This is an archive of the discontinued LLVM Phabricator instance.

Differential D88206

[AMDGPU] Fix v3f16 handling for getresinfo
ClosedPublic

Authored by Flakebi on Sep 24 2020, 2:41 AM.

Details

Reviewers
arsenm
Commits
rG6f7cd16d2971: [AMDGPU] Fix v3f16 handling for getresinfo
Summary

v3f32 should not be expanded to v4f32. getresinfo with a dmask of 7
created an image sample with a v3f32 return value, which was bitcasted
to a v4f32 in constructRetValue.

Diff Detail

Event Timeline

Flakebi created this revision.Sep 24 2020, 2:41 AM
Herald added a project: Restricted Project. · View Herald TranscriptSep 24 2020, 2:41 AM
Herald added subscribers: llvm-commits, kerbowa, hiraditya and 7 others. · View Herald Transcript
Flakebi requested review of this revision.Sep 24 2020, 2:41 AM
Herald added a subscriber: wdng. · View Herald TranscriptSep 24 2020, 2:41 AM
Harbormaster completed remote builds in B72787: Diff 293985.Sep 24 2020, 3:10 AM
arsenm accepted this revision.Sep 24 2020, 6:37 AM
This revision is now accepted and ready to land.Sep 24 2020, 6:37 AM
Closed by commit rG6f7cd16d2971: [AMDGPU] Fix v3f16 handling for getresinfo (authored by sebastian-ne). · Explain WhySep 24 2020, 7:15 AM
This revision was automatically updated to reflect the committed changes.
sebastian-ne added a commit: rG6f7cd16d2971: [AMDGPU] Fix v3f16 handling for getresinfo.

Revision Contents

PathSize
llvm/
lib/
Target/
AMDGPU/
3 lines
test/
CodeGen/
AMDGPU/
GlobalISel/
147 lines
195 lines

Diff 294048

llvm/lib/Target/AMDGPU/SIISelLowering.cpp

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 5,871 Lines • ▼ Show 20 Linesstatic SDValue constructRetValue(SelectionDAG &DAG,
if (IsD16) if (IsD16)
Data = adjustLoadValueTypeImpl(Data, ReqRetVT, DL, DAG, Unpacked); Data = adjustLoadValueTypeImpl(Data, ReqRetVT, DL, DAG, Unpacked);
EVT LegalReqRetVT = ReqRetVT; EVT LegalReqRetVT = ReqRetVT;
if (!ReqRetVT.isVector()) { if (!ReqRetVT.isVector()) {
Data = DAG.getNode(ISD::TRUNCATE, DL, ReqRetVT.changeTypeToInteger(), Data); Data = DAG.getNode(ISD::TRUNCATE, DL, ReqRetVT.changeTypeToInteger(), Data);
} else { } else {
// We need to widen the return vector to a legal type // We need to widen the return vector to a legal type
if ((ReqRetVT.getVectorNumElements() % 2) == 1) { if ((ReqRetVT.getVectorNumElements() % 2) == 1 &&
ReqRetVT.getVectorElementType().getSizeInBits() == 16) {
LegalReqRetVT = LegalReqRetVT =
EVT::getVectorVT(*DAG.getContext(), ReqRetVT.getVectorElementType(), EVT::getVectorVT(*DAG.getContext(), ReqRetVT.getVectorElementType(),
ReqRetVT.getVectorNumElements() + 1); ReqRetVT.getVectorNumElements() + 1);
} }
} }
Data = DAG.getNode(ISD::BITCAST, DL, LegalReqRetVT, Data); Data = DAG.getNode(ISD::BITCAST, DL, LegalReqRetVT, Data);
if (TexFail) if (TexFail)
▲ Show 20 Lines • Show All 6,024 Lines • Show Last 20 Lines

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.image.getresinfo.ll

Show First 20 Lines • Show All 381 Lines • ▼ Show 20 Lines
; GFX10-NEXT: image_get_resinfo v[0:3], v0, s[0:7] dmask:0xf dim:SQ_RSRC_IMG_2D_MSAA_ARRAY unorm ; GFX10-NEXT: image_get_resinfo v[0:3], v0, s[0:7] dmask:0xf dim:SQ_RSRC_IMG_2D_MSAA_ARRAY unorm
; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_waitcnt vmcnt(0)
; GFX10-NEXT: ; return to shader part epilog ; GFX10-NEXT: ; return to shader part epilog
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.getresinfo.2darraymsaa.v4f32.i32(i32 15, i32 %mip, <8 x i32> %rsrc, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.getresinfo.2darraymsaa.v4f32.i32(i32 15, i32 %mip, <8 x i32> %rsrc, i32 0, i32 0)
ret <4 x float> %v ret <4 x float> %v
} }
define amdgpu_ps <3 x float> @getresinfo_dmask7(<8 x i32> inreg %rsrc, <4 x float> %vdata, i32 %mip) {
; GFX6-LABEL: getresinfo_dmask7:
; GFX6: ; %bb.0: ; %main_body
; GFX6-NEXT: s_mov_b32 s0, s2
; GFX6-NEXT: s_mov_b32 s1, s3
; GFX6-NEXT: s_mov_b32 s2, s4
; GFX6-NEXT: s_mov_b32 s3, s5
; GFX6-NEXT: s_mov_b32 s4, s6
; GFX6-NEXT: s_mov_b32 s5, s7
; GFX6-NEXT: s_mov_b32 s6, s8
; GFX6-NEXT: s_mov_b32 s7, s9
; GFX6-NEXT: image_get_resinfo v[0:2], v0, s[0:7] dmask:0x7 unorm
; GFX6-NEXT: s_waitcnt vmcnt(0)
; GFX6-NEXT: ; return to shader part epilog
;
; GFX8-LABEL: getresinfo_dmask7:
; GFX8: ; %bb.0: ; %main_body
; GFX8-NEXT: s_mov_b32 s0, s2
; GFX8-NEXT: s_mov_b32 s1, s3
; GFX8-NEXT: s_mov_b32 s2, s4
; GFX8-NEXT: s_mov_b32 s3, s5
; GFX8-NEXT: s_mov_b32 s4, s6
; GFX8-NEXT: s_mov_b32 s5, s7
; GFX8-NEXT: s_mov_b32 s6, s8
; GFX8-NEXT: s_mov_b32 s7, s9
; GFX8-NEXT: image_get_resinfo v[0:2], v0, s[0:7] dmask:0x7 unorm
; GFX8-NEXT: s_waitcnt vmcnt(0)
; GFX8-NEXT: ; return to shader part epilog
;
; GFX10-LABEL: getresinfo_dmask7:
; GFX10: ; %bb.0: ; %main_body
; GFX10-NEXT: s_mov_b32 s0, s2
; GFX10-NEXT: s_mov_b32 s1, s3
; GFX10-NEXT: s_mov_b32 s2, s4
; GFX10-NEXT: s_mov_b32 s3, s5
; GFX10-NEXT: s_mov_b32 s4, s6
; GFX10-NEXT: s_mov_b32 s5, s7
; GFX10-NEXT: s_mov_b32 s6, s8
; GFX10-NEXT: s_mov_b32 s7, s9
; GFX10-NEXT: ; implicit-def: $vcc_hi
; GFX10-NEXT: image_get_resinfo v[0:2], v0, s[0:7] dmask:0x7 dim:SQ_RSRC_IMG_1D unorm
; GFX10-NEXT: s_waitcnt vmcnt(0)
; GFX10-NEXT: ; return to shader part epilog
main_body:
%r = call <3 x float> @llvm.amdgcn.image.getresinfo.1d.v3f32.i32(i32 7, i32 %mip, <8 x i32> %rsrc, i32 0, i32 0)
ret <3 x float> %r
}
define amdgpu_ps <2 x float> @getresinfo_dmask3(<8 x i32> inreg %rsrc, <4 x float> %vdata, i32 %mip) {
; GFX6-LABEL: getresinfo_dmask3:
; GFX6: ; %bb.0: ; %main_body
; GFX6-NEXT: s_mov_b32 s0, s2
; GFX6-NEXT: s_mov_b32 s1, s3
; GFX6-NEXT: s_mov_b32 s2, s4
; GFX6-NEXT: s_mov_b32 s3, s5
; GFX6-NEXT: s_mov_b32 s4, s6
; GFX6-NEXT: s_mov_b32 s5, s7
; GFX6-NEXT: s_mov_b32 s6, s8
; GFX6-NEXT: s_mov_b32 s7, s9
; GFX6-NEXT: image_get_resinfo v[0:1], v0, s[0:7] dmask:0x3 unorm
; GFX6-NEXT: s_waitcnt vmcnt(0)
; GFX6-NEXT: ; return to shader part epilog
;
; GFX8-LABEL: getresinfo_dmask3:
; GFX8: ; %bb.0: ; %main_body
; GFX8-NEXT: s_mov_b32 s0, s2
; GFX8-NEXT: s_mov_b32 s1, s3
; GFX8-NEXT: s_mov_b32 s2, s4
; GFX8-NEXT: s_mov_b32 s3, s5
; GFX8-NEXT: s_mov_b32 s4, s6
; GFX8-NEXT: s_mov_b32 s5, s7
; GFX8-NEXT: s_mov_b32 s6, s8
; GFX8-NEXT: s_mov_b32 s7, s9
; GFX8-NEXT: image_get_resinfo v[0:1], v0, s[0:7] dmask:0x3 unorm
; GFX8-NEXT: s_waitcnt vmcnt(0)
; GFX8-NEXT: ; return to shader part epilog
;
; GFX10-LABEL: getresinfo_dmask3:
; GFX10: ; %bb.0: ; %main_body
; GFX10-NEXT: s_mov_b32 s0, s2
; GFX10-NEXT: s_mov_b32 s1, s3
; GFX10-NEXT: s_mov_b32 s2, s4
; GFX10-NEXT: s_mov_b32 s3, s5
; GFX10-NEXT: s_mov_b32 s4, s6
; GFX10-NEXT: s_mov_b32 s5, s7
; GFX10-NEXT: s_mov_b32 s6, s8
; GFX10-NEXT: s_mov_b32 s7, s9
; GFX10-NEXT: ; implicit-def: $vcc_hi
; GFX10-NEXT: image_get_resinfo v[0:1], v0, s[0:7] dmask:0x3 dim:SQ_RSRC_IMG_1D unorm
; GFX10-NEXT: s_waitcnt vmcnt(0)
; GFX10-NEXT: ; return to shader part epilog
main_body:
%r = call <2 x float> @llvm.amdgcn.image.getresinfo.1d.v2f32.i32(i32 3, i32 %mip, <8 x i32> %rsrc, i32 0, i32 0)
ret <2 x float> %r
}
define amdgpu_ps float @getresinfo_dmask1(<8 x i32> inreg %rsrc, <4 x float> %vdata, i32 %mip) {
; GFX6-LABEL: getresinfo_dmask1:
; GFX6: ; %bb.0: ; %main_body
; GFX6-NEXT: s_mov_b32 s0, s2
; GFX6-NEXT: s_mov_b32 s1, s3
; GFX6-NEXT: s_mov_b32 s2, s4
; GFX6-NEXT: s_mov_b32 s3, s5
; GFX6-NEXT: s_mov_b32 s4, s6
; GFX6-NEXT: s_mov_b32 s5, s7
; GFX6-NEXT: s_mov_b32 s6, s8
; GFX6-NEXT: s_mov_b32 s7, s9
; GFX6-NEXT: image_get_resinfo v0, v0, s[0:7] dmask:0x1 unorm
; GFX6-NEXT: s_waitcnt vmcnt(0)
; GFX6-NEXT: ; return to shader part epilog
;
; GFX8-LABEL: getresinfo_dmask1:
; GFX8: ; %bb.0: ; %main_body
; GFX8-NEXT: s_mov_b32 s0, s2
; GFX8-NEXT: s_mov_b32 s1, s3
; GFX8-NEXT: s_mov_b32 s2, s4
; GFX8-NEXT: s_mov_b32 s3, s5
; GFX8-NEXT: s_mov_b32 s4, s6
; GFX8-NEXT: s_mov_b32 s5, s7
; GFX8-NEXT: s_mov_b32 s6, s8
; GFX8-NEXT: s_mov_b32 s7, s9
; GFX8-NEXT: image_get_resinfo v0, v0, s[0:7] dmask:0x1 unorm
; GFX8-NEXT: s_waitcnt vmcnt(0)
; GFX8-NEXT: ; return to shader part epilog
;
; GFX10-LABEL: getresinfo_dmask1:
; GFX10: ; %bb.0: ; %main_body
; GFX10-NEXT: s_mov_b32 s0, s2
; GFX10-NEXT: s_mov_b32 s1, s3
; GFX10-NEXT: s_mov_b32 s2, s4
; GFX10-NEXT: s_mov_b32 s3, s5
; GFX10-NEXT: s_mov_b32 s4, s6
; GFX10-NEXT: s_mov_b32 s5, s7
; GFX10-NEXT: s_mov_b32 s6, s8
; GFX10-NEXT: s_mov_b32 s7, s9
; GFX10-NEXT: ; implicit-def: $vcc_hi
; GFX10-NEXT: image_get_resinfo v0, v0, s[0:7] dmask:0x1 dim:SQ_RSRC_IMG_1D unorm
; GFX10-NEXT: s_waitcnt vmcnt(0)
; GFX10-NEXT: ; return to shader part epilog
main_body:
%r = call float @llvm.amdgcn.image.getresinfo.1d.f32.i32(i32 1, i32 %mip, <8 x i32> %rsrc, i32 0, i32 0)
ret float %r
}
define amdgpu_ps <4 x float> @getresinfo_dmask0(<8 x i32> inreg %rsrc, <4 x float> %vdata, i32 %mip) { define amdgpu_ps <4 x float> @getresinfo_dmask0(<8 x i32> inreg %rsrc, <4 x float> %vdata, i32 %mip) {
; GFX6-LABEL: getresinfo_dmask0: ; GFX6-LABEL: getresinfo_dmask0:
; GFX6: ; %bb.0: ; %main_body ; GFX6: ; %bb.0: ; %main_body
; GFX6-NEXT: ; return to shader part epilog ; GFX6-NEXT: ; return to shader part epilog
; ;
; GFX8-LABEL: getresinfo_dmask0: ; GFX8-LABEL: getresinfo_dmask0:
; GFX8: ; %bb.0: ; %main_body ; GFX8: ; %bb.0: ; %main_body
; GFX8-NEXT: ; return to shader part epilog ; GFX8-NEXT: ; return to shader part epilog
; ;
; GFX10-LABEL: getresinfo_dmask0: ; GFX10-LABEL: getresinfo_dmask0:
; GFX10: ; %bb.0: ; %main_body ; GFX10: ; %bb.0: ; %main_body
; GFX10-NEXT: ; implicit-def: $vcc_hi ; GFX10-NEXT: ; implicit-def: $vcc_hi
; GFX10-NEXT: ; return to shader part epilog ; GFX10-NEXT: ; return to shader part epilog
main_body: main_body:
%r = call <4 x float> @llvm.amdgcn.image.getresinfo.1d.v4f32.i32(i32 0, i32 %mip, <8 x i32> %rsrc, i32 0, i32 0) %r = call <4 x float> @llvm.amdgcn.image.getresinfo.1d.v4f32.i32(i32 0, i32 %mip, <8 x i32> %rsrc, i32 0, i32 0)
ret <4 x float> %r ret <4 x float> %r
} }
declare <4 x float> @llvm.amdgcn.image.getresinfo.1d.v4f32.i32(i32 immarg, i32, <8 x i32>, i32 immarg, i32 immarg) #1 declare <4 x float> @llvm.amdgcn.image.getresinfo.1d.v4f32.i32(i32 immarg, i32, <8 x i32>, i32 immarg, i32 immarg) #1
declare <3 x float> @llvm.amdgcn.image.getresinfo.1d.v3f32.i32(i32 immarg, i32, <8 x i32>, i32 immarg, i32 immarg) #1
declare <2 x float> @llvm.amdgcn.image.getresinfo.1d.v2f32.i32(i32 immarg, i32, <8 x i32>, i32 immarg, i32 immarg) #1
declare float @llvm.amdgcn.image.getresinfo.1d.f32.i32(i32 immarg, i32, <8 x i32>, i32 immarg, i32 immarg) #1
declare <4 x float> @llvm.amdgcn.image.getresinfo.2d.v4f32.i32(i32 immarg, i32, <8 x i32>, i32 immarg, i32 immarg) #1 declare <4 x float> @llvm.amdgcn.image.getresinfo.2d.v4f32.i32(i32 immarg, i32, <8 x i32>, i32 immarg, i32 immarg) #1
declare <4 x float> @llvm.amdgcn.image.getresinfo.3d.v4f32.i32(i32 immarg, i32, <8 x i32>, i32 immarg, i32 immarg) #1 declare <4 x float> @llvm.amdgcn.image.getresinfo.3d.v4f32.i32(i32 immarg, i32, <8 x i32>, i32 immarg, i32 immarg) #1
declare <4 x float> @llvm.amdgcn.image.getresinfo.cube.v4f32.i32(i32 immarg, i32, <8 x i32>, i32 immarg, i32 immarg) #1 declare <4 x float> @llvm.amdgcn.image.getresinfo.cube.v4f32.i32(i32 immarg, i32, <8 x i32>, i32 immarg, i32 immarg) #1
declare <4 x float> @llvm.amdgcn.image.getresinfo.1darray.v4f32.i32(i32 immarg, i32, <8 x i32>, i32 immarg, i32 immarg) #1 declare <4 x float> @llvm.amdgcn.image.getresinfo.1darray.v4f32.i32(i32 immarg, i32, <8 x i32>, i32 immarg, i32 immarg) #1
declare <4 x float> @llvm.amdgcn.image.getresinfo.2darray.v4f32.i32(i32 immarg, i32, <8 x i32>, i32 immarg, i32 immarg) #1 declare <4 x float> @llvm.amdgcn.image.getresinfo.2darray.v4f32.i32(i32 immarg, i32, <8 x i32>, i32 immarg, i32 immarg) #1
declare <4 x float> @llvm.amdgcn.image.getresinfo.2dmsaa.v4f32.i32(i32 immarg, i32, <8 x i32>, i32 immarg, i32 immarg) #1 declare <4 x float> @llvm.amdgcn.image.getresinfo.2dmsaa.v4f32.i32(i32 immarg, i32, <8 x i32>, i32 immarg, i32 immarg) #1
declare <4 x float> @llvm.amdgcn.image.getresinfo.2darraymsaa.v4f32.i32(i32 immarg, i32, <8 x i32>, i32 immarg, i32 immarg) #1 declare <4 x float> @llvm.amdgcn.image.getresinfo.2darraymsaa.v4f32.i32(i32 immarg, i32, <8 x i32>, i32 immarg, i32 immarg) #1
attributes #0 = { nounwind } attributes #0 = { nounwind }
attributes #1 = { nounwind readnone } attributes #1 = { nounwind readnone }

llvm/test/CodeGen/AMDGPU/llvm.amdgcn.image.dim.ll

Show First 20 Lines • Show All 3,161 Lines • ▼ Show 20 Lines
; GFX10-NEXT: ; implicit-def: $vcc_hi ; GFX10-NEXT: ; implicit-def: $vcc_hi
; GFX10-NEXT: image_store v[0:3], v4, s[0:7] dmask:0xf dim:SQ_RSRC_IMG_1D unorm glc slc ; encoding: [0x00,0x3f,0x20,0xf2,0x04,0x00,0x00,0x00] ; GFX10-NEXT: image_store v[0:3], v4, s[0:7] dmask:0xf dim:SQ_RSRC_IMG_1D unorm glc slc ; encoding: [0x00,0x3f,0x20,0xf2,0x04,0x00,0x00,0x00]
; GFX10-NEXT: s_endpgm ; encoding: [0x00,0x00,0x81,0xbf] ; GFX10-NEXT: s_endpgm ; encoding: [0x00,0x00,0x81,0xbf]
main_body: main_body:
call void @llvm.amdgcn.image.store.1d.v4f32.i32(<4 x float> %vdata, i32 15, i32 %s, <8 x i32> %rsrc, i32 0, i32 3) call void @llvm.amdgcn.image.store.1d.v4f32.i32(<4 x float> %vdata, i32 15, i32 %s, <8 x i32> %rsrc, i32 0, i32 3)
ret void ret void
} }
define amdgpu_ps <3 x float> @getresinfo_dmask7(<8 x i32> inreg %rsrc, <4 x float> %vdata, i32 %mip) {
; GFX6-LABEL: getresinfo_dmask7:
; GFX6: ; %bb.0: ; %main_body
; GFX6-NEXT: s_mov_b32 s0, s2
; GFX6-NEXT: s_mov_b32 s1, s3
; GFX6-NEXT: s_mov_b32 s2, s4
; GFX6-NEXT: s_mov_b32 s3, s5
; GFX6-NEXT: s_mov_b32 s4, s6
; GFX6-NEXT: s_mov_b32 s5, s7
; GFX6-NEXT: s_mov_b32 s6, s8
; GFX6-NEXT: s_mov_b32 s7, s9
; GFX6-NEXT: image_get_resinfo v[0:2], v0, s[0:7] dmask:0x7 unorm
; GFX6-NEXT: s_waitcnt vmcnt(0)
; GFX6-NEXT: ; return to shader part epilog
;
; GFX8-LABEL: getresinfo_dmask7:
; GFX8: ; %bb.0: ; %main_body
; GFX8-NEXT: s_mov_b32 s0, s2
; GFX8-NEXT: s_mov_b32 s1, s3
; GFX8-NEXT: s_mov_b32 s2, s4
; GFX8-NEXT: s_mov_b32 s3, s5
; GFX8-NEXT: s_mov_b32 s4, s6
; GFX8-NEXT: s_mov_b32 s5, s7
; GFX8-NEXT: s_mov_b32 s6, s8
; GFX8-NEXT: s_mov_b32 s7, s9
; GFX8-NEXT: image_get_resinfo v[0:2], v0, s[0:7] dmask:0x7 unorm
; GFX8-NEXT: s_waitcnt vmcnt(0)
; GFX8-NEXT: ; return to shader part epilog
;
; VERDE-LABEL: getresinfo_dmask7:
; VERDE: ; %bb.0: ; %main_body
; VERDE-NEXT: image_get_resinfo v[0:2], v0, s[0:7] dmask:0x7 unorm
; VERDE-NEXT: s_waitcnt vmcnt(0)
; VERDE-NEXT: ; return to shader part epilog
;
; FIJI-LABEL: getresinfo_dmask7:
; FIJI: ; %bb.0: ; %main_body
; FIJI-NEXT: image_get_resinfo v[0:2], v0, s[0:7] dmask:0x7 unorm
; FIJI-NEXT: s_waitcnt vmcnt(0)
; FIJI-NEXT: ; return to shader part epilog
;
; GFX6789-LABEL: getresinfo_dmask7:
; GFX6789: ; %bb.0: ; %main_body
; GFX6789-NEXT: image_get_resinfo v[0:2], v0, s[0:7] dmask:0x7 unorm
; GFX6789-NEXT: s_waitcnt vmcnt(0)
; GFX6789-NEXT: ; return to shader part epilog
;
; NOPRT-LABEL: getresinfo_dmask7:
; NOPRT: ; %bb.0: ; %main_body
; NOPRT-NEXT: image_get_resinfo v[0:2], v0, s[0:7] dmask:0x7 unorm
; NOPRT-NEXT: s_waitcnt vmcnt(0)
; NOPRT-NEXT: ; return to shader part epilog
;
; GFX10-LABEL: getresinfo_dmask7:
; GFX10: ; %bb.0: ; %main_body
; GFX10-NEXT: image_get_resinfo v[0:2], v0, s[0:7] dmask:0x7 dim:SQ_RSRC_IMG_1D unorm ; encoding: [0x00,0x17,0x38,0xf0,0x00,0x00,0x00,0x00]
; GFX10-NEXT: ; implicit-def: $vcc_hi
; GFX10-NEXT: s_waitcnt vmcnt(0) ; encoding: [0x70,0x3f,0x8c,0xbf]
; GFX10-NEXT: ; return to shader part epilog
main_body:
%r = call <3 x float> @llvm.amdgcn.image.getresinfo.1d.v3f32.i32(i32 7, i32 %mip, <8 x i32> %rsrc, i32 0, i32 0)
ret <3 x float> %r
}
define amdgpu_ps <2 x float> @getresinfo_dmask3(<8 x i32> inreg %rsrc, <4 x float> %vdata, i32 %mip) {
; GFX6-LABEL: getresinfo_dmask3:
; GFX6: ; %bb.0: ; %main_body
; GFX6-NEXT: s_mov_b32 s0, s2
; GFX6-NEXT: s_mov_b32 s1, s3
; GFX6-NEXT: s_mov_b32 s2, s4
; GFX6-NEXT: s_mov_b32 s3, s5
; GFX6-NEXT: s_mov_b32 s4, s6
; GFX6-NEXT: s_mov_b32 s5, s7
; GFX6-NEXT: s_mov_b32 s6, s8
; GFX6-NEXT: s_mov_b32 s7, s9
; GFX6-NEXT: image_get_resinfo v[0:1], v0, s[0:7] dmask:0x3 unorm
; GFX6-NEXT: s_waitcnt vmcnt(0)
; GFX6-NEXT: ; return to shader part epilog
;
; GFX8-LABEL: getresinfo_dmask3:
; GFX8: ; %bb.0: ; %main_body
; GFX8-NEXT: s_mov_b32 s0, s2
; GFX8-NEXT: s_mov_b32 s1, s3
; GFX8-NEXT: s_mov_b32 s2, s4
; GFX8-NEXT: s_mov_b32 s3, s5
; GFX8-NEXT: s_mov_b32 s4, s6
; GFX8-NEXT: s_mov_b32 s5, s7
; GFX8-NEXT: s_mov_b32 s6, s8
; GFX8-NEXT: s_mov_b32 s7, s9
; GFX8-NEXT: image_get_resinfo v[0:1], v0, s[0:7] dmask:0x3 unorm
; GFX8-NEXT: s_waitcnt vmcnt(0)
; GFX8-NEXT: ; return to shader part epilog
;
; VERDE-LABEL: getresinfo_dmask3:
; VERDE: ; %bb.0: ; %main_body
; VERDE-NEXT: image_get_resinfo v[0:1], v0, s[0:7] dmask:0x3 unorm
; VERDE-NEXT: s_waitcnt vmcnt(0)
; VERDE-NEXT: ; return to shader part epilog
;
; FIJI-LABEL: getresinfo_dmask3:
; FIJI: ; %bb.0: ; %main_body
; FIJI-NEXT: image_get_resinfo v[0:1], v0, s[0:7] dmask:0x3 unorm
; FIJI-NEXT: s_waitcnt vmcnt(0)
; FIJI-NEXT: ; return to shader part epilog
;
; GFX6789-LABEL: getresinfo_dmask3:
; GFX6789: ; %bb.0: ; %main_body
; GFX6789-NEXT: image_get_resinfo v[0:1], v0, s[0:7] dmask:0x3 unorm
; GFX6789-NEXT: s_waitcnt vmcnt(0)
; GFX6789-NEXT: ; return to shader part epilog
;
; NOPRT-LABEL: getresinfo_dmask3:
; NOPRT: ; %bb.0: ; %main_body
; NOPRT-NEXT: image_get_resinfo v[0:1], v0, s[0:7] dmask:0x3 unorm
; NOPRT-NEXT: s_waitcnt vmcnt(0)
; NOPRT-NEXT: ; return to shader part epilog
;
; GFX10-LABEL: getresinfo_dmask3:
; GFX10: ; %bb.0: ; %main_body
; GFX10-NEXT: image_get_resinfo v[0:1], v0, s[0:7] dmask:0x3 dim:SQ_RSRC_IMG_1D unorm ; encoding: [0x00,0x13,0x38,0xf0,0x00,0x00,0x00,0x00]
; GFX10-NEXT: ; implicit-def: $vcc_hi
; GFX10-NEXT: s_waitcnt vmcnt(0) ; encoding: [0x70,0x3f,0x8c,0xbf]
; GFX10-NEXT: ; return to shader part epilog
main_body:
%r = call <2 x float> @llvm.amdgcn.image.getresinfo.1d.v2f32.i32(i32 3, i32 %mip, <8 x i32> %rsrc, i32 0, i32 0)
ret <2 x float> %r
}
define amdgpu_ps float @getresinfo_dmask1(<8 x i32> inreg %rsrc, <4 x float> %vdata, i32 %mip) {
; GFX6-LABEL: getresinfo_dmask1:
; GFX6: ; %bb.0: ; %main_body
; GFX6-NEXT: s_mov_b32 s0, s2
; GFX6-NEXT: s_mov_b32 s1, s3
; GFX6-NEXT: s_mov_b32 s2, s4
; GFX6-NEXT: s_mov_b32 s3, s5
; GFX6-NEXT: s_mov_b32 s4, s6
; GFX6-NEXT: s_mov_b32 s5, s7
; GFX6-NEXT: s_mov_b32 s6, s8
; GFX6-NEXT: s_mov_b32 s7, s9
; GFX6-NEXT: image_get_resinfo v0, v0, s[0:7] dmask:0x1 unorm
; GFX6-NEXT: s_waitcnt vmcnt(0)
; GFX6-NEXT: ; return to shader part epilog
;
; GFX8-LABEL: getresinfo_dmask1:
; GFX8: ; %bb.0: ; %main_body
; GFX8-NEXT: s_mov_b32 s0, s2
; GFX8-NEXT: s_mov_b32 s1, s3
; GFX8-NEXT: s_mov_b32 s2, s4
; GFX8-NEXT: s_mov_b32 s3, s5
; GFX8-NEXT: s_mov_b32 s4, s6
; GFX8-NEXT: s_mov_b32 s5, s7
; GFX8-NEXT: s_mov_b32 s6, s8
; GFX8-NEXT: s_mov_b32 s7, s9
; GFX8-NEXT: image_get_resinfo v0, v0, s[0:7] dmask:0x1 unorm
; GFX8-NEXT: s_waitcnt vmcnt(0)
; GFX8-NEXT: ; return to shader part epilog
;
; VERDE-LABEL: getresinfo_dmask1:
; VERDE: ; %bb.0: ; %main_body
; VERDE-NEXT: image_get_resinfo v0, v0, s[0:7] dmask:0x1 unorm
; VERDE-NEXT: s_waitcnt vmcnt(0)
; VERDE-NEXT: ; return to shader part epilog
;
; FIJI-LABEL: getresinfo_dmask1:
; FIJI: ; %bb.0: ; %main_body
; FIJI-NEXT: image_get_resinfo v0, v0, s[0:7] dmask:0x1 unorm
; FIJI-NEXT: s_waitcnt vmcnt(0)
; FIJI-NEXT: ; return to shader part epilog
;
; GFX6789-LABEL: getresinfo_dmask1:
; GFX6789: ; %bb.0: ; %main_body
; GFX6789-NEXT: image_get_resinfo v0, v0, s[0:7] dmask:0x1 unorm
; GFX6789-NEXT: s_waitcnt vmcnt(0)
; GFX6789-NEXT: ; return to shader part epilog
;
; NOPRT-LABEL: getresinfo_dmask1:
; NOPRT: ; %bb.0: ; %main_body
; NOPRT-NEXT: image_get_resinfo v0, v0, s[0:7] dmask:0x1 unorm
; NOPRT-NEXT: s_waitcnt vmcnt(0)
; NOPRT-NEXT: ; return to shader part epilog
;
; GFX10-LABEL: getresinfo_dmask1:
; GFX10: ; %bb.0: ; %main_body
; GFX10-NEXT: image_get_resinfo v0, v0, s[0:7] dmask:0x1 dim:SQ_RSRC_IMG_1D unorm ; encoding: [0x00,0x11,0x38,0xf0,0x00,0x00,0x00,0x00]
; GFX10-NEXT: ; implicit-def: $vcc_hi
; GFX10-NEXT: s_waitcnt vmcnt(0) ; encoding: [0x70,0x3f,0x8c,0xbf]
; GFX10-NEXT: ; return to shader part epilog
main_body:
%r = call float @llvm.amdgcn.image.getresinfo.1d.f32.i32(i32 1, i32 %mip, <8 x i32> %rsrc, i32 0, i32 0)
ret float %r
}
define amdgpu_ps <4 x float> @getresinfo_dmask0(<8 x i32> inreg %rsrc, <4 x float> %vdata, i32 %mip) #0 { define amdgpu_ps <4 x float> @getresinfo_dmask0(<8 x i32> inreg %rsrc, <4 x float> %vdata, i32 %mip) #0 {
; VERDE-LABEL: getresinfo_dmask0: ; VERDE-LABEL: getresinfo_dmask0:
; VERDE: ; %bb.0: ; %main_body ; VERDE: ; %bb.0: ; %main_body
; VERDE-NEXT: ; return to shader part epilog ; VERDE-NEXT: ; return to shader part epilog
; ;
; FIJI-LABEL: getresinfo_dmask0: ; FIJI-LABEL: getresinfo_dmask0:
; FIJI: ; %bb.0: ; %main_body ; FIJI: ; %bb.0: ; %main_body
; FIJI-NEXT: ; return to shader part epilog ; FIJI-NEXT: ; return to shader part epilog
▲ Show 20 Lines • Show All 171 Lines • ▼ Show 20 Lines
declare void @llvm.amdgcn.image.store.mip.1d.v4f32.i32(<4 x float>, i32, i32, i32, <8 x i32>, i32, i32) #0 declare void @llvm.amdgcn.image.store.mip.1d.v4f32.i32(<4 x float>, i32, i32, i32, <8 x i32>, i32, i32) #0
declare void @llvm.amdgcn.image.store.mip.2d.v4f32.i32(<4 x float>, i32, i32, i32, i32, <8 x i32>, i32, i32) #0 declare void @llvm.amdgcn.image.store.mip.2d.v4f32.i32(<4 x float>, i32, i32, i32, i32, <8 x i32>, i32, i32) #0
declare void @llvm.amdgcn.image.store.mip.3d.v4f32.i32(<4 x float>, i32, i32, i32, i32, i32, <8 x i32>, i32, i32) #0 declare void @llvm.amdgcn.image.store.mip.3d.v4f32.i32(<4 x float>, i32, i32, i32, i32, i32, <8 x i32>, i32, i32) #0
declare void @llvm.amdgcn.image.store.mip.cube.v4f32.i32(<4 x float>, i32, i32, i32, i32, i32, <8 x i32>, i32, i32) #0 declare void @llvm.amdgcn.image.store.mip.cube.v4f32.i32(<4 x float>, i32, i32, i32, i32, i32, <8 x i32>, i32, i32) #0
declare void @llvm.amdgcn.image.store.mip.1darray.v4f32.i32(<4 x float>, i32, i32, i32, i32, <8 x i32>, i32, i32) #0 declare void @llvm.amdgcn.image.store.mip.1darray.v4f32.i32(<4 x float>, i32, i32, i32, i32, <8 x i32>, i32, i32) #0
declare void @llvm.amdgcn.image.store.mip.2darray.v4f32.i32(<4 x float>, i32, i32, i32, i32, i32, <8 x i32>, i32, i32) #0 declare void @llvm.amdgcn.image.store.mip.2darray.v4f32.i32(<4 x float>, i32, i32, i32, i32, i32, <8 x i32>, i32, i32) #0
declare <4 x float> @llvm.amdgcn.image.getresinfo.1d.v4f32.i32(i32, i32, <8 x i32>, i32, i32) #2 declare <4 x float> @llvm.amdgcn.image.getresinfo.1d.v4f32.i32(i32, i32, <8 x i32>, i32, i32) #2
declare <3 x float> @llvm.amdgcn.image.getresinfo.1d.v3f32.i32(i32 immarg, i32, <8 x i32>, i32 immarg, i32 immarg) #1
declare <2 x float> @llvm.amdgcn.image.getresinfo.1d.v2f32.i32(i32 immarg, i32, <8 x i32>, i32 immarg, i32 immarg) #1
declare float @llvm.amdgcn.image.getresinfo.1d.f32.i32(i32 immarg, i32, <8 x i32>, i32 immarg, i32 immarg) #1
declare <4 x float> @llvm.amdgcn.image.getresinfo.2d.v4f32.i32(i32, i32, <8 x i32>, i32, i32) #2 declare <4 x float> @llvm.amdgcn.image.getresinfo.2d.v4f32.i32(i32, i32, <8 x i32>, i32, i32) #2
declare <4 x float> @llvm.amdgcn.image.getresinfo.3d.v4f32.i32(i32, i32, <8 x i32>, i32, i32) #2 declare <4 x float> @llvm.amdgcn.image.getresinfo.3d.v4f32.i32(i32, i32, <8 x i32>, i32, i32) #2
declare <4 x float> @llvm.amdgcn.image.getresinfo.cube.v4f32.i32(i32, i32, <8 x i32>, i32, i32) #2 declare <4 x float> @llvm.amdgcn.image.getresinfo.cube.v4f32.i32(i32, i32, <8 x i32>, i32, i32) #2
declare <4 x float> @llvm.amdgcn.image.getresinfo.1darray.v4f32.i32(i32, i32, <8 x i32>, i32, i32) #2 declare <4 x float> @llvm.amdgcn.image.getresinfo.1darray.v4f32.i32(i32, i32, <8 x i32>, i32, i32) #2
declare <4 x float> @llvm.amdgcn.image.getresinfo.2darray.v4f32.i32(i32, i32, <8 x i32>, i32, i32) #2 declare <4 x float> @llvm.amdgcn.image.getresinfo.2darray.v4f32.i32(i32, i32, <8 x i32>, i32, i32) #2
declare <4 x float> @llvm.amdgcn.image.getresinfo.2dmsaa.v4f32.i32(i32, i32, <8 x i32>, i32, i32) #2 declare <4 x float> @llvm.amdgcn.image.getresinfo.2dmsaa.v4f32.i32(i32, i32, <8 x i32>, i32, i32) #2
declare <4 x float> @llvm.amdgcn.image.getresinfo.2darraymsaa.v4f32.i32(i32, i32, <8 x i32>, i32, i32) #2 declare <4 x float> @llvm.amdgcn.image.getresinfo.2darraymsaa.v4f32.i32(i32, i32, <8 x i32>, i32, i32) #2
Show All 9 Lines