[X86] Use WriteVecMove scheduler classes for VPMOVM2* instructions
authorSimon Pilgrim <llvm-dev@redking.me.uk>
Mon, 27 Dec 2021 13:21:20 +0000 (13:21 +0000)
committerSimon Pilgrim <llvm-dev@redking.me.uk>
Mon, 27 Dec 2021 13:21:29 +0000 (13:21 +0000)
These match the port behaviour of reg-reg predicated xmm/ymm/zmm moves

Fixes #34958

13 files changed:
llvm/lib/Target/X86/X86InstrAVX512.td
llvm/test/tools/llvm-mca/X86/Generic/resources-avx512bw.s
llvm/test/tools/llvm-mca/X86/Generic/resources-avx512bwvl.s
llvm/test/tools/llvm-mca/X86/Generic/resources-avx512dq.s
llvm/test/tools/llvm-mca/X86/Generic/resources-avx512dqvl.s
llvm/test/tools/llvm-mca/X86/IceLakeServer/resources-avx512bw.s
llvm/test/tools/llvm-mca/X86/IceLakeServer/resources-avx512bwvl.s
llvm/test/tools/llvm-mca/X86/IceLakeServer/resources-avx512dq.s
llvm/test/tools/llvm-mca/X86/IceLakeServer/resources-avx512dqvl.s
llvm/test/tools/llvm-mca/X86/SkylakeServer/resources-avx512bw.s
llvm/test/tools/llvm-mca/X86/SkylakeServer/resources-avx512bwvl.s
llvm/test/tools/llvm-mca/X86/SkylakeServer/resources-avx512dq.s
llvm/test/tools/llvm-mca/X86/SkylakeServer/resources-avx512dqvl.s

index ecd4777..bc67d1f 100644 (file)
@@ -10537,13 +10537,12 @@ def rr : AVX512XS8I<opc, MRMSrcReg, (outs Vec.RC:$dst), (ins Vec.KRC:$src),
 
 multiclass cvt_mask_by_elt_width<bits<8> opc, AVX512VLVectorVTInfo VTInfo,
                                  string OpcodeStr, Predicate prd> {
-// TODO - Replace WriteMove with WriteVecTrunc?
 let Predicates = [prd] in
-  defm Z : cvt_by_vec_width<opc, VTInfo.info512, OpcodeStr, WriteMove>, EVEX_V512;
+  defm Z : cvt_by_vec_width<opc, VTInfo.info512, OpcodeStr, WriteVecMoveZ>, EVEX_V512;
 
   let Predicates = [prd, HasVLX] in {
-    defm Z256 : cvt_by_vec_width<opc, VTInfo.info256, OpcodeStr, WriteMove>, EVEX_V256;
-    defm Z128 : cvt_by_vec_width<opc, VTInfo.info128, OpcodeStr, WriteMove>, EVEX_V128;
+    defm Z256 : cvt_by_vec_width<opc, VTInfo.info256, OpcodeStr, WriteVecMoveY>, EVEX_V256;
+    defm Z128 : cvt_by_vec_width<opc, VTInfo.info128, OpcodeStr, WriteVecMoveX>, EVEX_V128;
   }
 }
 
index da27526..7301694 100644 (file)
@@ -504,8 +504,8 @@ vpmovw2m          %zmm0, %k0
 # CHECK-NEXT:  2      8     1.00    *                   vpunpcklwd     (%rax), %zmm17, %zmm19 {%k1}
 # CHECK-NEXT:  1      1     1.00                        vpunpcklwd     %zmm16, %zmm17, %zmm19 {%k1} {z}
 # CHECK-NEXT:  2      8     1.00    *                   vpunpcklwd     (%rax), %zmm17, %zmm19 {%k1} {z}
-# CHECK-NEXT:  1      1     0.33                        vpmovm2b       %k0, %zmm0
-# CHECK-NEXT:  1      1     0.33                        vpmovm2w       %k0, %zmm0
+# CHECK-NEXT:  1      1     0.50                        vpmovm2b       %k0, %zmm0
+# CHECK-NEXT:  1      1     0.50                        vpmovm2w       %k0, %zmm0
 # CHECK-NEXT:  1      1     0.33                        vpmovb2m       %zmm0, %k0
 # CHECK-NEXT:  1      1     0.33                        vpmovw2m       %zmm0, %k0
 
@@ -521,7 +521,7 @@ vpmovw2m          %zmm0, %k0
 
 # CHECK:      Resource pressure per iteration:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6.0]  [6.1]
-# CHECK-NEXT:  -      -     13.67  47.67  4.00   156.67 52.50  52.50
+# CHECK-NEXT:  -      -     14.00  47.00  4.00   157.00 52.50  52.50
 
 # CHECK:      Resource pressure by instruction:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6.0]  [6.1]  Instructions:
@@ -749,7 +749,7 @@ vpmovw2m          %zmm0, %k0
 # CHECK-NEXT:  -      -      -      -      -     1.00   0.50   0.50   vpunpcklwd       (%rax), %zmm17, %zmm19 {%k1}
 # CHECK-NEXT:  -      -      -      -      -     1.00    -      -     vpunpcklwd       %zmm16, %zmm17, %zmm19 {%k1} {z}
 # CHECK-NEXT:  -      -      -      -      -     1.00   0.50   0.50   vpunpcklwd       (%rax), %zmm17, %zmm19 {%k1} {z}
-# CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     vpmovm2b %k0, %zmm0
-# CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     vpmovm2w %k0, %zmm0
+# CHECK-NEXT:  -      -     0.50    -      -     0.50    -      -     vpmovm2b %k0, %zmm0
+# CHECK-NEXT:  -      -     0.50    -      -     0.50    -      -     vpmovm2w %k0, %zmm0
 # CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     vpmovb2m %zmm0, %k0
 # CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     vpmovw2m %zmm0, %k0
index ad32d07..32c2856 100644 (file)
@@ -926,8 +926,8 @@ vpmovw2m          %ymm0, %k0
 # CHECK-NEXT:  2      8     1.00    *                   vpunpcklwd     (%rax), %ymm17, %ymm19 {%k1} {z}
 # CHECK-NEXT:  1      1     0.33                        vpmovm2b       %k0, %xmm0
 # CHECK-NEXT:  1      1     0.33                        vpmovm2w       %k0, %xmm0
-# CHECK-NEXT:  1      1     0.33                        vpmovm2b       %k0, %ymm0
-# CHECK-NEXT:  1      1     0.33                        vpmovm2w       %k0, %ymm0
+# CHECK-NEXT:  1      1     0.50                        vpmovm2b       %k0, %ymm0
+# CHECK-NEXT:  1      1     0.50                        vpmovm2w       %k0, %ymm0
 # CHECK-NEXT:  1      1     0.33                        vpmovb2m       %xmm0, %k0
 # CHECK-NEXT:  1      1     0.33                        vpmovw2m       %xmm0, %k0
 # CHECK-NEXT:  1      1     0.33                        vpmovb2m       %ymm0, %k0
@@ -945,7 +945,7 @@ vpmovw2m          %ymm0, %k0
 
 # CHECK:      Resource pressure per iteration:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6.0]  [6.1]
-# CHECK-NEXT:  -      -     18.33  133.33 8.00   248.33 105.00 105.00
+# CHECK-NEXT:  -      -     18.67  132.67 8.00   248.67 105.00 105.00
 
 # CHECK:      Resource pressure by instruction:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6.0]  [6.1]  Instructions:
@@ -1363,8 +1363,8 @@ vpmovw2m          %ymm0, %k0
 # CHECK-NEXT:  -      -      -      -      -     1.00   0.50   0.50   vpunpcklwd       (%rax), %ymm17, %ymm19 {%k1} {z}
 # CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     vpmovm2b %k0, %xmm0
 # CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     vpmovm2w %k0, %xmm0
-# CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     vpmovm2b %k0, %ymm0
-# CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     vpmovm2w %k0, %ymm0
+# CHECK-NEXT:  -      -     0.50    -      -     0.50    -      -     vpmovm2b %k0, %ymm0
+# CHECK-NEXT:  -      -     0.50    -      -     0.50    -      -     vpmovm2w %k0, %ymm0
 # CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     vpmovb2m %xmm0, %k0
 # CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     vpmovw2m %xmm0, %k0
 # CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     vpmovb2m %ymm0, %k0
index 9d08164..5849220 100644 (file)
@@ -336,8 +336,8 @@ vpmovq2m          %zmm0, %k0
 # CHECK-NEXT:  1      1     1.00                        vxorps %zmm16, %zmm17, %zmm19 {%k1} {z}
 # CHECK-NEXT:  2      8     1.00    *                   vxorps (%rax), %zmm17, %zmm19 {%k1} {z}
 # CHECK-NEXT:  2      8     1.00    *                   vxorps (%rax){1to16}, %zmm17, %zmm19 {%k1} {z}
-# CHECK-NEXT:  1      1     0.33                        vpmovm2d       %k0, %zmm0
-# CHECK-NEXT:  1      1     0.33                        vpmovm2q       %k0, %zmm0
+# CHECK-NEXT:  1      1     0.50                        vpmovm2d       %k0, %zmm0
+# CHECK-NEXT:  1      1     0.50                        vpmovm2q       %k0, %zmm0
 # CHECK-NEXT:  1      1     0.33                        vpmovd2m       %zmm0, %k0
 # CHECK-NEXT:  1      1     0.33                        vpmovq2m       %zmm0, %k0
 
@@ -353,7 +353,7 @@ vpmovq2m          %zmm0, %k0
 
 # CHECK:      Resource pressure per iteration:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6.0]  [6.1]
-# CHECK-NEXT:  -      -     9.67   41.67   -     110.67 46.50  46.50
+# CHECK-NEXT:  -      -     10.00  41.00   -     111.00 46.50  46.50
 
 # CHECK:      Resource pressure by instruction:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6.0]  [6.1]  Instructions:
@@ -506,7 +506,7 @@ vpmovq2m          %zmm0, %k0
 # CHECK-NEXT:  -      -      -      -      -     1.00    -      -     vxorps   %zmm16, %zmm17, %zmm19 {%k1} {z}
 # CHECK-NEXT:  -      -      -      -      -     1.00   0.50   0.50   vxorps   (%rax), %zmm17, %zmm19 {%k1} {z}
 # CHECK-NEXT:  -      -      -      -      -     1.00   0.50   0.50   vxorps   (%rax){1to16}, %zmm17, %zmm19 {%k1} {z}
-# CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     vpmovm2d %k0, %zmm0
-# CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     vpmovm2q %k0, %zmm0
+# CHECK-NEXT:  -      -     0.50    -      -     0.50    -      -     vpmovm2d %k0, %zmm0
+# CHECK-NEXT:  -      -     0.50    -      -     0.50    -      -     vpmovm2q %k0, %zmm0
 # CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     vpmovd2m %zmm0, %k0
 # CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     vpmovq2m %zmm0, %k0
index 7ab33c3..d7d0c3a 100644 (file)
@@ -459,8 +459,8 @@ vpmovq2m          %ymm0, %k0
 # CHECK-NEXT:  2      8     1.00    *                   vxorps (%rax){1to8}, %ymm17, %ymm19 {%k1} {z}
 # CHECK-NEXT:  1      1     0.33                        vpmovm2d       %k0, %xmm0
 # CHECK-NEXT:  1      1     0.33                        vpmovm2q       %k0, %xmm0
-# CHECK-NEXT:  1      1     0.33                        vpmovm2d       %k0, %ymm0
-# CHECK-NEXT:  1      1     0.33                        vpmovm2q       %k0, %ymm0
+# CHECK-NEXT:  1      1     0.50                        vpmovm2d       %k0, %ymm0
+# CHECK-NEXT:  1      1     0.50                        vpmovm2q       %k0, %ymm0
 # CHECK-NEXT:  1      1     0.33                        vpmovd2m       %xmm0, %k0
 # CHECK-NEXT:  1      1     0.33                        vpmovq2m       %xmm0, %k0
 # CHECK-NEXT:  1      1     0.33                        vpmovd2m       %ymm0, %k0
@@ -478,7 +478,7 @@ vpmovq2m          %ymm0, %k0
 
 # CHECK:      Resource pressure per iteration:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6.0]  [6.1]
-# CHECK-NEXT:  -      -     14.67  46.17   -     169.17 66.50  66.50
+# CHECK-NEXT:  -      -     15.00  45.50   -     169.50 66.50  66.50
 
 # CHECK:      Resource pressure by instruction:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6.0]  [6.1]  Instructions:
@@ -688,8 +688,8 @@ vpmovq2m          %ymm0, %k0
 # CHECK-NEXT:  -      -      -      -      -     1.00   0.50   0.50   vxorps   (%rax){1to8}, %ymm17, %ymm19 {%k1} {z}
 # CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     vpmovm2d %k0, %xmm0
 # CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     vpmovm2q %k0, %xmm0
-# CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     vpmovm2d %k0, %ymm0
-# CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     vpmovm2q %k0, %ymm0
+# CHECK-NEXT:  -      -     0.50    -      -     0.50    -      -     vpmovm2d %k0, %ymm0
+# CHECK-NEXT:  -      -     0.50    -      -     0.50    -      -     vpmovm2q %k0, %ymm0
 # CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     vpmovd2m %xmm0, %k0
 # CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     vpmovq2m %xmm0, %k0
 # CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     vpmovd2m %ymm0, %k0
index 9a41974..02a5152 100644 (file)
@@ -504,8 +504,8 @@ vpmovw2m          %zmm0, %k0
 # CHECK-NEXT:  2      8     1.00    *                   vpunpcklwd     (%rax), %zmm17, %zmm19 {%k1}
 # CHECK-NEXT:  1      1     1.00                        vpunpcklwd     %zmm16, %zmm17, %zmm19 {%k1} {z}
 # CHECK-NEXT:  2      8     1.00    *                   vpunpcklwd     (%rax), %zmm17, %zmm19 {%k1} {z}
-# CHECK-NEXT:  1      1     0.25                        vpmovm2b       %k0, %zmm0
-# CHECK-NEXT:  1      1     0.25                        vpmovm2w       %k0, %zmm0
+# CHECK-NEXT:  1      1     0.50                        vpmovm2b       %k0, %zmm0
+# CHECK-NEXT:  1      1     0.50                        vpmovm2w       %k0, %zmm0
 # CHECK-NEXT:  1      1     1.00                        vpmovb2m       %zmm0, %k0
 # CHECK-NEXT:  1      1     1.00                        vpmovw2m       %zmm0, %k0
 
@@ -525,7 +525,7 @@ vpmovw2m          %zmm0, %k0
 
 # CHECK:      Resource pressure per iteration:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]    [10]   [11]
-# CHECK-NEXT:  -      -     37.50  11.50  52.50  52.50  6.00   182.50 0.50   2.00    -      -
+# CHECK-NEXT:  -      -     38.00  11.00  52.50  52.50  6.00   183.00  -     2.00    -      -
 
 # CHECK:      Resource pressure by instruction:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]    [10]   [11]   Instructions:
@@ -753,7 +753,7 @@ vpmovw2m          %zmm0, %k0
 # CHECK-NEXT:  -      -      -      -     0.50   0.50    -     1.00    -      -      -      -     vpunpcklwd   (%rax), %zmm17, %zmm19 {%k1}
 # CHECK-NEXT:  -      -      -      -      -      -      -     1.00    -      -      -      -     vpunpcklwd   %zmm16, %zmm17, %zmm19 {%k1} {z}
 # CHECK-NEXT:  -      -      -      -     0.50   0.50    -     1.00    -      -      -      -     vpunpcklwd   (%rax), %zmm17, %zmm19 {%k1} {z}
-# CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -      -      -     vpmovm2b     %k0, %zmm0
-# CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -      -      -     vpmovm2w     %k0, %zmm0
+# CHECK-NEXT:  -      -     0.50    -      -      -      -     0.50    -      -      -      -     vpmovm2b     %k0, %zmm0
+# CHECK-NEXT:  -      -     0.50    -      -      -      -     0.50    -      -      -      -     vpmovm2w     %k0, %zmm0
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -      -      -     vpmovb2m     %zmm0, %k0
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -      -      -     vpmovw2m     %zmm0, %k0
index 0539e35..6e106c9 100644 (file)
@@ -924,10 +924,10 @@ vpmovw2m          %ymm0, %k0
 # CHECK-NEXT:  2      8     0.50    *                   vpunpcklwd     (%rax), %ymm17, %ymm19 {%k1}
 # CHECK-NEXT:  1      1     0.50                        vpunpcklwd     %ymm16, %ymm17, %ymm19 {%k1} {z}
 # CHECK-NEXT:  2      8     0.50    *                   vpunpcklwd     (%rax), %ymm17, %ymm19 {%k1} {z}
-# CHECK-NEXT:  1      1     0.25                        vpmovm2b       %k0, %xmm0
-# CHECK-NEXT:  1      1     0.25                        vpmovm2w       %k0, %xmm0
-# CHECK-NEXT:  1      1     0.25                        vpmovm2b       %k0, %ymm0
-# CHECK-NEXT:  1      1     0.25                        vpmovm2w       %k0, %ymm0
+# CHECK-NEXT:  1      1     0.33                        vpmovm2b       %k0, %xmm0
+# CHECK-NEXT:  1      1     0.33                        vpmovm2w       %k0, %xmm0
+# CHECK-NEXT:  1      1     0.33                        vpmovm2b       %k0, %ymm0
+# CHECK-NEXT:  1      1     0.33                        vpmovm2w       %k0, %ymm0
 # CHECK-NEXT:  1      1     1.00                        vpmovb2m       %xmm0, %k0
 # CHECK-NEXT:  1      1     1.00                        vpmovw2m       %xmm0, %k0
 # CHECK-NEXT:  1      1     1.00                        vpmovb2m       %ymm0, %k0
@@ -949,7 +949,7 @@ vpmovw2m          %ymm0, %k0
 
 # CHECK:      Resource pressure per iteration:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]    [10]   [11]
-# CHECK-NEXT:  -      -     41.00  89.00  103.67 103.67 8.00   293.00 1.00   2.67    -      -
+# CHECK-NEXT:  -      -     41.33  89.33  103.67 103.67 8.00   293.33  -     2.67    -      -
 
 # CHECK:      Resource pressure by instruction:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]    [10]   [11]   Instructions:
@@ -1365,10 +1365,10 @@ vpmovw2m          %ymm0, %k0
 # CHECK-NEXT:  -      -      -     0.50   0.50   0.50    -     0.50    -      -      -      -     vpunpcklwd   (%rax), %ymm17, %ymm19 {%k1}
 # CHECK-NEXT:  -      -      -     0.50    -      -      -     0.50    -      -      -      -     vpunpcklwd   %ymm16, %ymm17, %ymm19 {%k1} {z}
 # CHECK-NEXT:  -      -      -     0.50   0.50   0.50    -     0.50    -      -      -      -     vpunpcklwd   (%rax), %ymm17, %ymm19 {%k1} {z}
-# CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -      -      -     vpmovm2b     %k0, %xmm0
-# CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -      -      -     vpmovm2w     %k0, %xmm0
-# CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -      -      -     vpmovm2b     %k0, %ymm0
-# CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -      -      -     vpmovm2w     %k0, %ymm0
+# CHECK-NEXT:  -      -     0.33   0.33    -      -      -     0.33    -      -      -      -     vpmovm2b     %k0, %xmm0
+# CHECK-NEXT:  -      -     0.33   0.33    -      -      -     0.33    -      -      -      -     vpmovm2w     %k0, %xmm0
+# CHECK-NEXT:  -      -     0.33   0.33    -      -      -     0.33    -      -      -      -     vpmovm2b     %k0, %ymm0
+# CHECK-NEXT:  -      -     0.33   0.33    -      -      -     0.33    -      -      -      -     vpmovm2w     %k0, %ymm0
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -      -      -     vpmovb2m     %xmm0, %k0
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -      -      -     vpmovw2m     %xmm0, %k0
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -      -      -     vpmovb2m     %ymm0, %k0
index 102602b..b2b9a93 100644 (file)
@@ -336,8 +336,8 @@ vpmovq2m          %zmm0, %k0
 # CHECK-NEXT:  1      1     0.50                        vxorps %zmm16, %zmm17, %zmm19 {%k1} {z}
 # CHECK-NEXT:  2      8     0.50    *                   vxorps (%rax), %zmm17, %zmm19 {%k1} {z}
 # CHECK-NEXT:  2      8     0.50    *                   vxorps (%rax){1to16}, %zmm17, %zmm19 {%k1} {z}
-# CHECK-NEXT:  1      1     0.25                        vpmovm2d       %k0, %zmm0
-# CHECK-NEXT:  1      1     0.25                        vpmovm2q       %k0, %zmm0
+# CHECK-NEXT:  1      1     0.50                        vpmovm2d       %k0, %zmm0
+# CHECK-NEXT:  1      1     0.50                        vpmovm2q       %k0, %zmm0
 # CHECK-NEXT:  1      1     1.00                        vpmovd2m       %zmm0, %k0
 # CHECK-NEXT:  1      1     1.00                        vpmovq2m       %zmm0, %k0
 
@@ -357,7 +357,7 @@ vpmovq2m          %zmm0, %k0
 
 # CHECK:      Resource pressure per iteration:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]    [10]   [11]
-# CHECK-NEXT:  -      -     65.50  10.50  46.50  46.50   -     95.50  0.50    -      -      -
+# CHECK-NEXT:  -      -     66.00  10.00  46.50  46.50   -     96.00   -      -      -      -
 
 # CHECK:      Resource pressure by instruction:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]    [10]   [11]   Instructions:
@@ -510,7 +510,7 @@ vpmovq2m          %zmm0, %k0
 # CHECK-NEXT:  -      -     0.50    -      -      -      -     0.50    -      -      -      -     vxorps       %zmm16, %zmm17, %zmm19 {%k1} {z}
 # CHECK-NEXT:  -      -     0.50    -     0.50   0.50    -     0.50    -      -      -      -     vxorps       (%rax), %zmm17, %zmm19 {%k1} {z}
 # CHECK-NEXT:  -      -     0.50    -     0.50   0.50    -     0.50    -      -      -      -     vxorps       (%rax){1to16}, %zmm17, %zmm19 {%k1} {z}
-# CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -      -      -     vpmovm2d     %k0, %zmm0
-# CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -      -      -     vpmovm2q     %k0, %zmm0
+# CHECK-NEXT:  -      -     0.50    -      -      -      -     0.50    -      -      -      -     vpmovm2d     %k0, %zmm0
+# CHECK-NEXT:  -      -     0.50    -      -      -      -     0.50    -      -      -      -     vpmovm2q     %k0, %zmm0
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -      -      -     vpmovd2m     %zmm0, %k0
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -      -      -     vpmovq2m     %zmm0, %k0
index 647a5cf..9c7e93f 100644 (file)
@@ -457,10 +457,10 @@ vpmovq2m          %ymm0, %k0
 # CHECK-NEXT:  1      1     0.33                        vxorps %ymm16, %ymm17, %ymm19 {%k1} {z}
 # CHECK-NEXT:  2      8     0.50    *                   vxorps (%rax), %ymm17, %ymm19 {%k1} {z}
 # CHECK-NEXT:  2      8     0.50    *                   vxorps (%rax){1to8}, %ymm17, %ymm19 {%k1} {z}
-# CHECK-NEXT:  1      1     0.25                        vpmovm2d       %k0, %xmm0
-# CHECK-NEXT:  1      1     0.25                        vpmovm2q       %k0, %xmm0
-# CHECK-NEXT:  1      1     0.25                        vpmovm2d       %k0, %ymm0
-# CHECK-NEXT:  1      1     0.25                        vpmovm2q       %k0, %ymm0
+# CHECK-NEXT:  1      1     0.33                        vpmovm2d       %k0, %xmm0
+# CHECK-NEXT:  1      1     0.33                        vpmovm2q       %k0, %xmm0
+# CHECK-NEXT:  1      1     0.33                        vpmovm2d       %k0, %ymm0
+# CHECK-NEXT:  1      1     0.33                        vpmovm2q       %k0, %ymm0
 # CHECK-NEXT:  1      1     1.00                        vpmovd2m       %xmm0, %k0
 # CHECK-NEXT:  1      1     1.00                        vpmovq2m       %xmm0, %k0
 # CHECK-NEXT:  1      1     1.00                        vpmovd2m       %ymm0, %k0
@@ -482,7 +482,7 @@ vpmovq2m          %ymm0, %k0
 
 # CHECK:      Resource pressure per iteration:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]    [10]   [11]
-# CHECK-NEXT:  -      -     77.00  73.00  66.50  66.50   -     85.00  1.00    -      -      -
+# CHECK-NEXT:  -      -     77.33  73.33  66.50  66.50   -     85.33   -      -      -      -
 
 # CHECK:      Resource pressure by instruction:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]    [10]   [11]   Instructions:
@@ -690,10 +690,10 @@ vpmovq2m          %ymm0, %k0
 # CHECK-NEXT:  -      -     0.33   0.33    -      -      -     0.33    -      -      -      -     vxorps       %ymm16, %ymm17, %ymm19 {%k1} {z}
 # CHECK-NEXT:  -      -     0.33   0.33   0.50   0.50    -     0.33    -      -      -      -     vxorps       (%rax), %ymm17, %ymm19 {%k1} {z}
 # CHECK-NEXT:  -      -     0.33   0.33   0.50   0.50    -     0.33    -      -      -      -     vxorps       (%rax){1to8}, %ymm17, %ymm19 {%k1} {z}
-# CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -      -      -     vpmovm2d     %k0, %xmm0
-# CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -      -      -     vpmovm2q     %k0, %xmm0
-# CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -      -      -     vpmovm2d     %k0, %ymm0
-# CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -      -      -     vpmovm2q     %k0, %ymm0
+# CHECK-NEXT:  -      -     0.33   0.33    -      -      -     0.33    -      -      -      -     vpmovm2d     %k0, %xmm0
+# CHECK-NEXT:  -      -     0.33   0.33    -      -      -     0.33    -      -      -      -     vpmovm2q     %k0, %xmm0
+# CHECK-NEXT:  -      -     0.33   0.33    -      -      -     0.33    -      -      -      -     vpmovm2d     %k0, %ymm0
+# CHECK-NEXT:  -      -     0.33   0.33    -      -      -     0.33    -      -      -      -     vpmovm2q     %k0, %ymm0
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -      -      -     vpmovd2m     %xmm0, %k0
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -      -      -     vpmovq2m     %xmm0, %k0
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -      -      -     vpmovd2m     %ymm0, %k0
index a789988..fed721e 100644 (file)
@@ -504,8 +504,8 @@ vpmovw2m          %zmm0, %k0
 # CHECK-NEXT:  2      8     1.00    *                   vpunpcklwd     (%rax), %zmm17, %zmm19 {%k1}
 # CHECK-NEXT:  1      1     1.00                        vpunpcklwd     %zmm16, %zmm17, %zmm19 {%k1} {z}
 # CHECK-NEXT:  2      8     1.00    *                   vpunpcklwd     (%rax), %zmm17, %zmm19 {%k1} {z}
-# CHECK-NEXT:  1      1     0.25                        vpmovm2b       %k0, %zmm0
-# CHECK-NEXT:  1      1     0.25                        vpmovm2w       %k0, %zmm0
+# CHECK-NEXT:  1      1     0.50                        vpmovm2b       %k0, %zmm0
+# CHECK-NEXT:  1      1     0.50                        vpmovm2w       %k0, %zmm0
 # CHECK-NEXT:  1      1     1.00                        vpmovb2m       %zmm0, %k0
 # CHECK-NEXT:  1      1     1.00                        vpmovw2m       %zmm0, %k0
 
@@ -523,7 +523,7 @@ vpmovw2m          %zmm0, %k0
 
 # CHECK:      Resource pressure per iteration:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]
-# CHECK-NEXT:  -      -     37.50  11.50  52.50  52.50  6.00   182.50 0.50   2.00
+# CHECK-NEXT:  -      -     38.00  11.00  52.50  52.50  6.00   183.00  -     2.00
 
 # CHECK:      Resource pressure by instruction:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]    Instructions:
@@ -751,7 +751,7 @@ vpmovw2m          %zmm0, %k0
 # CHECK-NEXT:  -      -      -      -     0.50   0.50    -     1.00    -      -     vpunpcklwd (%rax), %zmm17, %zmm19 {%k1}
 # CHECK-NEXT:  -      -      -      -      -      -      -     1.00    -      -     vpunpcklwd %zmm16, %zmm17, %zmm19 {%k1} {z}
 # CHECK-NEXT:  -      -      -      -     0.50   0.50    -     1.00    -      -     vpunpcklwd (%rax), %zmm17, %zmm19 {%k1} {z}
-# CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     vpmovm2b   %k0, %zmm0
-# CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     vpmovm2w   %k0, %zmm0
+# CHECK-NEXT:  -      -     0.50    -      -      -      -     0.50    -      -     vpmovm2b   %k0, %zmm0
+# CHECK-NEXT:  -      -     0.50    -      -      -      -     0.50    -      -     vpmovm2w   %k0, %zmm0
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -     vpmovb2m   %zmm0, %k0
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -     vpmovw2m   %zmm0, %k0
index 2744d28..cbdeb14 100644 (file)
@@ -924,10 +924,10 @@ vpmovw2m          %ymm0, %k0
 # CHECK-NEXT:  2      8     1.00    *                   vpunpcklwd     (%rax), %ymm17, %ymm19 {%k1}
 # CHECK-NEXT:  1      1     1.00                        vpunpcklwd     %ymm16, %ymm17, %ymm19 {%k1} {z}
 # CHECK-NEXT:  2      8     1.00    *                   vpunpcklwd     (%rax), %ymm17, %ymm19 {%k1} {z}
-# CHECK-NEXT:  1      1     0.25                        vpmovm2b       %k0, %xmm0
-# CHECK-NEXT:  1      1     0.25                        vpmovm2w       %k0, %xmm0
-# CHECK-NEXT:  1      1     0.25                        vpmovm2b       %k0, %ymm0
-# CHECK-NEXT:  1      1     0.25                        vpmovm2w       %k0, %ymm0
+# CHECK-NEXT:  1      1     0.33                        vpmovm2b       %k0, %xmm0
+# CHECK-NEXT:  1      1     0.33                        vpmovm2w       %k0, %xmm0
+# CHECK-NEXT:  1      1     0.33                        vpmovm2b       %k0, %ymm0
+# CHECK-NEXT:  1      1     0.33                        vpmovm2w       %k0, %ymm0
 # CHECK-NEXT:  1      1     1.00                        vpmovb2m       %xmm0, %k0
 # CHECK-NEXT:  1      1     1.00                        vpmovw2m       %xmm0, %k0
 # CHECK-NEXT:  1      1     1.00                        vpmovb2m       %ymm0, %k0
@@ -947,7 +947,7 @@ vpmovw2m          %ymm0, %k0
 
 # CHECK:      Resource pressure per iteration:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]
-# CHECK-NEXT:  -      -     41.00  37.00  103.67 103.67 8.00   345.00 1.00   2.67
+# CHECK-NEXT:  -      -     41.33  37.33  103.67 103.67 8.00   345.33  -     2.67
 
 # CHECK:      Resource pressure by instruction:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]    Instructions:
@@ -1363,10 +1363,10 @@ vpmovw2m          %ymm0, %k0
 # CHECK-NEXT:  -      -      -      -     0.50   0.50    -     1.00    -      -     vpunpcklwd (%rax), %ymm17, %ymm19 {%k1}
 # CHECK-NEXT:  -      -      -      -      -      -      -     1.00    -      -     vpunpcklwd %ymm16, %ymm17, %ymm19 {%k1} {z}
 # CHECK-NEXT:  -      -      -      -     0.50   0.50    -     1.00    -      -     vpunpcklwd (%rax), %ymm17, %ymm19 {%k1} {z}
-# CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     vpmovm2b   %k0, %xmm0
-# CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     vpmovm2w   %k0, %xmm0
-# CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     vpmovm2b   %k0, %ymm0
-# CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     vpmovm2w   %k0, %ymm0
+# CHECK-NEXT:  -      -     0.33   0.33    -      -      -     0.33    -      -     vpmovm2b   %k0, %xmm0
+# CHECK-NEXT:  -      -     0.33   0.33    -      -      -     0.33    -      -     vpmovm2w   %k0, %xmm0
+# CHECK-NEXT:  -      -     0.33   0.33    -      -      -     0.33    -      -     vpmovm2b   %k0, %ymm0
+# CHECK-NEXT:  -      -     0.33   0.33    -      -      -     0.33    -      -     vpmovm2w   %k0, %ymm0
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -     vpmovb2m   %xmm0, %k0
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -     vpmovw2m   %xmm0, %k0
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -     vpmovb2m   %ymm0, %k0
index 18750b0..046bc8e 100644 (file)
@@ -336,8 +336,8 @@ vpmovq2m          %zmm0, %k0
 # CHECK-NEXT:  1      1     0.50                        vxorps %zmm16, %zmm17, %zmm19 {%k1} {z}
 # CHECK-NEXT:  2      8     0.50    *                   vxorps (%rax), %zmm17, %zmm19 {%k1} {z}
 # CHECK-NEXT:  2      8     0.50    *                   vxorps (%rax){1to16}, %zmm17, %zmm19 {%k1} {z}
-# CHECK-NEXT:  1      1     0.25                        vpmovm2d       %k0, %zmm0
-# CHECK-NEXT:  1      1     0.25                        vpmovm2q       %k0, %zmm0
+# CHECK-NEXT:  1      1     0.50                        vpmovm2d       %k0, %zmm0
+# CHECK-NEXT:  1      1     0.50                        vpmovm2q       %k0, %zmm0
 # CHECK-NEXT:  1      1     1.00                        vpmovd2m       %zmm0, %k0
 # CHECK-NEXT:  1      1     1.00                        vpmovq2m       %zmm0, %k0
 
@@ -355,7 +355,7 @@ vpmovq2m          %zmm0, %k0
 
 # CHECK:      Resource pressure per iteration:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]
-# CHECK-NEXT:  -      -     65.50  10.50  46.50  46.50   -     95.50  0.50    -
+# CHECK-NEXT:  -      -     66.00  10.00  46.50  46.50   -     96.00   -      -
 
 # CHECK:      Resource pressure by instruction:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]    Instructions:
@@ -508,7 +508,7 @@ vpmovq2m          %zmm0, %k0
 # CHECK-NEXT:  -      -     0.50    -      -      -      -     0.50    -      -     vxorps     %zmm16, %zmm17, %zmm19 {%k1} {z}
 # CHECK-NEXT:  -      -     0.50    -     0.50   0.50    -     0.50    -      -     vxorps     (%rax), %zmm17, %zmm19 {%k1} {z}
 # CHECK-NEXT:  -      -     0.50    -     0.50   0.50    -     0.50    -      -     vxorps     (%rax){1to16}, %zmm17, %zmm19 {%k1} {z}
-# CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     vpmovm2d   %k0, %zmm0
-# CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     vpmovm2q   %k0, %zmm0
+# CHECK-NEXT:  -      -     0.50    -      -      -      -     0.50    -      -     vpmovm2d   %k0, %zmm0
+# CHECK-NEXT:  -      -     0.50    -      -      -      -     0.50    -      -     vpmovm2q   %k0, %zmm0
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -     vpmovd2m   %zmm0, %k0
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -     vpmovq2m   %zmm0, %k0
index 4f2b0d9..6233999 100644 (file)
@@ -457,10 +457,10 @@ vpmovq2m          %ymm0, %k0
 # CHECK-NEXT:  1      1     0.33                        vxorps %ymm16, %ymm17, %ymm19 {%k1} {z}
 # CHECK-NEXT:  2      8     0.50    *                   vxorps (%rax), %ymm17, %ymm19 {%k1} {z}
 # CHECK-NEXT:  2      8     0.50    *                   vxorps (%rax){1to8}, %ymm17, %ymm19 {%k1} {z}
-# CHECK-NEXT:  1      1     0.25                        vpmovm2d       %k0, %xmm0
-# CHECK-NEXT:  1      1     0.25                        vpmovm2q       %k0, %xmm0
-# CHECK-NEXT:  1      1     0.25                        vpmovm2d       %k0, %ymm0
-# CHECK-NEXT:  1      1     0.25                        vpmovm2q       %k0, %ymm0
+# CHECK-NEXT:  1      1     0.33                        vpmovm2d       %k0, %xmm0
+# CHECK-NEXT:  1      1     0.33                        vpmovm2q       %k0, %xmm0
+# CHECK-NEXT:  1      1     0.33                        vpmovm2d       %k0, %ymm0
+# CHECK-NEXT:  1      1     0.33                        vpmovm2q       %k0, %ymm0
 # CHECK-NEXT:  1      1     1.00                        vpmovd2m       %xmm0, %k0
 # CHECK-NEXT:  1      1     1.00                        vpmovq2m       %xmm0, %k0
 # CHECK-NEXT:  1      1     1.00                        vpmovd2m       %ymm0, %k0
@@ -480,7 +480,7 @@ vpmovq2m          %ymm0, %k0
 
 # CHECK:      Resource pressure per iteration:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]
-# CHECK-NEXT:  -      -     77.00  73.00  66.50  66.50   -     85.00  1.00    -
+# CHECK-NEXT:  -      -     77.33  73.33  66.50  66.50   -     85.33   -      -
 
 # CHECK:      Resource pressure by instruction:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]    Instructions:
@@ -688,10 +688,10 @@ vpmovq2m          %ymm0, %k0
 # CHECK-NEXT:  -      -     0.33   0.33    -      -      -     0.33    -      -     vxorps     %ymm16, %ymm17, %ymm19 {%k1} {z}
 # CHECK-NEXT:  -      -     0.33   0.33   0.50   0.50    -     0.33    -      -     vxorps     (%rax), %ymm17, %ymm19 {%k1} {z}
 # CHECK-NEXT:  -      -     0.33   0.33   0.50   0.50    -     0.33    -      -     vxorps     (%rax){1to8}, %ymm17, %ymm19 {%k1} {z}
-# CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     vpmovm2d   %k0, %xmm0
-# CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     vpmovm2q   %k0, %xmm0
-# CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     vpmovm2d   %k0, %ymm0
-# CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     vpmovm2q   %k0, %ymm0
+# CHECK-NEXT:  -      -     0.33   0.33    -      -      -     0.33    -      -     vpmovm2d   %k0, %xmm0
+# CHECK-NEXT:  -      -     0.33   0.33    -      -      -     0.33    -      -     vpmovm2q   %k0, %xmm0
+# CHECK-NEXT:  -      -     0.33   0.33    -      -      -     0.33    -      -     vpmovm2d   %k0, %ymm0
+# CHECK-NEXT:  -      -     0.33   0.33    -      -      -     0.33    -      -     vpmovm2q   %k0, %ymm0
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -     vpmovd2m   %xmm0, %k0
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -     vpmovq2m   %xmm0, %k0
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -     vpmovd2m   %ymm0, %k0