,10,11,12],ymm0[13,14,15]
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4],xmm0[5,6,7] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX2-NEXT: vpalignr {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15,0,1,2,3,4,5,6,7,8,9] ; AVX2-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
; AVX2-NEXT: vpalignr {{.*#+}} xmm1 = xmm1[10,11,12,13,14,15,0,1,2,3,4,5,6,7,8,9] ; AVX2-NEXT: vpalignr {{.*#+}} ymm0 = ymm0[10,11,12,13,14,15,0,1,2,3,4,5,6,7,8,9,26,27,28,29,30,31,16,17,18,19,20,21,22,23,24,25]
; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
; AVX2-NEXT: retq ; AVX2-NEXT: retq
; ;
; AVX512VL-LABEL: shuffle_v16i16_05_06_07_16_17_18_19_28_13_14_15_24_25_26_27_28: ; AVX512VL-LABEL: shuffle_v16i16_05_06_07_16_17_18_19_28_13_14_15_24_25_26_27_28:
; AVX512VL: # BB#0: ; AVX512VL: # BB#0:
; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [21,22,23,0,1,2,3,12,29,30,31,8,9,10,11,12] ; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm2 = [21,22,23,0,1,2,3,12,29,30,31,8,9,10,11,12]
; AVX512VL-NEXT: vpermi2w %ymm0, %ymm1, %ymm2 ; AVX512VL-NEXT: vpermi2w %ymm0, %ymm1, %ymm2
; AVX512VL-NEXT: vmovdqa %ymm2, %ymm0 ; AVX512VL-NEXT: vmovdqa %ymm2, %ymm0
; AVX512VL-NEXT: retq ; AVX512VL-NEXT: retq
▲ Show 20 Lines • Show All 382 Lines • Show Last 20 Lines

test/CodeGen/X86/vector-shuffle-512-v32.ll

Show First 20 Lines • Show All 57 Lines • ▼ Show 20 Lines
; SKX-NEXT: retq ; SKX-NEXT: retq
%c = shufflevector <32 x i16> %a, <32 x i16> undef, <32 x i32> <i32 2, i32 5, i32 undef, i32 undef, i32 7, i32 undef, i32 10, i32 1, i32 0, i32 5, i32 undef, i32 4, i32 7, i32 undef, i32 10, i32 1, i32 2, i32 5, i32 undef, i32 undef, i32 7, i32 undef, i32 10, i32 1, i32 0, i32 5, i32 undef, i32 4, i32 7, i32 undef, i32 10, i32 31> %c = shufflevector <32 x i16> %a, <32 x i16> undef, <32 x i32> <i32 2, i32 5, i32 undef, i32 undef, i32 7, i32 undef, i32 10, i32 1, i32 0, i32 5, i32 undef, i32 4, i32 7, i32 undef, i32 10, i32 1, i32 2, i32 5, i32 undef, i32 undef, i32 7, i32 undef, i32 10, i32 1, i32 0, i32 5, i32 undef, i32 4, i32 7, i32 undef, i32 10, i32 31>
ret <32 x i16> %c ret <32 x i16> %c
} }
define <32 x i16> @shuffle_v32i16_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_18_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_38(<32 x i16> %a, <32 x i16> %b) { define <32 x i16> @shuffle_v32i16_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_18_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_38(<32 x i16> %a, <32 x i16> %b) {
; KNL-LABEL: shuffle_v32i16_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_18_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_38: ; KNL-LABEL: shuffle_v32i16_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_18_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_38:
; KNL: ## BB#0: ; KNL: ## BB#0:
; KNL-NEXT: vextracti128 $1, %ymm1, %xmm2 ; KNL-NEXT: vperm2i128 {{.*#+}} ymm2 = ymm1[2,3,0,1]
; KNL-NEXT: vpunpckhwd {{.*#+}} xmm1 = xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7] ; KNL-NEXT: vpblendw {{.*#+}} ymm2 = ymm1[0,1,2,3],ymm2[4,5],ymm1[6],ymm2[7],ymm1[8,9,10,11],ymm2[12,13],ymm1[14],ymm2[15]
; KNL-NEXT: vpshufb {{.*#+}} xmm4 = xmm1[8,9,12,13,12,13,10,11,0,1,4,5,4,5,0,1] ; KNL-NEXT: vpshufb {{.*#+}} ymm1 = ymm2[u,u,14,15,u,u,12,13,u,u,10,11,u,u,8,9,u,u,22,23,u,u,20,21,u,u,18,19,u,u,u,u]
; KNL-NEXT: vpshufd {{.*#+}} xmm1 = xmm2[0,1,0,3] ; KNL-NEXT: vperm2i128 {{.*#+}} ymm4 = ymm0[2,3,0,1]
; KNL-NEXT: vpshuflw {{.*#+}} xmm2 = xmm1[0,3,2,2,4,5,6,7] ; KNL-NEXT: vpblendw {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm4[5,6,7],ymm0[8,9,10,11,12],ymm4[13,14,15]
; KNL-NEXT: vextracti128 $1, %ymm0, %xmm1 ; KNL-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[14,15,u,u,12,13,u,u,10,11,u,u,8,9,u,u,22,23,u,u,20,21,u,u,18,19,u,u,16,17,u,u]
; KNL-NEXT: vpunpckhwd {{.*#+}} xmm0 = xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7] ; KNL-NEXT: vpblendw {{.*#+}} ymm1 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7],ymm0[8],ymm1[9],ymm0[10],ymm1[11],ymm0[12],ymm1[13],ymm0[14],ymm1[15]
; KNL-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[12,13,10,11,8,9,14,15,4,5,2,3,2,3,6,7]
; KNL-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[6,7,2,3,4,5,6,7,2,3,2,3,0,1,14,15]
; KNL-NEXT: vinserti128 $1, %xmm2, %ymm4, %ymm5
; KNL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
; KNL-NEXT: vpblendw {{.*#+}} ymm1 = ymm0[0],ymm5[1],ymm0[2],ymm5[3],ymm0[4],ymm5[5],ymm0[6],ymm5[7],ymm0[8],ymm5[9],ymm0[10],ymm5[11],ymm0[12],ymm5[13],ymm0[14],ymm5[15]
; KNL-NEXT: vextracti128 $1, %ymm3, %xmm3 ; KNL-NEXT: vextracti128 $1, %ymm3, %xmm3
; KNL-NEXT: vpbroadcastw %xmm3, %ymm3 ; KNL-NEXT: vpbroadcastw %xmm3, %ymm3
; KNL-NEXT: vmovdqa {{.*#+}} ymm5 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0] ; KNL-NEXT: vmovdqa {{.*#+}} ymm4 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0]
; KNL-NEXT: vpblendvb %ymm5, %ymm1, %ymm3, %ymm1 ; KNL-NEXT: vpblendvb %ymm4, %ymm1, %ymm3, %ymm1
; KNL-NEXT: vpshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,4,5,6,4] ; KNL-NEXT: vpshufb {{.*#+}} ymm2 = ymm2[u,u,14,15,u,u,12,13,u,u,10,11,u,u,8,9,u,u,22,23,u,u,20,21,u,u,18,19,u,u,16,17]
; KNL-NEXT: vinserti128 $1, %xmm2, %ymm4, %ymm2
; KNL-NEXT: vpblendw {{.*#+}} ymm0 = ymm0[0],ymm2[1],ymm0[2],ymm2[3],ymm0[4],ymm2[5],ymm0[6],ymm2[7],ymm0[8],ymm2[9],ymm0[10],ymm2[11],ymm0[12],ymm2[13],ymm0[14],ymm2[15] ; KNL-NEXT: vpblendw {{.*#+}} ymm0 = ymm0[0],ymm2[1],ymm0[2],ymm2[3],ymm0[4],ymm2[5],ymm0[6],ymm2[7],ymm0[8],ymm2[9],ymm0[10],ymm2[11],ymm0[12],ymm2[13],ymm0[14],ymm2[15]
; KNL-NEXT: retq ; KNL-NEXT: retq
; ;
; SKX-LABEL: shuffle_v32i16_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_18_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_38: ; SKX-LABEL: shuffle_v32i16_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_18_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_38:
; SKX: ## BB#0: ; SKX: ## BB#0:
; SKX-NEXT: vmovdqa64 {{.*#+}} zmm2 = [15,31,14,22,13,29,4,28,11,27,10,26,9,25,8,24,15,31,14,22,13,29,4,28,11,27,10,26,9,25,8,56] ; SKX-NEXT: vmovdqa64 {{.*#+}} zmm2 = [15,31,14,22,13,29,4,28,11,27,10,26,9,25,8,24,15,31,14,22,13,29,4,28,11,27,10,26,9,25,8,56]
; SKX-NEXT: vpermt2w %zmm1, %zmm2, %zmm0 ; SKX-NEXT: vpermt2w %zmm1, %zmm2, %zmm0
; SKX-NEXT: retq ; SKX-NEXT: retq
▲ Show 20 Lines • Show All 269 Lines • Show Last 20 Lines

test/CodeGen/X86/vector-shuffle-combining-avx2.ll

Show First 20 Lines • Show All 877 Lines • ▼ Show 20 Lines
; X64-NEXT: retq ; X64-NEXT: retq
%1 = tail call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> <i8 15, i8 14, i8 13, i8 12, i8 11, i8 10, i8 9, i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 -1, i8 -2, i8 -3, i8 -4, i8 -5, i8 -6, i8 -7, i8 -8, i8 -9, i8 -10, i8 -11, i8 -12, i8 -13, i8 -14, i8 -15>, <32 x i8> <i8 1, i8 -1, i8 -1, i8 -1, i8 undef, i8 undef, i8 -1, i8 -1, i8 15, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 7, i8 6, i8 1, i8 -1, i8 -1, i8 -1, i8 undef, i8 undef, i8 -1, i8 -1, i8 15, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 7, i8 6>) %1 = tail call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> <i8 15, i8 14, i8 13, i8 12, i8 11, i8 10, i8 9, i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 -1, i8 -2, i8 -3, i8 -4, i8 -5, i8 -6, i8 -7, i8 -8, i8 -9, i8 -10, i8 -11, i8 -12, i8 -13, i8 -14, i8 -15>, <32 x i8> <i8 1, i8 -1, i8 -1, i8 -1, i8 undef, i8 undef, i8 -1, i8 -1, i8 15, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 7, i8 6, i8 1, i8 -1, i8 -1, i8 -1, i8 undef, i8 undef, i8 -1, i8 -1, i8 15, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 7, i8 6>)
ret <32 x i8> %1 ret <32 x i8> %1
} }
define <32 x i8> @PR27320(<8 x i32> %a0) { define <32 x i8> @PR27320(<8 x i32> %a0) {
; X32-LABEL: PR27320: ; X32-LABEL: PR27320:
; X32: # BB#0: ; X32: # BB#0:
; X32-NEXT: vpshufb {{.*#+}} xmm1 = xmm0[12,13,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero ; X32-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,1]
; X32-NEXT: vextracti128 $1, %ymm0, %xmm2 ; X32-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,1,2,3,4,4,5,6,7,7,8,9,10,10,11,28,29,29,30,31,16,16,17,18,19,19,20,21,22,22,23]
; X32-NEXT: vpshufb {{.*#+}} xmm2 = zero,zero,zero,zero,zero,xmm2[0,0,1,2,3,3,4,5,6,6,7]
; X32-NEXT: vpor %xmm1, %xmm2, %xmm1
; X32-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,1,2,3,4,4,5,6,7,7,8,9,10,10,11]
; X32-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
; X32-NEXT: retl ; X32-NEXT: retl
; ;
; X64-LABEL: PR27320: ; X64-LABEL: PR27320:
; X64: # BB#0: ; X64: # BB#0:
; X64-NEXT: vpshufb {{.*#+}} xmm1 = xmm0[12,13,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero ; X64-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,1]
; X64-NEXT: vextracti128 $1, %ymm0, %xmm2 ; X64-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,1,2,3,4,4,5,6,7,7,8,9,10,10,11,28,29,29,30,31,16,16,17,18,19,19,20,21,22,22,23]
; X64-NEXT: vpshufb {{.*#+}} xmm2 = zero,zero,zero,zero,zero,xmm2[0,0,1,2,3,3,4,5,6,6,7]
; X64-NEXT: vpor %xmm1, %xmm2, %xmm1
; X64-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,1,2,3,4,4,5,6,7,7,8,9,10,10,11]
; X64-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
; X64-NEXT: retq ; X64-NEXT: retq
%1 = shufflevector <8 x i32> %a0, <8 x i32> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 undef, i32 3, i32 4, i32 5, i32 undef> %1 = shufflevector <8 x i32> %a0, <8 x i32> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 undef, i32 3, i32 4, i32 5, i32 undef>
%2 = bitcast <8 x i32> %1 to <32 x i8> %2 = bitcast <8 x i32> %1 to <32 x i8>
%3 = shufflevector <32 x i8> %2, <32 x i8> undef, <32 x i32> <i32 0, i32 1, i32 1, i32 2, i32 3, i32 4, i32 4, i32 5, i32 6, i32 7, i32 7, i32 8, i32 9, i32 10, i32 10, i32 11, i32 16, i32 17, i32 17, i32 18, i32 19, i32 20, i32 20, i32 21, i32 22, i32 23, i32 23, i32 24, i32 25, i32 26, i32 26, i32 27> %3 = shufflevector <32 x i8> %2, <32 x i8> undef, <32 x i32> <i32 0, i32 1, i32 1, i32 2, i32 3, i32 4, i32 4, i32 5, i32 6, i32 7, i32 7, i32 8, i32 9, i32 10, i32 10, i32 11, i32 16, i32 17, i32 17, i32 18, i32 19, i32 20, i32 20, i32 21, i32 22, i32 23, i32 23, i32 24, i32 25, i32 26, i32 26, i32 27>
ret <32 x i8> %3 ret <32 x i8> %3
} }

test/CodeGen/X86/vector-shuffle-v48.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=x86_64-pc-linux -mattr=+avx2 < %s | FileCheck %s ; RUN: llc -mtriple=x86_64-pc-linux -mattr=+avx2 < %s | FileCheck %s
define <32 x i8> @foo(<48 x i8>* %x0, <16 x i32> %x1, <16 x i32> %x2) { define <32 x i8> @foo(<48 x i8>* %x0, <16 x i32> %x1, <16 x i32> %x2) {
; CHECK-LABEL: foo: ; CHECK-LABEL: foo:
; CHECK: # BB#0: ; CHECK: # BB#0:
; CHECK-NEXT: vmovdqu (%rdi), %ymm0 ; CHECK-NEXT: vmovdqu 32(%rdi), %xmm0
; CHECK-NEXT: vmovdqu 32(%rdi), %xmm1 ; CHECK-NEXT: vmovdqu (%rdi), %ymm1
; CHECK-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[u,u,u,u,u,u,1,2,4,5,7,8,10,11,13,14] ; CHECK-NEXT: vperm2i128 {{.*#+}} ymm2 = ymm1[2,3,0,1]
; CHECK-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm1 ; CHECK-NEXT: vpshufb {{.*#+}} xmm2 = xmm2[u,u,u,u,u,u,u,u,u,u,u,0,2,3,5,6]
; CHECK-NEXT: vextracti128 $1, %ymm0, %xmm2 ; CHECK-NEXT: vpshufb {{.*#+}} ymm1 = ymm1[0,1,3,4,6,7,9,10,12,13,15,u,u,u,u,u,24,25,27,28,30,31,u,u,u,u,u,u,u,u,u,u]
; CHECK-NEXT: vpshufb {{.*#+}} xmm3 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm2[0,2,3,5,6] ; CHECK-NEXT: vmovdqa {{.*#+}} ymm3 = <255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,255,255,255,255,255,255,u,u,u,u,u,u,u,u,u,u>
; CHECK-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,3,4,6,7,9,10,12,13,15],zero,zero,zero,zero,zero ; CHECK-NEXT: vpblendvb %ymm3, %ymm1, %ymm2, %ymm1
; CHECK-NEXT: vpor %xmm3, %xmm0, %xmm0 ; CHECK-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[u,u,u,u,u,u,1,2,4,5,7,8,10,11,13,14]
; CHECK-NEXT: vpshufb {{.*#+}} xmm2 = xmm2[8,9,11,12,14,15,u,u,u,u,u,u,u,u,u,u] ; CHECK-NEXT: vinserti128 $1, %xmm0, %ymm0, %ymm0
; CHECK-NEXT: vinserti128 $1, %xmm2, %ymm0, %ymm0
; CHECK-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0] ; CHECK-NEXT: vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0]
; CHECK-NEXT: vpblendvb %ymm2, %ymm0, %ymm1, %ymm0 ; CHECK-NEXT: vpblendvb %ymm2, %ymm1, %ymm0, %ymm0
; CHECK-NEXT: retq ; CHECK-NEXT: retq
%1 = load <48 x i8>, <48 x i8>* %x0, align 1 %1 = load <48 x i8>, <48 x i8>* %x0, align 1
%2 = shufflevector <48 x i8> %1, <48 x i8> undef, <32 x i32> <i32 0, i32 1, i32 3, i32 4, i32 6, i32 7, i32 9, i32 10, i32 12, i32 13, i32 15, i32 16, i32 18, i32 19, i32 21, i32 22, i32 24, i32 25, i32 27, i32 28, i32 30, i32 31, i32 33, i32 34, i32 36, i32 37, i32 39, i32 40, i32 42, i32 43, i32 45, i32 46> %2 = shufflevector <48 x i8> %1, <48 x i8> undef, <32 x i32> <i32 0, i32 1, i32 3, i32 4, i32 6, i32 7, i32 9, i32 10, i32 12, i32 13, i32 15, i32 16, i32 18, i32 19, i32 21, i32 22, i32 24, i32 25, i32 27, i32 28, i32 30, i32 31, i32 33, i32 34, i32 36, i32 37, i32 39, i32 40, i32 42, i32 43, i32 45, i32 46>
ret <32 x i8> %2 ret <32 x i8> %2
} }

test/CodeGen/X86/x86-interleaved-access.ll

Show First 20 Lines • Show All 732 Lines • ▼ Show 20 Lines
; AVX1-NEXT: vpand %xmm1, %xmm4, %xmm3 ; AVX1-NEXT: vpand %xmm1, %xmm4, %xmm3
; AVX1-NEXT: vpand %xmm1, %xmm2, %xmm1 ; AVX1-NEXT: vpand %xmm1, %xmm2, %xmm1
; AVX1-NEXT: vpcmpeqb %xmm3, %xmm1, %xmm1 ; AVX1-NEXT: vpcmpeqb %xmm3, %xmm1, %xmm1
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2-LABEL: interleaved_load_vf32_i8_stride4: ; AVX2-LABEL: interleaved_load_vf32_i8_stride4:
; AVX2: # BB#0: ; AVX2: # BB#0:
; AVX2-NEXT: vmovdqa (%rdi), %ymm9 ; AVX2-NEXT: vmovdqa (%rdi), %ymm11
; AVX2-NEXT: vmovdqa 32(%rdi), %ymm11 ; AVX2-NEXT: vmovdqa 32(%rdi), %ymm1
; AVX2-NEXT: vmovdqa 64(%rdi), %ymm14 ; AVX2-NEXT: vmovdqa 64(%rdi), %ymm4
; AVX2-NEXT: vmovdqa 96(%rdi), %ymm3 ; AVX2-NEXT: vmovdqa 96(%rdi), %ymm6
; AVX2-NEXT: vmovdqa {{.*#+}} ymm4 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
; AVX2-NEXT: vpshufb %ymm4, %ymm3, %ymm5 ; AVX2-NEXT: vpshufb %ymm2, %ymm6, %ymm3
; AVX2-NEXT: vpermq {{.*#+}} ymm5 = ymm5[0,2,2,3] ; AVX2-NEXT: vpermq {{.*#+}} ymm3 = ymm3[0,2,2,3]
; AVX2-NEXT: vmovdqa {{.*#+}} xmm6 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u> ; AVX2-NEXT: vmovdqa {{.*#+}} xmm5 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
; AVX2-NEXT: vpshufb %xmm6, %xmm5, %xmm5 ; AVX2-NEXT: vpshufb %xmm5, %xmm3, %xmm3
; AVX2-NEXT: vpshufb %ymm4, %ymm14, %ymm7 ; AVX2-NEXT: vpshufb %ymm2, %ymm4, %ymm7
; AVX2-NEXT: vpermq {{.*#+}} ymm7 = ymm7[0,2,2,3] ; AVX2-NEXT: vpermq {{.*#+}} ymm7 = ymm7[0,2,2,3]
; AVX2-NEXT: vpshufb %xmm6, %xmm7, %xmm7 ; AVX2-NEXT: vpshufb %xmm5, %xmm7, %xmm7
; AVX2-NEXT: vpunpcklqdq {{.*#+}} xmm5 = xmm7[0],xmm5[0] ; AVX2-NEXT: vpunpcklqdq {{.*#+}} xmm3 = xmm7[0],xmm3[0]
; AVX2-NEXT: vpshufb %ymm4, %ymm11, %ymm7 ; AVX2-NEXT: vpshufb %ymm2, %ymm1, %ymm7
; AVX2-NEXT: vpermq {{.*#+}} ymm7 = ymm7[0,2,2,3] ; AVX2-NEXT: vpermq {{.*#+}} ymm7 = ymm7[0,2,2,3]
; AVX2-NEXT: vpshufb %xmm6, %xmm7, %xmm7 ; AVX2-NEXT: vpshufb %xmm5, %xmm7, %xmm7
; AVX2-NEXT: vpshufb %ymm4, %ymm9, %ymm4 ; AVX2-NEXT: vpshufb %ymm2, %ymm11, %ymm2
; AVX2-NEXT: vpermq {{.*#+}} ymm4 = ymm4[0,2,2,3] ; AVX2-NEXT: vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
; AVX2-NEXT: vpshufb %xmm6, %xmm4, %xmm4 ; AVX2-NEXT: vpshufb %xmm5, %xmm2, %xmm2
; AVX2-NEXT: vpunpcklqdq {{.*#+}} xmm4 = xmm4[0],xmm7[0] ; AVX2-NEXT: vpunpcklqdq {{.*#+}} xmm2 = xmm2[0],xmm7[0]
; AVX2-NEXT: vinserti128 $1, %xmm5, %ymm4, %ymm8 ; AVX2-NEXT: vinserti128 $1, %xmm3, %ymm2, %ymm8
; AVX2-NEXT: vextracti128 $1, %ymm3, %xmm10 ; AVX2-NEXT: vextracti128 $1, %ymm1, %xmm9
; AVX2-NEXT: vmovdqa {{.*#+}} xmm7 = <u,u,u,u,1,5,9,13,u,u,u,u,u,u,u,u> ; AVX2-NEXT: vmovdqa {{.*#+}} xmm7 = <u,u,u,u,1,5,9,13,u,u,u,u,u,u,u,u>
; AVX2-NEXT: vpshufb %xmm7, %xmm10, %xmm5 ; AVX2-NEXT: vpshufb %xmm7, %xmm9, %xmm3
; AVX2-NEXT: vpshufb %xmm7, %xmm3, %xmm6 ; AVX2-NEXT: vpshufb %xmm7, %xmm1, %xmm5
; AVX2-NEXT: vpunpckldq {{.*#+}} xmm5 = xmm6[0],xmm5[0],xmm6[1],xmm5[1] ; AVX2-NEXT: vpunpckldq {{.*#+}} xmm5 = xmm5[0],xmm3[0],xmm5[1],xmm3[1]
; AVX2-NEXT: vinserti128 $1, %xmm5, %ymm0, %ymm6 ; AVX2-NEXT: vextracti128 $1, %ymm11, %xmm10
; AVX2-NEXT: vextracti128 $1, %ymm14, %xmm13 ; AVX2-NEXT: vmovdqa {{.*#+}} xmm2 = <1,5,9,13,u,u,u,u,u,u,u,u,u,u,u,u>
; AVX2-NEXT: vmovdqa {{.*#+}} xmm0 = <1,5,9,13,u,u,u,u,u,u,u,u,u,u,u,u> ; AVX2-NEXT: vpshufb %xmm2, %xmm10, %xmm3
; AVX2-NEXT: vpshufb %xmm0, %xmm13, %xmm4 ; AVX2-NEXT: vpshufb %xmm2, %xmm11, %xmm0
; AVX2-NEXT: vpshufb %xmm0, %xmm14, %xmm1 ; AVX2-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
; AVX2-NEXT: vpunpckldq {{.*#+}} xmm1 = xmm1[0],xmm4[0],xmm1[1],xmm4[1] ; AVX2-NEXT: vpblendd {{.*#+}} xmm12 = xmm0[0,1],xmm5[2,3]
; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm1 ; AVX2-NEXT: vextracti128 $1, %ymm6, %xmm13
; AVX2-NEXT: vpblendd {{.*#+}} ymm12 = ymm1[0,1,2,3,4,5],ymm6[6,7] ; AVX2-NEXT: vpshufb %xmm7, %xmm13, %xmm3
; AVX2-NEXT: vextracti128 $1, %ymm11, %xmm6 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm6 = ymm6[2,3,0,1]
; AVX2-NEXT: vpshufb %xmm7, %xmm6, %xmm4 ; AVX2-NEXT: vextracti128 $1, %ymm6, %xmm14
; AVX2-NEXT: vpshufb %xmm7, %xmm11, %xmm7 ; AVX2-NEXT: vpshufb %xmm7, %xmm14, %xmm7
; AVX2-NEXT: vpunpckldq {{.*#+}} xmm4 = xmm7[0],xmm4[0],xmm7[1],xmm4[1] ; AVX2-NEXT: vpunpckldq {{.*#+}} xmm3 = xmm7[0],xmm3[0],xmm7[1],xmm3[1]
; AVX2-NEXT: vextracti128 $1, %ymm9, %xmm7 ; AVX2-NEXT: vinserti128 $1, %xmm3, %ymm0, %ymm3
; AVX2-NEXT: vpshufb %xmm0, %xmm7, %xmm1 ; AVX2-NEXT: vextracti128 $1, %ymm4, %xmm7
; AVX2-NEXT: vpshufb %xmm0, %xmm9, %xmm0 ; AVX2-NEXT: vpshufb %xmm2, %xmm7, %xmm0
; AVX2-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm4 = ymm4[2,3,0,1]
; AVX2-NEXT: vpblendd {{.*#+}} xmm0 = xmm0[0,1],xmm4[2,3] ; AVX2-NEXT: vextracti128 $1, %ymm4, %xmm4
; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm12[4,5,6,7] ; AVX2-NEXT: vpshufb %xmm2, %xmm4, %xmm2
; AVX2-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm2[0],xmm0[0],xmm2[1],xmm0[1]
; AVX2-NEXT: vinserti128 $1, %xmm0, %ymm0, %ymm0
; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm3[6,7]
; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm12[0,1,2,3],ymm0[4,5,6,7]
; AVX2-NEXT: vpcmpeqb %ymm0, %ymm8, %ymm8 ; AVX2-NEXT: vpcmpeqb %ymm0, %ymm8, %ymm8
; AVX2-NEXT: vmovdqa {{.*#+}} xmm0 = <u,u,u,u,2,6,10,14,u,u,u,u,u,u,u,u> ; AVX2-NEXT: vmovdqa {{.*#+}} xmm0 = <u,u,u,u,2,6,10,14,u,u,u,u,u,u,u,u>
; AVX2-NEXT: vpshufb %xmm0, %xmm10, %xmm1 ; AVX2-NEXT: vpshufb %xmm0, %xmm9, %xmm2
; AVX2-NEXT: vpshufb %xmm0, %xmm3, %xmm4 ; AVX2-NEXT: vpshufb %xmm0, %xmm1, %xmm3
; AVX2-NEXT: vpunpckldq {{.*#+}} xmm1 = xmm4[0],xmm1[0],xmm4[1],xmm1[1]
; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm1
; AVX2-NEXT: vmovdqa {{.*#+}} xmm4 = <2,6,10,14,u,u,u,u,u,u,u,u,u,u,u,u>
; AVX2-NEXT: vpshufb %xmm4, %xmm13, %xmm5
; AVX2-NEXT: vpshufb %xmm4, %xmm14, %xmm2
; AVX2-NEXT: vpunpckldq {{.*#+}} xmm2 = xmm2[0],xmm5[0],xmm2[1],xmm5[1]
; AVX2-NEXT: vinserti128 $1, %xmm2, %ymm0, %ymm2
; AVX2-NEXT: vpblendd {{.*#+}} ymm1 = ymm2[0,1,2,3,4,5],ymm1[6,7]
; AVX2-NEXT: vpshufb %xmm0, %xmm6, %xmm2
; AVX2-NEXT: vpshufb %xmm0, %xmm11, %xmm0
; AVX2-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
; AVX2-NEXT: vpshufb %xmm4, %xmm7, %xmm2
; AVX2-NEXT: vpshufb %xmm4, %xmm9, %xmm4
; AVX2-NEXT: vpunpckldq {{.*#+}} xmm2 = xmm4[0],xmm2[0],xmm4[1],xmm2[1]
; AVX2-NEXT: vpblendd {{.*#+}} xmm0 = xmm2[0,1],xmm0[2,3]
; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
; AVX2-NEXT: vmovdqa {{.*#+}} xmm1 = <u,u,u,u,3,7,11,15,u,u,u,u,u,u,u,u>
; AVX2-NEXT: vpshufb %xmm1, %xmm10, %xmm2
; AVX2-NEXT: vpshufb %xmm1, %xmm3, %xmm3
; AVX2-NEXT: vpunpckldq {{.*#+}} xmm2 = xmm3[0],xmm2[0],xmm3[1],xmm2[1] ; AVX2-NEXT: vpunpckldq {{.*#+}} xmm2 = xmm3[0],xmm2[0],xmm3[1],xmm2[1]
; AVX2-NEXT: vinserti128 $1, %xmm2, %ymm0, %ymm2 ; AVX2-NEXT: vmovdqa {{.*#+}} xmm3 = <2,6,10,14,u,u,u,u,u,u,u,u,u,u,u,u>
; AVX2-NEXT: vpshufb %xmm3, %xmm10, %xmm5
; AVX2-NEXT: vpshufb %xmm3, %xmm11, %xmm6
; AVX2-NEXT: vpunpckldq {{.*#+}} xmm5 = xmm6[0],xmm5[0],xmm6[1],xmm5[1]
; AVX2-NEXT: vpblendd {{.*#+}} xmm2 = xmm5[0,1],xmm2[2,3]
; AVX2-NEXT: vpshufb %xmm0, %xmm13, %xmm5
; AVX2-NEXT: vpshufb %xmm0, %xmm14, %xmm0
; AVX2-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm5[0],xmm0[1],xmm5[1]
; AVX2-NEXT: vinserti128 $1, %xmm0, %ymm0, %ymm0
; AVX2-NEXT: vpshufb %xmm3, %xmm7, %xmm5
; AVX2-NEXT: vpshufb %xmm3, %xmm4, %xmm3
; AVX2-NEXT: vpunpckldq {{.*#+}} xmm3 = xmm3[0],xmm5[0],xmm3[1],xmm5[1]
; AVX2-NEXT: vinserti128 $1, %xmm3, %ymm0, %ymm3
; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm3[0,1,2,3,4,5],ymm0[6,7]
; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm2[0,1,2,3],ymm0[4,5,6,7]
; AVX2-NEXT: vmovdqa {{.*#+}} xmm2 = <u,u,u,u,3,7,11,15,u,u,u,u,u,u,u,u>
; AVX2-NEXT: vpshufb %xmm2, %xmm9, %xmm3
; AVX2-NEXT: vpshufb %xmm2, %xmm1, %xmm1
; AVX2-NEXT: vpunpckldq {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1]
; AVX2-NEXT: vmovdqa {{.*#+}} xmm3 = <3,7,11,15,u,u,u,u,u,u,u,u,u,u,u,u> ; AVX2-NEXT: vmovdqa {{.*#+}} xmm3 = <3,7,11,15,u,u,u,u,u,u,u,u,u,u,u,u>
; AVX2-NEXT: vpshufb %xmm3, %xmm13, %xmm4 ; AVX2-NEXT: vpshufb %xmm3, %xmm10, %xmm5
; AVX2-NEXT: vpshufb %xmm3, %xmm14, %xmm5 ; AVX2-NEXT: vpshufb %xmm3, %xmm11, %xmm6
; AVX2-NEXT: vpunpckldq {{.*#+}} xmm4 = xmm5[0],xmm4[0],xmm5[1],xmm4[1] ; AVX2-NEXT: vpunpckldq {{.*#+}} xmm5 = xmm6[0],xmm5[0],xmm6[1],xmm5[1]
; AVX2-NEXT: vinserti128 $1, %xmm4, %ymm0, %ymm4 ; AVX2-NEXT: vpblendd {{.*#+}} xmm1 = xmm5[0,1],xmm1[2,3]
; AVX2-NEXT: vpblendd {{.*#+}} ymm2 = ymm4[0,1,2,3,4,5],ymm2[6,7] ; AVX2-NEXT: vpshufb %xmm2, %xmm13, %xmm5
; AVX2-NEXT: vpshufb %xmm1, %xmm6, %xmm4 ; AVX2-NEXT: vpshufb %xmm2, %xmm14, %xmm2
; AVX2-NEXT: vpshufb %xmm1, %xmm11, %xmm1 ; AVX2-NEXT: vpunpckldq {{.*#+}} xmm2 = xmm2[0],xmm5[0],xmm2[1],xmm5[1]
; AVX2-NEXT: vpunpckldq {{.*#+}} xmm1 = xmm1[0],xmm4[0],xmm1[1],xmm4[1] ; AVX2-NEXT: vinserti128 $1, %xmm2, %ymm0, %ymm2
; AVX2-NEXT: vpshufb %xmm3, %xmm7, %xmm4 ; AVX2-NEXT: vpshufb %xmm3, %xmm7, %xmm5
; AVX2-NEXT: vpshufb %xmm3, %xmm9, %xmm3 ; AVX2-NEXT: vpshufb %xmm3, %xmm4, %xmm3
; AVX2-NEXT: vpunpckldq {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1] ; AVX2-NEXT: vpunpckldq {{.*#+}} xmm3 = xmm3[0],xmm5[0],xmm3[1],xmm5[1]
; AVX2-NEXT: vpblendd {{.*#+}} xmm1 = xmm3[0,1],xmm1[2,3] ; AVX2-NEXT: vinserti128 $1, %xmm3, %ymm0, %ymm3
; AVX2-NEXT: vpblendd {{.*#+}} ymm2 = ymm3[0,1,2,3,4,5],ymm2[6,7]
; AVX2-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm2[4,5,6,7] ; AVX2-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm2[4,5,6,7]
; AVX2-NEXT: vpcmpeqb %ymm1, %ymm0, %ymm0 ; AVX2-NEXT: vpcmpeqb %ymm1, %ymm0, %ymm0
; AVX2-NEXT: vmovdqa {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
; AVX2-NEXT: vpand %ymm1, %ymm8, %ymm2 ; AVX2-NEXT: vpand %ymm1, %ymm8, %ymm2
; AVX2-NEXT: vpand %ymm1, %ymm0, %ymm0 ; AVX2-NEXT: vpand %ymm1, %ymm0, %ymm0
; AVX2-NEXT: vpcmpeqb %ymm0, %ymm2, %ymm0 ; AVX2-NEXT: vpcmpeqb %ymm0, %ymm2, %ymm0
; AVX2-NEXT: retq ; AVX2-NEXT: retq
; ;
; AVX512-LABEL: interleaved_load_vf32_i8_stride4: ; AVX512-LABEL: interleaved_load_vf32_i8_stride4:
; AVX512: # BB#0: ; AVX512: # BB#0:
; AVX512-NEXT: vmovdqa64 (%rdi), %zmm0 ; AVX512-NEXT: vmovdqa64 (%rdi), %zmm0
; AVX512-NEXT: vmovdqa64 64(%rdi), %zmm1 ; AVX512-NEXT: vmovdqa64 64(%rdi), %zmm7
; AVX512-NEXT: vpmovdw %zmm0, %ymm2 ; AVX512-NEXT: vpmovdw %zmm0, %ymm1
; AVX512-NEXT: vpmovdw %zmm1, %ymm3 ; AVX512-NEXT: vpmovdw %zmm7, %ymm2
; AVX512-NEXT: vinserti64x4 $1, %ymm3, %zmm2, %zmm2 ; AVX512-NEXT: vinserti64x4 $1, %ymm2, %zmm1, %zmm1
; AVX512-NEXT: vpmovwb %zmm2, %ymm8 ; AVX512-NEXT: vpmovwb %zmm1, %ymm8
; AVX512-NEXT: vextracti64x4 $1, %zmm1, %ymm14 ; AVX512-NEXT: vextracti64x4 $1, %zmm0, %ymm1
; AVX512-NEXT: vextracti128 $1, %ymm14, %xmm9 ; AVX512-NEXT: vextracti128 $1, %ymm1, %xmm9
; AVX512-NEXT: vmovdqa {{.*#+}} xmm7 = <u,u,u,u,1,5,9,13,u,u,u,u,u,u,u,u> ; AVX512-NEXT: vmovdqa {{.*#+}} xmm6 = <u,u,u,u,1,5,9,13,u,u,u,u,u,u,u,u>
; AVX512-NEXT: vpshufb %xmm7, %xmm9, %xmm4 ; AVX512-NEXT: vpshufb %xmm6, %xmm9, %xmm3
; AVX512-NEXT: vpshufb %xmm7, %xmm14, %xmm5 ; AVX512-NEXT: vpshufb %xmm6, %xmm1, %xmm4
; AVX512-NEXT: vpunpckldq {{.*#+}} xmm4 = xmm5[0],xmm4[0],xmm5[1],xmm4[1] ; AVX512-NEXT: vpunpckldq {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1]
; AVX512-NEXT: vinserti128 $1, %xmm4, %ymm0, %ymm5 ; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm10
; AVX512-NEXT: vextracti128 $1, %ymm1, %xmm10 ; AVX512-NEXT: vmovdqa {{.*#+}} xmm2 = <1,5,9,13,u,u,u,u,u,u,u,u,u,u,u,u>
; AVX512-NEXT: vmovdqa {{.*#+}} xmm3 = <1,5,9,13,u,u,u,u,u,u,u,u,u,u,u,u> ; AVX512-NEXT: vpshufb %xmm2, %xmm10, %xmm5
; AVX512-NEXT: vpshufb %xmm3, %xmm10, %xmm6 ; AVX512-NEXT: vpshufb %xmm2, %xmm0, %xmm3
; AVX512-NEXT: vpshufb %xmm3, %xmm1, %xmm4 ; AVX512-NEXT: vpunpckldq {{.*#+}} xmm3 = xmm3[0],xmm5[0],xmm3[1],xmm5[1]
; AVX512-NEXT: vpunpckldq {{.*#+}} xmm4 = xmm4[0],xmm6[0],xmm4[1],xmm6[1] ; AVX512-NEXT: vpblendd {{.*#+}} xmm11 = xmm3[0,1],xmm4[2,3]
; AVX512-NEXT: vinserti128 $1, %xmm4, %ymm0, %ymm4 ; AVX512-NEXT: vextracti64x4 $1, %zmm7, %ymm5
; AVX512-NEXT: vpblendd {{.*#+}} ymm11 = ymm4[0,1,2,3,4,5],ymm5[6,7]
; AVX512-NEXT: vextracti64x4 $1, %zmm0, %ymm5
; AVX512-NEXT: vextracti128 $1, %ymm5, %xmm12 ; AVX512-NEXT: vextracti128 $1, %ymm5, %xmm12
; AVX512-NEXT: vpshufb %xmm7, %xmm12, %xmm4 ; AVX512-NEXT: vpshufb %xmm6, %xmm12, %xmm3
; AVX512-NEXT: vpshufb %xmm7, %xmm5, %xmm7 ; AVX512-NEXT: vperm2i128 {{.*#+}} ymm5 = ymm5[2,3,0,1]
; AVX512-NEXT: vpunpckldq {{.*#+}} xmm4 = xmm7[0],xmm4[0],xmm7[1],xmm4[1] ; AVX512-NEXT: vextracti128 $1, %ymm5, %xmm13
; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm13 ; AVX512-NEXT: vpshufb %xmm6, %xmm13, %xmm6
; AVX512-NEXT: vpshufb %xmm3, %xmm13, %xmm6 ; AVX512-NEXT: vpunpckldq {{.*#+}} xmm3 = xmm6[0],xmm3[0],xmm6[1],xmm3[1]
; AVX512-NEXT: vpshufb %xmm3, %xmm0, %xmm3 ; AVX512-NEXT: vinserti128 $1, %xmm3, %ymm0, %ymm3
; AVX512-NEXT: vpunpckldq {{.*#+}} xmm3 = xmm3[0],xmm6[0],xmm3[1],xmm6[1] ; AVX512-NEXT: vextracti128 $1, %ymm7, %xmm14
; AVX512-NEXT: vpblendd {{.*#+}} xmm3 = xmm3[0,1],xmm4[2,3] ; AVX512-NEXT: vpshufb %xmm2, %xmm14, %xmm4
; AVX512-NEXT: vpblendd {{.*#+}} ymm3 = ymm3[0,1,2,3],ymm11[4,5,6,7] ; AVX512-NEXT: vperm2i128 {{.*#+}} ymm7 = ymm7[2,3,0,1]
; AVX512-NEXT: vpcmpeqb %ymm3, %ymm8, %ymm8 ; AVX512-NEXT: vextracti128 $1, %ymm7, %xmm7
; AVX512-NEXT: vmovdqa {{.*#+}} xmm3 = <u,u,u,u,2,6,10,14,u,u,u,u,u,u,u,u> ; AVX512-NEXT: vpshufb %xmm2, %xmm7, %xmm2
; AVX512-NEXT: vpshufb %xmm3, %xmm9, %xmm4 ; AVX512-NEXT: vpunpckldq {{.*#+}} xmm2 = xmm2[0],xmm4[0],xmm2[1],xmm4[1]
; AVX512-NEXT: vpshufb %xmm3, %xmm14, %xmm6 ; AVX512-NEXT: vinserti128 $1, %xmm2, %ymm0, %ymm2
; AVX512-NEXT: vpunpckldq {{.*#+}} xmm4 = xmm6[0],xmm4[0],xmm6[1],xmm4[1] ; AVX512-NEXT: vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3,4,5],ymm3[6,7]
; AVX512-NEXT: vinserti128 $1, %xmm4, %ymm0, %ymm4 ; AVX512-NEXT: vpblendd {{.*#+}} ymm2 = ymm11[0,1,2,3],ymm2[4,5,6,7]
; AVX512-NEXT: vmovdqa {{.*#+}} xmm6 = <2,6,10,14,u,u,u,u,u,u,u,u,u,u,u,u> ; AVX512-NEXT: vpcmpeqb %ymm2, %ymm8, %ymm8
; AVX512-NEXT: vpshufb %xmm6, %xmm10, %xmm7 ; AVX512-NEXT: vmovdqa {{.*#+}} xmm2 = <u,u,u,u,2,6,10,14,u,u,u,u,u,u,u,u>
; AVX512-NEXT: vpshufb %xmm6, %xmm1, %xmm2 ; AVX512-NEXT: vpshufb %xmm2, %xmm9, %xmm3
; AVX512-NEXT: vpunpckldq {{.*#+}} xmm2 = xmm2[0],xmm7[0],xmm2[1],xmm7[1] ; AVX512-NEXT: vpshufb %xmm2, %xmm1, %xmm4
; AVX512-NEXT: vpunpckldq {{.*#+}} xmm3 = xmm4[0],xmm3[0],xmm4[1],xmm3[1]
; AVX512-NEXT: vmovdqa {{.*#+}} xmm4 = <2,6,10,14,u,u,u,u,u,u,u,u,u,u,u,u>
; AVX512-NEXT: vpshufb %xmm4, %xmm10, %xmm5
; AVX512-NEXT: vpshufb %xmm4, %xmm0, %xmm6
; AVX512-NEXT: vpunpckldq {{.*#+}} xmm5 = xmm6[0],xmm5[0],xmm6[1],xmm5[1]
; AVX512-NEXT: vpblendd {{.*#+}} xmm3 = xmm5[0,1],xmm3[2,3]
; AVX512-NEXT: vpshufb %xmm2, %xmm12, %xmm5
; AVX512-NEXT: vpshufb %xmm2, %xmm13, %xmm2
; AVX512-NEXT: vpunpckldq {{.*#+}} xmm2 = xmm2[0],xmm5[0],xmm2[1],xmm5[1]
; AVX512-NEXT: vinserti128 $1, %xmm2, %ymm0, %ymm2 ; AVX512-NEXT: vinserti128 $1, %xmm2, %ymm0, %ymm2
; AVX512-NEXT: vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3,4,5],ymm4[6,7] ; AVX512-NEXT: vpshufb %xmm4, %xmm14, %xmm5
; AVX512-NEXT: vpshufb %xmm3, %xmm12, %xmm4 ; AVX512-NEXT: vpshufb %xmm4, %xmm7, %xmm4
; AVX512-NEXT: vpshufb %xmm3, %xmm5, %xmm3 ; AVX512-NEXT: vpunpckldq {{.*#+}} xmm4 = xmm4[0],xmm5[0],xmm4[1],xmm5[1]
; AVX512-NEXT: vpunpckldq {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1] ; AVX512-NEXT: vinserti128 $1, %xmm4, %ymm0, %ymm4
; AVX512-NEXT: vpshufb %xmm6, %xmm13, %xmm4 ; AVX512-NEXT: vpblendd {{.*#+}} ymm2 = ymm4[0,1,2,3,4,5],ymm2[6,7]
; AVX512-NEXT: vpshufb %xmm6, %xmm0, %xmm6
; AVX512-NEXT: vpunpckldq {{.*#+}} xmm4 = xmm6[0],xmm4[0],xmm6[1],xmm4[1]
; AVX512-NEXT: vpblendd {{.*#+}} xmm3 = xmm4[0,1],xmm3[2,3]
; AVX512-NEXT: vpblendd {{.*#+}} ymm2 = ymm3[0,1,2,3],ymm2[4,5,6,7] ; AVX512-NEXT: vpblendd {{.*#+}} ymm2 = ymm3[0,1,2,3],ymm2[4,5,6,7]
; AVX512-NEXT: vmovdqa {{.*#+}} xmm3 = <u,u,u,u,3,7,11,15,u,u,u,u,u,u,u,u> ; AVX512-NEXT: vmovdqa {{.*#+}} xmm3 = <u,u,u,u,3,7,11,15,u,u,u,u,u,u,u,u>
; AVX512-NEXT: vpshufb %xmm3, %xmm9, %xmm4 ; AVX512-NEXT: vpshufb %xmm3, %xmm9, %xmm4
; AVX512-NEXT: vpshufb %xmm3, %xmm14, %xmm6 ; AVX512-NEXT: vpshufb %xmm3, %xmm1, %xmm1
; AVX512-NEXT: vpunpckldq {{.*#+}} xmm4 = xmm6[0],xmm4[0],xmm6[1],xmm4[1] ; AVX512-NEXT: vpunpckldq {{.*#+}} xmm1 = xmm1[0],xmm4[0],xmm1[1],xmm4[1]
; AVX512-NEXT: vinserti128 $1, %xmm4, %ymm0, %ymm4 ; AVX512-NEXT: vmovdqa {{.*#+}} xmm4 = <3,7,11,15,u,u,u,u,u,u,u,u,u,u,u,u>
; AVX512-NEXT: vmovdqa {{.*#+}} xmm6 = <3,7,11,15,u,u,u,u,u,u,u,u,u,u,u,u> ; AVX512-NEXT: vpshufb %xmm4, %xmm10, %xmm5
; AVX512-NEXT: vpshufb %xmm6, %xmm10, %xmm7 ; AVX512-NEXT: vpshufb %xmm4, %xmm0, %xmm0
; AVX512-NEXT: vpshufb %xmm6, %xmm1, %xmm1 ; AVX512-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm5[0],xmm0[1],xmm5[1]
; AVX512-NEXT: vpunpckldq {{.*#+}} xmm1 = xmm1[0],xmm7[0],xmm1[1],xmm7[1] ; AVX512-NEXT: vpblendd {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3]
; AVX512-NEXT: vpshufb %xmm3, %xmm12, %xmm1
; AVX512-NEXT: vpshufb %xmm3, %xmm13, %xmm3
; AVX512-NEXT: vpunpckldq {{.*#+}} xmm1 = xmm3[0],xmm1[0],xmm3[1],xmm1[1]
; AVX512-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm1 ; AVX512-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm1
; AVX512-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3,4,5],ymm4[6,7] ; AVX512-NEXT: vpshufb %xmm4, %xmm14, %xmm3
; AVX512-NEXT: vpshufb %xmm3, %xmm12, %xmm4 ; AVX512-NEXT: vpshufb %xmm4, %xmm7, %xmm4
; AVX512-NEXT: vpshufb %xmm3, %xmm5, %xmm3 ; AVX512-NEXT: vpunpckldq {{.*#+}} xmm3 = xmm4[0],xmm3[0],xmm4[1],xmm3[1]
; AVX512-NEXT: vpunpckldq {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1] ; AVX512-NEXT: vinserti128 $1, %xmm3, %ymm0, %ymm3
; AVX512-NEXT: vpshufb %xmm6, %xmm13, %xmm4 ; AVX512-NEXT: vpblendd {{.*#+}} ymm1 = ymm3[0,1,2,3,4,5],ymm1[6,7]
; AVX512-NEXT: vpshufb %xmm6, %xmm0, %xmm0
; AVX512-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1]
; AVX512-NEXT: vpblendd {{.*#+}} xmm0 = xmm0[0,1],xmm3[2,3]
; AVX512-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7] ; AVX512-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
; AVX512-NEXT: vpcmpeqb %ymm0, %ymm2, %ymm0 ; AVX512-NEXT: vpcmpeqb %ymm0, %ymm2, %ymm0
; AVX512-NEXT: vpsllw $7, %ymm8, %ymm1 ; AVX512-NEXT: vpsllw $7, %ymm8, %ymm1
; AVX512-NEXT: vpmovb2m %zmm1, %k0 ; AVX512-NEXT: vpmovb2m %zmm1, %k0
; AVX512-NEXT: vpsllw $7, %ymm0, %ymm0 ; AVX512-NEXT: vpsllw $7, %ymm0, %ymm0
; AVX512-NEXT: vpmovb2m %zmm0, %k1 ; AVX512-NEXT: vpmovb2m %zmm0, %k1
; AVX512-NEXT: kxnord %k1, %k0, %k0 ; AVX512-NEXT: kxnord %k1, %k0, %k0
; AVX512-NEXT: vpmovm2b %k0, %zmm0 ; AVX512-NEXT: vpmovm2b %k0, %zmm0
▲ Show 20 Lines • Show All 129 Lines • ▼ Show 20 Lines
; AVX1-NEXT: vpaddb %xmm1, %xmm3, %xmm1 ; AVX1-NEXT: vpaddb %xmm1, %xmm3, %xmm1
; AVX1-NEXT: vpaddb %xmm0, %xmm2, %xmm0 ; AVX1-NEXT: vpaddb %xmm0, %xmm2, %xmm0
; AVX1-NEXT: vpaddb %xmm0, %xmm8, %xmm0 ; AVX1-NEXT: vpaddb %xmm0, %xmm8, %xmm0
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq ; AVX1-NEXT: retq
; ;
; AVX2-LABEL: interleaved_load_vf32_i8_stride3: ; AVX2-LABEL: interleaved_load_vf32_i8_stride3:
; AVX2: # BB#0: ; AVX2: # BB#0:
; AVX2-NEXT: vmovdqa (%rdi), %ymm1 ; AVX2-NEXT: vmovdqa (%rdi), %ymm0
; AVX2-NEXT: vmovdqa 32(%rdi), %ymm2 ; AVX2-NEXT: vmovdqa 32(%rdi), %ymm1
; AVX2-NEXT: vmovdqa 64(%rdi), %ymm0 ; AVX2-NEXT: vmovdqa 64(%rdi), %ymm2
; AVX2-NEXT: vmovdqa {{.*#+}} ymm3 = <255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0> ; AVX2-NEXT: vmovdqa {{.*#+}} ymm3 = <255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0>
; AVX2-NEXT: vpblendvb %ymm3, %ymm1, %ymm2, %ymm3 ; AVX2-NEXT: vpblendvb %ymm3, %ymm0, %ymm1, %ymm3
; AVX2-NEXT: vextracti128 $1, %ymm3, %xmm4 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm4 = ymm3[2,3,0,1]
; AVX2-NEXT: vpshufb {{.*#+}} xmm5 = zero,zero,zero,zero,zero,zero,xmm4[2,5,8,11,14],zero,zero,zero,zero,zero ; AVX2-NEXT: vmovdqa {{.*#+}} ymm5 = <255,255,0,255,255,0,255,255,0,255,255,0,255,255,0,255,255,u,u,255,u,u,255,u,u,255,u,u,255,u,u,255>
; AVX2-NEXT: vpshufb {{.*#+}} xmm3 = xmm3[0,3,6,9,12,15],zero,zero,zero,zero,zero,xmm3[1,4,7,10,13] ; AVX2-NEXT: vpblendvb %ymm5, %ymm3, %ymm4, %ymm3
; AVX2-NEXT: vpor %xmm5, %xmm3, %xmm3 ; AVX2-NEXT: vpshufb {{.*#+}} ymm3 = ymm3[0,3,6,9,12,15,2,5,8,11,14,1,4,7,10,13,16,19,22,25,28,31,u,u,u,u,u,u,u,u,u,u]
; AVX2-NEXT: vpshufb {{.*#+}} xmm4 = xmm4[0,3,6,9,12,15,u,u,u,u,u,u,u,u,u,u] ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm4 = ymm2[2,3,0,1]
; AVX2-NEXT: vinserti128 $1, %xmm4, %ymm3, %ymm4 ; AVX2-NEXT: vextracti128 $1, %ymm4, %xmm4
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm3 ; AVX2-NEXT: vpshufb {{.*#+}} xmm5 = xmm4[u,u,u,u,u,u,2,5,8,11,14],zero,zero,zero,zero,zero
; AVX2-NEXT: vpshufb {{.*#+}} xmm5 = xmm3[u,u,u,u,u,u],zero,zero,zero,zero,zero,xmm3[1,4,7,10,13] ; AVX2-NEXT: vextracti128 $1, %ymm2, %xmm2
; AVX2-NEXT: vpshufb {{.*#+}} xmm6 = xmm0[u,u,u,u,u,u,2,5,8,11,14],zero,zero,zero,zero,zero ; AVX2-NEXT: vpshufb {{.*#+}} xmm6 = xmm2[u,u,u,u,u,u],zero,zero,zero,zero,zero,xmm2[1,4,7,10,13]
; AVX2-NEXT: vpor %xmm5, %xmm6, %xmm5 ; AVX2-NEXT: vpor %xmm6, %xmm5, %xmm5
; AVX2-NEXT: vinserti128 $1, %xmm5, %ymm0, %ymm5 ; AVX2-NEXT: vinserti128 $1, %xmm5, %ymm0, %ymm5
; AVX2-NEXT: vmovdqa {{.*#+}} ymm6 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm6 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0]
; AVX2-NEXT: vpblendvb %ymm6, %ymm4, %ymm5, %ymm4 ; AVX2-NEXT: vpblendvb %ymm6, %ymm3, %ymm5, %ymm3
; AVX2-NEXT: vmovdqa {{.*#+}} ymm5 = <u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255> ; AVX2-NEXT: vmovdqa {{.*#+}} ymm5 = <u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255>
; AVX2-NEXT: vpblendvb %ymm5, %ymm1, %ymm2, %ymm5 ; AVX2-NEXT: vpblendvb %ymm5, %ymm0, %ymm1, %ymm5
; AVX2-NEXT: vextracti128 $1, %ymm5, %xmm6 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm6 = ymm5[2,3,0,1]
; AVX2-NEXT: vpshufb {{.*#+}} xmm7 = zero,zero,zero,zero,zero,xmm6[0,3,6,9,12,15],zero,zero,zero,zero,zero ; AVX2-NEXT: vmovdqa {{.*#+}} ymm7 = <0,255,255,0,255,255,0,255,255,0,255,255,0,255,255,0,u,255,u,u,255,u,u,255,u,u,255,u,u,255,u,u>
; AVX2-NEXT: vpshufb {{.*#+}} xmm5 = xmm5[1,4,7,10,13],zero,zero,zero,zero,zero,zero,xmm5[2,5,8,11,14] ; AVX2-NEXT: vpblendvb %ymm7, %ymm5, %ymm6, %ymm5
; AVX2-NEXT: vpor %xmm7, %xmm5, %xmm5 ; AVX2-NEXT: vpshufb {{.*#+}} ymm5 = ymm5[1,4,7,10,13,0,3,6,9,12,15,2,5,8,11,14,17,20,23,26,29,u,u,u,u,u,u,u,u,u,u,u]
; AVX2-NEXT: vpshufb {{.*#+}} xmm6 = xmm6[1,4,7,10,13,u,u,u,u,u,u,u,u,u,u,u] ; AVX2-NEXT: vpshufb {{.*#+}} xmm6 = xmm4[u,u,u,u,u,0,3,6,9,12,15],zero,zero,zero,zero,zero
; AVX2-NEXT: vinserti128 $1, %xmm6, %ymm5, %ymm5 ; AVX2-NEXT: vpshufb {{.*#+}} xmm7 = xmm2[u,u,u,u,u],zero,zero,zero,zero,zero,zero,xmm2[2,5,8,11,14]
; AVX2-NEXT: vpshufb {{.*#+}} xmm6 = xmm3[u,u,u,u,u],zero,zero,zero,zero,zero,zero,xmm3[2,5,8,11,14] ; AVX2-NEXT: vpor %xmm7, %xmm6, %xmm6
; AVX2-NEXT: vpshufb {{.*#+}} xmm7 = xmm0[u,u,u,u,u,0,3,6,9,12,15],zero,zero,zero,zero,zero
; AVX2-NEXT: vpor %xmm6, %xmm7, %xmm6
; AVX2-NEXT: vinserti128 $1, %xmm6, %ymm0, %ymm6 ; AVX2-NEXT: vinserti128 $1, %xmm6, %ymm0, %ymm6
; AVX2-NEXT: vmovdqa {{.*#+}} ymm7 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0,0] ; AVX2-NEXT: vmovdqa {{.*#+}} ymm7 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0,0]
; AVX2-NEXT: vpblendvb %ymm7, %ymm5, %ymm6, %ymm5 ; AVX2-NEXT: vpblendvb %ymm7, %ymm5, %ymm6, %ymm5
; AVX2-NEXT: vmovdqa {{.*#+}} ymm6 = <255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u> ; AVX2-NEXT: vmovdqa {{.*#+}} ymm6 = <255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u>
; AVX2-NEXT: vpblendvb %ymm6, %ymm2, %ymm1, %ymm1 ; AVX2-NEXT: vpblendvb %ymm6, %ymm1, %ymm0, %ymm0
; AVX2-NEXT: vextracti128 $1, %ymm1, %xmm2 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX2-NEXT: vpshufb {{.*#+}} xmm6 = zero,zero,zero,zero,zero,xmm2[1,4,7,10,13],zero,zero,zero,zero,zero,zero ; AVX2-NEXT: vmovdqa {{.*#+}} ymm6 = <255,0,255,255,0,255,255,0,255,255,0,255,255,0,255,255,u,u,255,u,u,255,u,u,255,u,u,255,u,u,255,u>
; AVX2-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[2,5,8,11,14],zero,zero,zero,zero,zero,xmm1[0,3,6,9,12,15] ; AVX2-NEXT: vpblendvb %ymm6, %ymm0, %ymm1, %ymm0
; AVX2-NEXT: vpor %xmm6, %xmm1, %xmm1 ; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[2,5,8,11,14,1,4,7,10,13,0,3,6,9,12,15,18,21,24,27,30,u,u,u,u,u,u,u,u,u,u,u]
; AVX2-NEXT: vpshufb {{.*#+}} xmm2 = xmm2[2,5,8,11,14,u,u,u,u,u,u,u,u,u,u,u] ; AVX2-NEXT: vpshufb {{.*#+}} xmm1 = xmm4[u,u,u,u,u,1,4,7,10,13],zero,zero,zero,zero,zero,zero
; AVX2-NEXT: vinserti128 $1, %xmm2, %ymm1, %ymm1 ; AVX2-NEXT: vpshufb {{.*#+}} xmm2 = xmm2[u,u,u,u,u],zero,zero,zero,zero,zero,xmm2[0,3,6,9,12,15]
; AVX2-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[u,u,u,u,u,1,4,7,10,13],zero,zero,zero,zero,zero,zero ; AVX2-NEXT: vpor %xmm1, %xmm2, %xmm1
; AVX2-NEXT: vpshufb {{.*#+}} xmm2 = xmm3[u,u,u,u,u],zero,zero,zero,zero,zero,xmm3[0,3,6,9,12,15] ; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm1
; AVX2-NEXT: vpor %xmm0, %xmm2, %xmm0 ; AVX2-NEXT: vpblendvb %ymm7, %ymm0, %ymm1, %ymm0
; AVX2-NEXT: vinserti128 $1, %xmm0, %ymm0, %ymm0
; AVX2-NEXT: vpblendvb %ymm7, %ymm1, %ymm0, %ymm0
; AVX2-NEXT: vpaddb %ymm0, %ymm5, %ymm0 ; AVX2-NEXT: vpaddb %ymm0, %ymm5, %ymm0
; AVX2-NEXT: vpaddb %ymm0, %ymm4, %ymm0 ; AVX2-NEXT: vpaddb %ymm0, %ymm3, %ymm0
; AVX2-NEXT: retq ; AVX2-NEXT: retq
; ;
; AVX512-LABEL: interleaved_load_vf32_i8_stride3: ; AVX512-LABEL: interleaved_load_vf32_i8_stride3:
; AVX512: # BB#0: ; AVX512: # BB#0:
; AVX512-NEXT: vmovdqa64 (%rdi), %zmm1 ; AVX512-NEXT: vmovdqa64 (%rdi), %zmm0
; AVX512-NEXT: vmovdqa64 64(%rdi), %zmm0 ; AVX512-NEXT: vmovdqa64 64(%rdi), %zmm1
; AVX512-NEXT: vmovdqa {{.*#+}} ymm3 = <255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0> ; AVX512-NEXT: vmovdqa {{.*#+}} ymm2 = <255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0>
; AVX512-NEXT: vextracti64x4 $1, %zmm1, %ymm2 ; AVX512-NEXT: vextracti64x4 $1, %zmm0, %ymm3
; AVX512-NEXT: vpblendvb %ymm3, %ymm1, %ymm2, %ymm3 ; AVX512-NEXT: vpblendvb %ymm2, %ymm0, %ymm3, %ymm2
; AVX512-NEXT: vextracti128 $1, %ymm3, %xmm4 ; AVX512-NEXT: vperm2i128 {{.*#+}} ymm4 = ymm2[2,3,0,1]
; AVX512-NEXT: vpshufb {{.*#+}} xmm5 = zero,zero,zero,zero,zero,zero,xmm4[2,5,8,11,14],zero,zero,zero,zero,zero ; AVX512-NEXT: vmovdqa {{.*#+}} ymm5 = <255,255,0,255,255,0,255,255,0,255,255,0,255,255,0,255,255,u,u,255,u,u,255,u,u,255,u,u,255,u,u,255>
; AVX512-NEXT: vpshufb {{.*#+}} xmm3 = xmm3[0,3,6,9,12,15],zero,zero,zero,zero,zero,xmm3[1,4,7,10,13] ; AVX512-NEXT: vpblendvb %ymm5, %ymm2, %ymm4, %ymm2
; AVX512-NEXT: vpor %xmm5, %xmm3, %xmm3 ; AVX512-NEXT: vpshufb {{.*#+}} ymm2 = ymm2[0,3,6,9,12,15,2,5,8,11,14,1,4,7,10,13,16,19,22,25,28,31,u,u,u,u,u,u,u,u,u,u]
; AVX512-NEXT: vpshufb {{.*#+}} xmm4 = xmm4[0,3,6,9,12,15,u,u,u,u,u,u,u,u,u,u] ; AVX512-NEXT: vperm2i128 {{.*#+}} ymm4 = ymm1[2,3,0,1]
; AVX512-NEXT: vinserti128 $1, %xmm4, %ymm3, %ymm4 ; AVX512-NEXT: vextracti128 $1, %ymm4, %xmm4
; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm3 ; AVX512-NEXT: vpshufb {{.*#+}} xmm5 = xmm4[u,u,u,u,u,u,2,5,8,11,14],zero,zero,zero,zero,zero
; AVX512-NEXT: vpshufb {{.*#+}} xmm5 = xmm3[u,u,u,u,u,u],zero,zero,zero,zero,zero,xmm3[1,4,7,10,13] ; AVX512-NEXT: vextracti128 $1, %ymm1, %xmm1
; AVX512-NEXT: vpshufb {{.*#+}} xmm6 = xmm0[u,u,u,u,u,u,2,5,8,11,14],zero,zero,zero,zero,zero ; AVX512-NEXT: vpshufb {{.*#+}} xmm6 = xmm1[u,u,u,u,u,u],zero,zero,zero,zero,zero,xmm1[1,4,7,10,13]
; AVX512-NEXT: vpor %xmm5, %xmm6, %xmm5 ; AVX512-NEXT: vpor %xmm6, %xmm5, %xmm5
; AVX512-NEXT: vinserti128 $1, %xmm5, %ymm0, %ymm5 ; AVX512-NEXT: vinserti128 $1, %xmm5, %ymm0, %ymm5
; AVX512-NEXT: vmovdqa {{.*#+}} ymm6 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0] ; AVX512-NEXT: vmovdqa {{.*#+}} ymm6 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0]
; AVX512-NEXT: vpblendvb %ymm6, %ymm4, %ymm5, %ymm4 ; AVX512-NEXT: vpblendvb %ymm6, %ymm2, %ymm5, %ymm2
; AVX512-NEXT: vmovdqa {{.*#+}} ymm5 = <u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255> ; AVX512-NEXT: vmovdqa {{.*#+}} ymm5 = <u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255>
; AVX512-NEXT: vpblendvb %ymm5, %ymm1, %ymm2, %ymm5 ; AVX512-NEXT: vpblendvb %ymm5, %ymm0, %ymm3, %ymm5
; AVX512-NEXT: vextracti128 $1, %ymm5, %xmm6 ; AVX512-NEXT: vperm2i128 {{.*#+}} ymm6 = ymm5[2,3,0,1]
; AVX512-NEXT: vpshufb {{.*#+}} xmm7 = zero,zero,zero,zero,zero,xmm6[0,3,6,9,12,15],zero,zero,zero,zero,zero ; AVX512-NEXT: vmovdqa {{.*#+}} ymm7 = <0,255,255,0,255,255,0,255,255,0,255,255,0,255,255,0,u,255,u,u,255,u,u,255,u,u,255,u,u,255,u,u>
; AVX512-NEXT: vpshufb {{.*#+}} xmm5 = xmm5[1,4,7,10,13],zero,zero,zero,zero,zero,zero,xmm5[2,5,8,11,14] ; AVX512-NEXT: vpblendvb %ymm7, %ymm5, %ymm6, %ymm5
; AVX512-NEXT: vpor %xmm7, %xmm5, %xmm5 ; AVX512-NEXT: vpshufb {{.*#+}} ymm5 = ymm5[1,4,7,10,13,0,3,6,9,12,15,2,5,8,11,14,17,20,23,26,29,u,u,u,u,u,u,u,u,u,u,u]
; AVX512-NEXT: vpshufb {{.*#+}} xmm6 = xmm6[1,4,7,10,13,u,u,u,u,u,u,u,u,u,u,u] ; AVX512-NEXT: vpshufb {{.*#+}} xmm6 = xmm4[u,u,u,u,u,0,3,6,9,12,15],zero,zero,zero,zero,zero
; AVX512-NEXT: vinserti128 $1, %xmm6, %ymm5, %ymm5 ; AVX512-NEXT: vpshufb {{.*#+}} xmm7 = xmm1[u,u,u,u,u],zero,zero,zero,zero,zero,zero,xmm1[2,5,8,11,14]
; AVX512-NEXT: vpshufb {{.*#+}} xmm6 = xmm3[u,u,u,u,u],zero,zero,zero,zero,zero,zero,xmm3[2,5,8,11,14] ; AVX512-NEXT: vpor %xmm7, %xmm6, %xmm6
; AVX512-NEXT: vpshufb {{.*#+}} xmm7 = xmm0[u,u,u,u,u,0,3,6,9,12,15],zero,zero,zero,zero,zero
; AVX512-NEXT: vpor %xmm6, %xmm7, %xmm6
; AVX512-NEXT: vinserti128 $1, %xmm6, %ymm0, %ymm6 ; AVX512-NEXT: vinserti128 $1, %xmm6, %ymm0, %ymm6
; AVX512-NEXT: vmovdqa {{.*#+}} ymm7 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0,0] ; AVX512-NEXT: vmovdqa {{.*#+}} ymm7 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0,0]
; AVX512-NEXT: vpblendvb %ymm7, %ymm5, %ymm6, %ymm5 ; AVX512-NEXT: vpblendvb %ymm7, %ymm5, %ymm6, %ymm5
; AVX512-NEXT: vmovdqa {{.*#+}} ymm6 = <255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u> ; AVX512-NEXT: vmovdqa {{.*#+}} ymm6 = <255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u>
; AVX512-NEXT: vpblendvb %ymm6, %ymm2, %ymm1, %ymm1 ; AVX512-NEXT: vpblendvb %ymm6, %ymm3, %ymm0, %ymm0
; AVX512-NEXT: vextracti128 $1, %ymm1, %xmm2 ; AVX512-NEXT: vperm2i128 {{.*#+}} ymm3 = ymm0[2,3,0,1]
; AVX512-NEXT: vpshufb {{.*#+}} xmm6 = zero,zero,zero,zero,zero,xmm2[1,4,7,10,13],zero,zero,zero,zero,zero,zero ; AVX512-NEXT: vmovdqa {{.*#+}} ymm6 = <255,0,255,255,0,255,255,0,255,255,0,255,255,0,255,255,u,u,255,u,u,255,u,u,255,u,u,255,u,u,255,u>
; AVX512-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[2,5,8,11,14],zero,zero,zero,zero,zero,xmm1[0,3,6,9,12,15] ; AVX512-NEXT: vpblendvb %ymm6, %ymm0, %ymm3, %ymm0
; AVX512-NEXT: vpor %xmm6, %xmm1, %xmm1 ; AVX512-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[2,5,8,11,14,1,4,7,10,13,0,3,6,9,12,15,18,21,24,27,30,u,u,u,u,u,u,u,u,u,u,u]
; AVX512-NEXT: vpshufb {{.*#+}} xmm2 = xmm2[2,5,8,11,14,u,u,u,u,u,u,u,u,u,u,u] ; AVX512-NEXT: vpshufb {{.*#+}} xmm3 = xmm4[u,u,u,u,u,1,4,7,10,13],zero,zero,zero,zero,zero,zero
; AVX512-NEXT: vinserti128 $1, %xmm2, %ymm1, %ymm1 ; AVX512-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[u,u,u,u,u],zero,zero,zero,zero,zero,xmm1[0,3,6,9,12,15]
; AVX512-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[u,u,u,u,u,1,4,7,10,13],zero,zero,zero,zero,zero,zero ; AVX512-NEXT: vpor %xmm3, %xmm1, %xmm1
; AVX512-NEXT: vpshufb {{.*#+}} xmm2 = xmm3[u,u,u,u,u],zero,zero,zero,zero,zero,xmm3[0,3,6,9,12,15] ; AVX512-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm1
; AVX512-NEXT: vpor %xmm0, %xmm2, %xmm0 ; AVX512-NEXT: vpblendvb %ymm7, %ymm0, %ymm1, %ymm0
; AVX512-NEXT: vinserti128 $1, %xmm0, %ymm0, %ymm0
; AVX512-NEXT: vpblendvb %ymm7, %ymm1, %ymm0, %ymm0
; AVX512-NEXT: vpaddb %ymm0, %ymm5, %ymm0 ; AVX512-NEXT: vpaddb %ymm0, %ymm5, %ymm0
; AVX512-NEXT: vpaddb %ymm0, %ymm4, %ymm0 ; AVX512-NEXT: vpaddb %ymm0, %ymm2, %ymm0
; AVX512-NEXT: retq ; AVX512-NEXT: retq
%wide.vec = load <96 x i8>, <96 x i8>* %ptr %wide.vec = load <96 x i8>, <96 x i8>* %ptr
%v1 = shufflevector <96 x i8> %wide.vec, <96 x i8> undef,<32 x i32> <i32 0,i32 3,i32 6,i32 9,i32 12,i32 15,i32 18,i32 21,i32 24,i32 27,i32 30,i32 33,i32 36,i32 39,i32 42,i32 45,i32 48,i32 51,i32 54,i32 57,i32 60,i32 63,i32 66,i32 69,i32 72,i32 75,i32 78,i32 81,i32 84,i32 87,i32 90,i32 93> %v1 = shufflevector <96 x i8> %wide.vec, <96 x i8> undef,<32 x i32> <i32 0,i32 3,i32 6,i32 9,i32 12,i32 15,i32 18,i32 21,i32 24,i32 27,i32 30,i32 33,i32 36,i32 39,i32 42,i32 45,i32 48,i32 51,i32 54,i32 57,i32 60,i32 63,i32 66,i32 69,i32 72,i32 75,i32 78,i32 81,i32 84,i32 87,i32 90,i32 93>
%v2 = shufflevector <96 x i8> %wide.vec, <96 x i8> undef,<32 x i32> <i32 1,i32 4,i32 7,i32 10,i32 13,i32 16,i32 19,i32 22,i32 25,i32 28,i32 31,i32 34,i32 37,i32 40,i32 43,i32 46,i32 49,i32 52,i32 55,i32 58,i32 61,i32 64,i32 67,i32 70,i32 73,i32 76,i32 79,i32 82,i32 85,i32 88,i32 91,i32 94> %v2 = shufflevector <96 x i8> %wide.vec, <96 x i8> undef,<32 x i32> <i32 1,i32 4,i32 7,i32 10,i32 13,i32 16,i32 19,i32 22,i32 25,i32 28,i32 31,i32 34,i32 37,i32 40,i32 43,i32 46,i32 49,i32 52,i32 55,i32 58,i32 61,i32 64,i32 67,i32 70,i32 73,i32 76,i32 79,i32 82,i32 85,i32 88,i32 91,i32 94>
%v3 = shufflevector <96 x i8> %wide.vec, <96 x i8> undef,<32 x i32> <i32 2,i32 5,i32 8,i32 11,i32 14,i32 17,i32 20,i32 23,i32 26,i32 29,i32 32,i32 35,i32 38,i32 41,i32 44,i32 47,i32 50,i32 53,i32 56,i32 59,i32 62,i32 65,i32 68,i32 71,i32 74,i32 77,i32 80,i32 83,i32 86,i32 89,i32 92,i32 95> %v3 = shufflevector <96 x i8> %wide.vec, <96 x i8> undef,<32 x i32> <i32 2,i32 5,i32 8,i32 11,i32 14,i32 17,i32 20,i32 23,i32 26,i32 29,i32 32,i32 35,i32 38,i32 41,i32 44,i32 47,i32 50,i32 53,i32 56,i32 59,i32 62,i32 65,i32 68,i32 71,i32 74,i32 77,i32 80,i32 83,i32 86,i32 89,i32 92,i32 95>
%add1 = add <32 x i8> %v1, %v2 %add1 = add <32 x i8> %v1, %v2
%add2 = add <32 x i8> %v3, %add1 %add2 = add <32 x i8> %v3, %add1
ret <32 x i8> %add2 ret <32 x i8> %add2
▲ Show 20 Lines • Show All 400 Lines • Show Last 20 Lines