[X86][BtVer2] Fix WriteFShuffle256 schedule write info.

[andreadb]

Summary

This patch fixes the number of micro opcodes, and processor resource cycles for the following AVX instructions:

vinsertf128rr/rm
vperm2f128rr/rm

Tests have been regenerated using the usual scripts in the llvm/utils directory.

Please let me know if okay to commit.

Thanks @RKSimon for spotting the issue with WriteFShuffle256.

Andrea

[RKSimon]

test/tools/llvm-mca/X86/BtVer2/resources-avx1.s

1791 ↗

(On Diff #163346)

vbroadcastf128 looks dodgy as well

[lebedev.ri]

[MCA can get the latency, and micro-ops for the most part, but ]

processor resource cycles

Really ignorant question: how do you decide what is the right values?
Is there some documentation?

[RKSimon]

In D51492#1219429, @lebedev.ri wrote:

Really ignorant question: how do you decide what is the right values?
Is there some documentation?

AMD's Fam16h SOG which includes a latency spreadsheet - https://developer.amd.com/resources/developer-guides-manuals/

The Fam15h SOG have similar charts at the back of the PDF as do many of the Intel CPUs in their AOM docs

[lebedev.ri]

In D51492#1219441, @RKSimon wrote:

In D51492#1219429, @lebedev.ri wrote:

Really ignorant question: how do you decide what is the right values?
Is there some documentation?

AMD's Fam16h SOG which includes a latency spreadsheet - https://developer.amd.com/resources/developer-guides-manuals/

The Fam15h SOG have similar charts at the back of the PDF as do many of the Intel CPUs in their AOM docs

Hm, perhaps i wrote too much :)
I was specifically asking *not* about the latency/micro-ops, but about the resource cycles.

[avt77]

I don't see any changes for VEXTRACTF128 in tests. Do you really need this JWriteVecExtractF128? If YES you should add the corresponding test.

[andreadb]

In D51492#1219494, @avt77 wrote:

I don't see any changes for VEXTRACTF128 in tests. Do you really need this JWriteVecExtractF128? If YES you should add the corresponding test.

There is no change in tests for VEXTRACTF128 because I didn't touch its latency/throughput profile info.

I need JWriteVecExtractF128 because otherwise I would affect profile info for VEXTRACTF128, which is already correct.
I don't think that I need more tests for it, as there are already existing tests for that instruction (both in llvm-mca and CodeGen/X86).

@lebedev.ri , the amdfam16h SOG also reports that data paths are 128-bits wide. AVX 256-bit instructions are effectively split into two opcodes, and consume twice as many pipeline resources as their 128-bit counterpart. The spreadsheet in the SOG should also report the reciprocal throughput (that's how I derived those resource cycles).

[andreadb]

Patch updated with a fix for the throughput information for VBROADCASTF128.

This should address Simon's review comment.

[RKSimon]

LGTM - thanks

llvm/trunk/lib/Target/X86/X86ScheduleBtVer2.td

 defm : JWriteResYMMPair<WriteFVarShuffleY,[JFPU01, JFPX],  3, [2, 6], 6>;
 defm : X86WriteResPairUnsupported<WriteFVarShuffleZ>;
 defm : JWriteResFpuPair<WriteFBlend,      [JFPU01, JFPX],  1>;
 defm : JWriteResYMMPair<WriteFBlendY,     [JFPU01, JFPX],  1, [2, 2], 2>;
 defm : X86WriteResPairUnsupported<WriteFBlendZ>;
 defm : JWriteResFpuPair<WriteFVarBlend,   [JFPU01, JFPX],  2, [1, 4], 3>;
 defm : JWriteResYMMPair<WriteFVarBlendY,  [JFPU01, JFPX],  3, [2, 6], 6>;
 defm : X86WriteResPairUnsupported<WriteFVarBlendZ>;
-defm : JWriteResFpuPair<WriteFShuffle256, [JFPU01, JFPX],  1>;
+defm : JWriteResFpuPair<WriteFShuffle256, [JFPU01, JFPX],  1, [2, 2], 2>;
 defm : X86WriteResPairUnsupported<WriteFVarShuffle256>;
 
 ////////////////////////////////////////////////////////////////////////////////
 // Conversions.
 ////////////////////////////////////////////////////////////////////////////////
 
 defm : JWriteResFpuPair<WriteCvtSS2I,      [JFPU1, JSTC, JFPA, JALU0], 7, [1,1,1,1], 2>;
 defm : JWriteResFpuPair<WriteCvtPS2I,      [JFPU1, JSTC], 3, [1,1], 1>;
def JWriteINSERTQ: SchedWriteRes<[JFPU01, JVALU]> {
   let ResourceCycles = [1, 4];
 }
 def : InstRW<[JWriteINSERTQ], (instrs INSERTQ, INSERTQI)>;
 
 ////////////////////////////////////////////////////////////////////////////////
 // AVX instructions.
 ////////////////////////////////////////////////////////////////////////////////
 
+def JWriteVecExtractF128: SchedWriteRes<[JFPU01, JFPX]>;
+def : InstRW<[JWriteVecExtractF128], (instrs VEXTRACTF128rr)>;
+
 def JWriteVBROADCASTYLd: SchedWriteRes<[JLAGU, JFPU01, JFPX]> {
   let Latency = 6;
   let ResourceCycles = [1, 2, 4];
   let NumMicroOps = 2;
 }
 def : InstRW<[JWriteVBROADCASTYLd, ReadAfterLd], (instrs VBROADCASTSDYrm,
-                                                         VBROADCASTSSYrm)>;
+                                                         VBROADCASTSSYrm,
+                                                         VBROADCASTF128)>;
 
 def JWriteJVZEROALL: SchedWriteRes<[]> {
   let Latency = 90;
   let NumMicroOps = 73;
 }
 def : InstRW<[JWriteJVZEROALL], (instrs VZEROALL)>;
 
 def JWriteJVZEROUPPER: SchedWriteRes<[]> {

llvm/trunk/test/CodeGen/X86/avx-schedule.ll

 ;
 ; SKX-LABEL: test_broadcastf128:
 ; SKX:       # %bb.0:
 ; SKX-NEXT:    vbroadcastf128 {{.*#+}} ymm0 = mem[0,1,0,1] sched: [7:0.50]
 ; SKX-NEXT:    retq # sched: [7:1.00]
 ;
 ; BTVER2-LABEL: test_broadcastf128:
 ; BTVER2:       # %bb.0:
-; BTVER2-NEXT:    vbroadcastf128 {{.*#+}} ymm0 = mem[0,1,0,1] sched: [6:1.00]
+; BTVER2-NEXT:    vbroadcastf128 {{.*#+}} ymm0 = mem[0,1,0,1] sched: [6:2.00]
 ; BTVER2-NEXT:    retq # sched: [4:1.00]
 ;
 ; ZNVER1-LABEL: test_broadcastf128:
 ; ZNVER1:       # %bb.0:
 ; ZNVER1-NEXT:    vbroadcastf128 {{.*#+}} ymm0 = mem[0,1,0,1] sched: [8:0.50]
 ; ZNVER1-NEXT:    retq # sched: [1:0.50]
   %1 = load <4 x float>, <4 x float> *%a0, align 32
   %2 = shufflevector <4 x float> %1, <4 x float> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3>

 ; SKX-NEXT:    vcvtpd2dqy (%rdi), %xmm1 # sched: [8:0.50]
 ; SKX-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0 # sched: [3:1.00]
 ; SKX-NEXT:    retq # sched: [7:1.00]
 ;
 ; BTVER2-LABEL: test_cvtpd2dq:
 ; BTVER2:       # %bb.0:
 ; BTVER2-NEXT:    vcvtpd2dqy (%rdi), %xmm1 # sched: [11:2.00]
 ; BTVER2-NEXT:    vcvtpd2dq %ymm0, %xmm0 # sched: [6:2.00]
-; BTVER2-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0 # sched: [1:0.50]
+; BTVER2-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0 # sched: [1:1.00]
 ; BTVER2-NEXT:    retq # sched: [4:1.00]
 ;
 ; ZNVER1-LABEL: test_cvtpd2dq:
 ; ZNVER1:       # %bb.0:
 ; ZNVER1-NEXT:    vcvtpd2dqy (%rdi), %xmm1 # sched: [12:1.00]
 ; ZNVER1-NEXT:    vcvtpd2dq %ymm0, %xmm0 # sched: [5:1.00]
 ; ZNVER1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0 # sched: [2:0.67]
 ; ZNVER1-NEXT:    retq # sched: [1:0.50]

 ; SKX-NEXT:    vcvttpd2dqy (%rdi), %xmm1 # sched: [8:0.50]
 ; SKX-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0 # sched: [3:1.00]
 ; SKX-NEXT:    retq # sched: [7:1.00]
 ;
 ; BTVER2-LABEL: test_cvttpd2dq:
 ; BTVER2:       # %bb.0:
 ; BTVER2-NEXT:    vcvttpd2dqy (%rdi), %xmm1 # sched: [11:2.00]
 ; BTVER2-NEXT:    vcvttpd2dq %ymm0, %xmm0 # sched: [6:2.00]
-; BTVER2-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0 # sched: [1:0.50]
+; BTVER2-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0 # sched: [1:1.00]
 ; BTVER2-NEXT:    retq # sched: [4:1.00]
 ;
 ; ZNVER1-LABEL: test_cvttpd2dq:
 ; ZNVER1:       # %bb.0:
 ; ZNVER1-NEXT:    vcvttpd2dqy (%rdi), %xmm1 # sched: [12:1.00]
 ; ZNVER1-NEXT:    vcvttpd2dq %ymm0, %xmm0 # sched: [5:1.00]
 ; ZNVER1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0 # sched: [2:0.67]
 ; ZNVER1-NEXT:    retq # sched: [1:0.50]

 ; SKX-NEXT:    vcvtpd2psy (%rdi), %xmm1 # sched: [8:1.00]
 ; SKX-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0 # sched: [3:1.00]
 ; SKX-NEXT:    retq # sched: [7:1.00]
 ;
 ; BTVER2-LABEL: test_cvtpd2ps:
 ; BTVER2:       # %bb.0:
 ; BTVER2-NEXT:    vcvtpd2psy (%rdi), %xmm1 # sched: [11:2.00]
 ; BTVER2-NEXT:    vcvtpd2ps %ymm0, %xmm0 # sched: [6:2.00]
-; BTVER2-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0 # sched: [1:0.50]
+; BTVER2-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0 # sched: [1:1.00]
 ; BTVER2-NEXT:    retq # sched: [4:1.00]
 ;
 ; ZNVER1-LABEL: test_cvtpd2ps:
 ; ZNVER1:       # %bb.0:
 ; ZNVER1-NEXT:    vcvtpd2psy (%rdi), %xmm1 # sched: [11:1.00]
 ; ZNVER1-NEXT:    vcvtpd2ps %ymm0, %xmm0 # sched: [5:1.00]
 ; ZNVER1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0 # sched: [2:0.67]
 ; ZNVER1-NEXT:    retq # sched: [1:0.50]

 ; SKX:       # %bb.0:
 ; SKX-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm1 # sched: [3:1.00]
 ; SKX-NEXT:    vinsertf128 $1, (%rdi), %ymm0, %ymm0 # sched: [7:0.50]
 ; SKX-NEXT:    vaddps %ymm0, %ymm1, %ymm0 # sched: [4:0.50]
 ; SKX-NEXT:    retq # sched: [7:1.00]
 ;
 ; BTVER2-LABEL: test_insertf128:
 ; BTVER2:       # %bb.0:
-; BTVER2-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm1 # sched: [1:0.50]
+; BTVER2-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm1 # sched: [1:1.00]
 ; BTVER2-NEXT:    vinsertf128 $1, (%rdi), %ymm0, %ymm0 # sched: [6:1.00]
 ; BTVER2-NEXT:    vaddps %ymm0, %ymm1, %ymm0 # sched: [3:2.00]
 ; BTVER2-NEXT:    retq # sched: [4:1.00]
 ;
 ; ZNVER1-LABEL: test_insertf128:
 ; ZNVER1:       # %bb.0:
 ; ZNVER1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm1 # sched: [2:0.67]
 ; ZNVER1-NEXT:    vinsertf128 $1, (%rdi), %ymm0, %ymm0 # sched: [9:0.67]

 ; SKX:       # %bb.0:
 ; SKX-NEXT:    vperm2f128 {{.*#+}} ymm1 = ymm0[2,3],ymm1[0,1] sched: [3:1.00]
 ; SKX-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],mem[0,1] sched: [10:1.00]
 ; SKX-NEXT:    vaddpd %ymm0, %ymm1, %ymm0 # sched: [4:0.50]
 ; SKX-NEXT:    retq # sched: [7:1.00]
 ;
 ; BTVER2-LABEL: test_perm2f128:
 ; BTVER2:       # %bb.0:
-; BTVER2-NEXT:    vperm2f128 {{.*#+}} ymm1 = ymm0[2,3],ymm1[0,1] sched: [1:0.50]
+; BTVER2-NEXT:    vperm2f128 {{.*#+}} ymm1 = ymm0[2,3],ymm1[0,1] sched: [1:1.00]
 ; BTVER2-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],mem[0,1] sched: [6:1.00]
 ; BTVER2-NEXT:    vaddpd %ymm0, %ymm1, %ymm0 # sched: [3:2.00]
 ; BTVER2-NEXT:    retq # sched: [4:1.00]
 ;
 ; ZNVER1-LABEL: test_perm2f128:
 ; ZNVER1:       # %bb.0:
 ; ZNVER1-NEXT:    vperm2f128 {{.*#+}} ymm1 = ymm0[2,3],ymm1[0,1] sched: [100:0.25]
 ; ZNVER1-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],mem[0,1] sched: [100:0.25]

llvm/trunk/test/tools/llvm-mca/X86/BtVer2/resources-avx1.s

 # CHECK-NEXT:  3      2     2.00                        vblendvpd %xmm3, %xmm0, %xmm1, %xmm2
 # CHECK-NEXT:  3      7     2.00    *                   vblendvpd %xmm3, (%rax), %xmm1, %xmm2
 # CHECK-NEXT:  6      3     3.00                        vblendvpd %ymm3, %ymm0, %ymm1, %ymm2
 # CHECK-NEXT:  6      8     3.00    *                   vblendvpd %ymm3, (%rax), %ymm1, %ymm2
 # CHECK-NEXT:  3      2     2.00                        vblendvps %xmm3, %xmm0, %xmm1, %xmm2
 # CHECK-NEXT:  3      7     2.00    *                   vblendvps %xmm3, (%rax), %xmm1, %xmm2
 # CHECK-NEXT:  6      3     3.00                        vblendvps %ymm3, %ymm0, %ymm1, %ymm2
 # CHECK-NEXT:  6      8     3.00    *                   vblendvps %ymm3, (%rax), %ymm1, %ymm2
-# CHECK-NEXT:  1      6     1.00    *                   vbroadcastf128  (%rax), %ymm2
+# CHECK-NEXT:  2      6     2.00    *                   vbroadcastf128  (%rax), %ymm2
 # CHECK-NEXT:  2      6     2.00    *                   vbroadcastsd  (%rax), %ymm2
 # CHECK-NEXT:  1      6     1.00    *                   vbroadcastss  (%rax), %xmm2
 # CHECK-NEXT:  2      6     2.00    *                   vbroadcastss  (%rax), %ymm2
 # CHECK-NEXT:  1      2     1.00                        vcmppd  $0, %xmm0, %xmm1, %xmm2
 # CHECK-NEXT:  1      7     1.00    *                   vcmppd  $0, (%rax), %xmm1, %xmm2
 # CHECK-NEXT:  2      2     2.00                        vcmppd  $0, %ymm0, %ymm1, %ymm2
 # CHECK-NEXT:  2      7     2.00    *                   vcmppd  $0, (%rax), %ymm1, %ymm2
 # CHECK-NEXT:  1      2     1.00                        vcmpps  $0, %xmm0, %xmm1, %xmm2

 # CHECK-NEXT:  1      3     1.00                        vhsubpd %xmm0, %xmm1, %xmm2
 # CHECK-NEXT:  1      8     1.00    *                   vhsubpd (%rax), %xmm1, %xmm2
 # CHECK-NEXT:  2      3     2.00                        vhsubpd %ymm0, %ymm1, %ymm2
 # CHECK-NEXT:  2      8     2.00    *                   vhsubpd (%rax), %ymm1, %ymm2
 # CHECK-NEXT:  1      3     1.00                        vhsubps %xmm0, %xmm1, %xmm2
 # CHECK-NEXT:  1      8     1.00    *                   vhsubps (%rax), %xmm1, %xmm2
 # CHECK-NEXT:  2      3     2.00                        vhsubps %ymm0, %ymm1, %ymm2
 # CHECK-NEXT:  2      8     2.00    *                   vhsubps (%rax), %ymm1, %ymm2
-# CHECK-NEXT:  1      1     0.50                        vinsertf128 $1, %xmm0, %ymm1, %ymm2
-# CHECK-NEXT:  1      6     1.00    *                   vinsertf128 $1, (%rax), %ymm1, %ymm2
+# CHECK-NEXT:  2      1     1.00                        vinsertf128 $1, %xmm0, %ymm1, %ymm2
+# CHECK-NEXT:  2      6     1.00    *                   vinsertf128 $1, (%rax), %ymm1, %ymm2
 # CHECK-NEXT:  1      1     0.50                        vinsertps $1, %xmm0, %xmm1, %xmm2
 # CHECK-NEXT:  1      6     1.00    *                   vinsertps $1, (%rax), %xmm1, %xmm2
 # CHECK-NEXT:  1      5     1.00    *                   vlddqu  (%rax), %xmm2
 # CHECK-NEXT:  1      5     1.00    *                   vlddqu  (%rax), %ymm2
 # CHECK-NEXT:  1      5     1.00    *      *      U     vldmxcsr  (%rax)
 # CHECK-NEXT:  1      1     1.00    *      *      U     vmaskmovdqu %xmm0, %xmm1
 # CHECK-NEXT:  1      6     1.00    *                   vmaskmovpd  (%rax), %xmm0, %xmm2
 # CHECK-NEXT:  2      6     2.00    *                   vmaskmovpd  (%rax), %ymm0, %ymm2

 # CHECK-NEXT:  1      1     0.50                        vpcmpgtb  %xmm0, %xmm1, %xmm2
 # CHECK-NEXT:  1      6     1.00    *                   vpcmpgtb  (%rax), %xmm1, %xmm2
 # CHECK-NEXT:  1      1     0.50                        vpcmpgtd  %xmm0, %xmm1, %xmm2
 # CHECK-NEXT:  1      6     1.00    *                   vpcmpgtd  (%rax), %xmm1, %xmm2
 # CHECK-NEXT:  1      1     0.50                        vpcmpgtq  %xmm0, %xmm1, %xmm2
 # CHECK-NEXT:  1      6     1.00    *                   vpcmpgtq  (%rax), %xmm1, %xmm2
 # CHECK-NEXT:  1      1     0.50                        vpcmpgtw  %xmm0, %xmm1, %xmm2
 # CHECK-NEXT:  1      6     1.00    *                   vpcmpgtw  (%rax), %xmm1, %xmm2
-# CHECK-NEXT:  1      1     0.50                        vperm2f128  $1, %ymm0, %ymm1, %ymm2
-# CHECK-NEXT:  1      6     1.00    *                   vperm2f128  $1, (%rax), %ymm1, %ymm2
+# CHECK-NEXT:  2      1     1.00                        vperm2f128  $1, %ymm0, %ymm1, %ymm2
+# CHECK-NEXT:  2      6     1.00    *                   vperm2f128  $1, (%rax), %ymm1, %ymm2
 # CHECK-NEXT:  1      1     0.50                        vpermilpd $1, %xmm0, %xmm2
 # CHECK-NEXT:  1      6     1.00    *                   vpermilpd $1, (%rax), %xmm2
 # CHECK-NEXT:  3      2     2.00                        vpermilpd %xmm0, %xmm1, %xmm2
 # CHECK-NEXT:  3      7     2.00    *                   vpermilpd (%rax), %xmm1, %xmm2
 # CHECK-NEXT:  2      1     1.00                        vpermilpd $1, %ymm0, %ymm2
 # CHECK-NEXT:  2      6     2.00    *                   vpermilpd $1, (%rax), %ymm2
 # CHECK-NEXT:  6      3     3.00                        vpermilpd %ymm0, %ymm1, %ymm2
 # CHECK-NEXT:  6      8     3.00    *                   vpermilpd (%rax), %ymm1, %ymm2

 # CHECK-NEXT: [9]   - JSAGU
 # CHECK-NEXT: [10]  - JSTC
 # CHECK-NEXT: [11]  - JVALU0
 # CHECK-NEXT: [12]  - JVALU1
 # CHECK-NEXT: [13]  - JVIMUL
 
 # CHECK:      Resource pressure per iteration:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]    [10]   [11]   [12]   [13]
-# CHECK-NEXT: 48.00   -      -     353.50 911.50 402.00 422.00 382.00  -     43.00  132.00 119.50 119.50 38.00
+# CHECK-NEXT: 48.00   -      -     357.00 915.00 404.50 424.50 382.00  -     43.00  132.00 119.50 119.50 38.00
 
 # CHECK:      Resource pressure by instruction:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]    [10]   [11]   [12]   [13]   Instructions:
 # CHECK-NEXT:  -      -      -     1.00    -     1.00    -      -      -      -      -      -      -      -     vaddpd  %xmm0, %xmm1, %xmm2
 # CHECK-NEXT:  -      -      -     1.00    -     1.00    -     1.00    -      -      -      -      -      -     vaddpd  (%rax), %xmm1, %xmm2
 # CHECK-NEXT:  -      -      -     2.00    -     2.00    -      -      -      -      -      -      -      -     vaddpd  %ymm0, %ymm1, %ymm2
 # CHECK-NEXT:  -      -      -     2.00    -     2.00    -     2.00    -      -      -      -      -      -     vaddpd  (%rax), %ymm1, %ymm2
 # CHECK-NEXT:  -      -      -     1.00    -     1.00    -      -      -      -      -      -      -      -     vaddps  %xmm0, %xmm1, %xmm2

 # CHECK-NEXT:  -      -      -     2.00   2.00   0.50   0.50    -      -      -      -      -      -      -     vblendvpd %xmm3, %xmm0, %xmm1, %xmm2
 # CHECK-NEXT:  -      -      -     2.00   2.00   0.50   0.50   1.00    -      -      -      -      -      -     vblendvpd %xmm3, (%rax), %xmm1, %xmm2
 # CHECK-NEXT:  -      -      -     3.00   3.00   1.00   1.00    -      -      -      -      -      -      -     vblendvpd %ymm3, %ymm0, %ymm1, %ymm2
 # CHECK-NEXT:  -      -      -     3.00   3.00   1.00   1.00   2.00    -      -      -      -      -      -     vblendvpd %ymm3, (%rax), %ymm1, %ymm2
 # CHECK-NEXT:  -      -      -     2.00   2.00   0.50   0.50    -      -      -      -      -      -      -     vblendvps %xmm3, %xmm0, %xmm1, %xmm2
 # CHECK-NEXT:  -      -      -     2.00   2.00   0.50   0.50   1.00    -      -      -      -      -      -     vblendvps %xmm3, (%rax), %xmm1, %xmm2
 # CHECK-NEXT:  -      -      -     3.00   3.00   1.00   1.00    -      -      -      -      -      -      -     vblendvps %ymm3, %ymm0, %ymm1, %ymm2
 # CHECK-NEXT:  -      -      -     3.00   3.00   1.00   1.00   2.00    -      -      -      -      -      -     vblendvps %ymm3, (%rax), %ymm1, %ymm2
-# CHECK-NEXT:  -      -      -     0.50   0.50   0.50   0.50   1.00    -      -      -      -      -      -     vbroadcastf128  (%rax), %ymm2
+# CHECK-NEXT:  -      -      -     2.00   2.00   1.00   1.00   1.00    -      -      -      -      -      -     vbroadcastf128  (%rax), %ymm2
 # CHECK-NEXT:  -      -      -     2.00   2.00   1.00   1.00   1.00    -      -      -      -      -      -     vbroadcastsd  (%rax), %ymm2
 # CHECK-NEXT:  -      -      -     0.50   0.50   0.50   0.50   1.00    -      -      -      -      -      -     vbroadcastss  (%rax), %xmm2
 # CHECK-NEXT:  -      -      -     2.00   2.00   1.00   1.00   1.00    -      -      -      -      -      -     vbroadcastss  (%rax), %ymm2
 # CHECK-NEXT:  -      -      -     1.00    -     1.00    -      -      -      -      -      -      -      -     vcmppd  $0, %xmm0, %xmm1, %xmm2
 # CHECK-NEXT:  -      -      -     1.00    -     1.00    -     1.00    -      -      -      -      -      -     vcmppd  $0, (%rax), %xmm1, %xmm2
 # CHECK-NEXT:  -      -      -     2.00    -     2.00    -      -      -      -      -      -      -      -     vcmppd  $0, %ymm0, %ymm1, %ymm2
 # CHECK-NEXT:  -      -      -     2.00    -     2.00    -     2.00    -      -      -      -      -      -     vcmppd  $0, (%rax), %ymm1, %ymm2
 # CHECK-NEXT:  -      -      -     1.00    -     1.00    -      -      -      -      -      -      -      -     vcmpps  $0, %xmm0, %xmm1, %xmm2

 # CHECK-NEXT:  -      -      -     1.00    -     1.00    -      -      -      -      -      -      -      -     vhsubpd %xmm0, %xmm1, %xmm2
 # CHECK-NEXT:  -      -      -     1.00    -     1.00    -     1.00    -      -      -      -      -      -     vhsubpd (%rax), %xmm1, %xmm2
 # CHECK-NEXT:  -      -      -     2.00    -     2.00    -      -      -      -      -      -      -      -     vhsubpd %ymm0, %ymm1, %ymm2
 # CHECK-NEXT:  -      -      -     2.00    -     2.00    -     2.00    -      -      -      -      -      -     vhsubpd (%rax), %ymm1, %ymm2
 # CHECK-NEXT:  -      -      -     1.00    -     1.00    -      -      -      -      -      -      -      -     vhsubps %xmm0, %xmm1, %xmm2
 # CHECK-NEXT:  -      -      -     1.00    -     1.00    -     1.00    -      -      -      -      -      -     vhsubps (%rax), %xmm1, %xmm2
 # CHECK-NEXT:  -      -      -     2.00    -     2.00    -      -      -      -      -      -      -      -     vhsubps %ymm0, %ymm1, %ymm2
 # CHECK-NEXT:  -      -      -     2.00    -     2.00    -     2.00    -      -      -      -      -      -     vhsubps (%rax), %ymm1, %ymm2
-# CHECK-NEXT:  -      -      -     0.50   0.50   0.50   0.50    -      -      -      -      -      -      -     vinsertf128  $1, %xmm0, %ymm1, %ymm2
-# CHECK-NEXT:  -      -      -     0.50   0.50   0.50   0.50   1.00    -      -      -      -      -      -     vinsertf128  $1, (%rax), %ymm1, %ymm2
+# CHECK-NEXT:  -      -      -     1.00   1.00   1.00   1.00    -      -      -      -      -      -      -     vinsertf128  $1, %xmm0, %ymm1, %ymm2
+# CHECK-NEXT:  -      -      -     1.00   1.00   1.00   1.00   1.00    -      -      -      -      -      -     vinsertf128  $1, (%rax), %ymm1, %ymm2
 # CHECK-NEXT:  -      -      -     0.50   0.50   0.50   0.50    -      -      -      -      -      -      -     vinsertps $1, %xmm0, %xmm1, %xmm2
 # CHECK-NEXT:  -      -      -     0.50   0.50   0.50   0.50   1.00    -      -      -      -      -      -     vinsertps $1, (%rax), %xmm1, %xmm2
 # CHECK-NEXT:  -      -      -      -      -     0.50   0.50   1.00    -      -      -     0.50   0.50    -     vlddqu  (%rax), %xmm2
 # CHECK-NEXT:  -      -      -      -      -     0.50   0.50   1.00    -      -      -     0.50   0.50    -     vlddqu  (%rax), %ymm2
 # CHECK-NEXT:  -      -      -      -      -      -      -     1.00    -      -      -      -      -      -     vldmxcsr  (%rax)
 # CHECK-NEXT:  -      -      -      -      -      -     1.00    -      -     1.00   1.00    -      -      -     vmaskmovdqu %xmm0, %xmm1
 # CHECK-NEXT:  -      -      -     1.00   1.00   0.50   0.50   1.00    -      -      -      -      -      -     vmaskmovpd  (%rax), %xmm0, %xmm2
 # CHECK-NEXT:  -      -      -     2.00   2.00   1.00   1.00   2.00    -      -      -      -      -      -     vmaskmovpd  (%rax), %ymm0, %ymm2

 # CHECK-NEXT:  -      -      -      -      -     0.50   0.50    -      -      -      -     0.50   0.50    -     vpcmpgtb  %xmm0, %xmm1, %xmm2
 # CHECK-NEXT:  -      -      -      -      -     0.50   0.50   1.00    -      -      -     0.50   0.50    -     vpcmpgtb  (%rax), %xmm1, %xmm2
 # CHECK-NEXT:  -      -      -      -      -     0.50   0.50    -      -      -      -     0.50   0.50    -     vpcmpgtd  %xmm0, %xmm1, %xmm2
 # CHECK-NEXT:  -      -      -      -      -     0.50   0.50   1.00    -      -      -     0.50   0.50    -     vpcmpgtd  (%rax), %xmm1, %xmm2
 # CHECK-NEXT:  -      -      -      -      -     0.50   0.50    -      -      -      -     0.50   0.50    -     vpcmpgtq  %xmm0, %xmm1, %xmm2
 # CHECK-NEXT:  -      -      -      -      -     0.50   0.50   1.00    -      -      -     0.50   0.50    -     vpcmpgtq  (%rax), %xmm1, %xmm2
 # CHECK-NEXT:  -      -      -      -      -     0.50   0.50    -      -      -      -     0.50   0.50    -     vpcmpgtw  %xmm0, %xmm1, %xmm2
 # CHECK-NEXT:  -      -      -      -      -     0.50   0.50   1.00    -      -      -     0.50   0.50    -     vpcmpgtw  (%rax), %xmm1, %xmm2
-# CHECK-NEXT:  -      -      -     0.50   0.50   0.50   0.50    -      -      -      -      -      -      -     vperm2f128  $1, %ymm0, %ymm1, %ymm2
-# CHECK-NEXT:  -      -      -     0.50   0.50   0.50   0.50   1.00    -      -      -      -      -      -     vperm2f128  $1, (%rax), %ymm1, %ymm2
+# CHECK-NEXT:  -      -      -     1.00   1.00   1.00   1.00    -      -      -      -      -      -      -     vperm2f128  $1, %ymm0, %ymm1, %ymm2
+# CHECK-NEXT:  -      -      -     1.00   1.00   1.00   1.00   1.00    -      -      -      -      -      -     vperm2f128  $1, (%rax), %ymm1, %ymm2
 # CHECK-NEXT:  -      -      -     0.50   0.50   0.50   0.50    -      -      -      -      -      -      -     vpermilpd $1, %xmm0, %xmm2
 # CHECK-NEXT:  -      -      -     0.50   0.50   0.50   0.50   1.00    -      -      -      -      -      -     vpermilpd $1, (%rax), %xmm2
 # CHECK-NEXT:  -      -      -     2.00   2.00   0.50   0.50    -      -      -      -      -      -      -     vpermilpd %xmm0, %xmm1, %xmm2
 # CHECK-NEXT:  -      -      -     2.00   2.00   0.50   0.50   1.00    -      -      -      -      -      -     vpermilpd (%rax), %xmm1, %xmm2
 # CHECK-NEXT:  -      -      -     1.00   1.00   1.00   1.00    -      -      -      -      -      -      -     vpermilpd $1, %ymm0, %ymm2
 # CHECK-NEXT:  -      -      -     1.00   1.00   1.00   1.00   2.00    -      -      -      -      -      -     vpermilpd $1, (%rax), %ymm2
 # CHECK-NEXT:  -      -      -     3.00   3.00   1.00   1.00    -      -      -      -      -      -      -     vpermilpd %ymm0, %ymm1, %ymm2
 # CHECK-NEXT:  -      -      -     3.00   3.00   1.00   1.00   2.00    -      -      -      -      -      -     vpermilpd (%rax), %ymm1, %ymm2