test/tools/llvm-mca/X86/Generic/avx512-super-registers-1.s

   1 # NOTE: Assertions have been autogenerated by utils/update_mca_test_checks.py
   2 # RUN: llvm-mca -mtriple=x86_64-unknown-unknown -mcpu=x86-64 -timeline -timeline-max-iterations=2 < %s | FileCheck %s
   3
   4   vmulps  %zmm0, %zmm1, %zmm2
   5   vaddps  %xmm1, %xmm1, %xmm2
   6   vmulps  %ymm2, %ymm3, %ymm4
   7   vaddps  %xmm4, %xmm5, %xmm6
   8   vmulps  %xmm6, %xmm3, %xmm4
   9   vaddps  %xmm4, %xmm5, %xmm0
  10
  11 # CHECK:      Iterations:        100
  12 # CHECK-NEXT: Instructions:      600
  13 # CHECK-NEXT: Total Cycles:      318
  14 # CHECK-NEXT: Total uOps:        600
  15
  16 # CHECK:      Dispatch Width:    4
  17 # CHECK-NEXT: uOps Per Cycle:    1.89
  18 # CHECK-NEXT: IPC:               1.89
  19 # CHECK-NEXT: Block RThroughput: 3.0
  20
  21 # CHECK:      Instruction Info:
  22 # CHECK-NEXT: [1]: #uOps
  23 # CHECK-NEXT: [2]: Latency
  24 # CHECK-NEXT: [3]: RThroughput
  25 # CHECK-NEXT: [4]: MayLoad
  26 # CHECK-NEXT: [5]: MayStore
  27 # CHECK-NEXT: [6]: HasSideEffects (U)
  28
  29 # CHECK:      [1]    [2]    [3]    [4]    [5]    [6]    Instructions:
  30 # CHECK-NEXT:  1      5     1.00                        vmulps  %zmm0, %zmm1, %zmm2
  31 # CHECK-NEXT:  1      3     1.00                        vaddps  %xmm1, %xmm1, %xmm2
  32 # CHECK-NEXT:  1      5     1.00                        vmulps  %ymm2, %ymm3, %ymm4
  33 # CHECK-NEXT:  1      3     1.00                        vaddps  %xmm4, %xmm5, %xmm6
  34 # CHECK-NEXT:  1      5     1.00                        vmulps  %xmm6, %xmm3, %xmm4
  35 # CHECK-NEXT:  1      3     1.00                        vaddps  %xmm4, %xmm5, %xmm0
  36
  37 # CHECK:      Resources:
  38 # CHECK-NEXT: [0]   - SBDivider
  39 # CHECK-NEXT: [1]   - SBFPDivider
  40 # CHECK-NEXT: [2]   - SBPort0
  41 # CHECK-NEXT: [3]   - SBPort1
  42 # CHECK-NEXT: [4]   - SBPort4
  43 # CHECK-NEXT: [5]   - SBPort5
  44 # CHECK-NEXT: [6.0] - SBPort23
  45 # CHECK-NEXT: [6.1] - SBPort23
  46
  47 # CHECK:      Resource pressure per iteration:
  48 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6.0]  [6.1]
  49 # CHECK-NEXT:  -      -     3.00   3.00    -      -      -      -
  50
  51 # CHECK:      Resource pressure by instruction:
  52 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6.0]  [6.1]  Instructions:
  53 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -     vmulps    %zmm0, %zmm1, %zmm2
  54 # CHECK-NEXT:  -      -      -     1.00    -      -      -      -     vaddps    %xmm1, %xmm1, %xmm2
  55 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -     vmulps    %ymm2, %ymm3, %ymm4
  56 # CHECK-NEXT:  -      -      -     1.00    -      -      -      -     vaddps    %xmm4, %xmm5, %xmm6
  57 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -     vmulps    %xmm6, %xmm3, %xmm4
  58 # CHECK-NEXT:  -      -      -     1.00    -      -      -      -     vaddps    %xmm4, %xmm5, %xmm0
  59
  60 # CHECK:      Timeline view:
  61 # CHECK-NEXT:                     0123456789
  62 # CHECK-NEXT: Index     0123456789          0123456
  63
  64 # CHECK:      [0,0]     DeeeeeER  .    .    .    ..   vmulps    %zmm0, %zmm1, %zmm2
  65 # CHECK-NEXT: [0,1]     DeeeE--R  .    .    .    ..   vaddps    %xmm1, %xmm1, %xmm2
  66 # CHECK-NEXT: [0,2]     D===eeeeeER    .    .    ..   vmulps    %ymm2, %ymm3, %ymm4
  67 # CHECK-NEXT: [0,3]     D========eeeER .    .    ..   vaddps    %xmm4, %xmm5, %xmm6
  68 # CHECK-NEXT: [0,4]     .D==========eeeeeER .    ..   vmulps    %xmm6, %xmm3, %xmm4
  69 # CHECK-NEXT: [0,5]     .D===============eeeER   ..   vaddps    %xmm4, %xmm5, %xmm0
  70 # CHECK-NEXT: [1,0]     .D==================eeeeeER   vmulps    %zmm0, %zmm1, %zmm2
  71 # CHECK-NEXT: [1,1]     .DeeeE--------------------R   vaddps    %xmm1, %xmm1, %xmm2
  72 # CHECK-NEXT: [1,2]     . D==eeeeeE---------------R   vmulps    %ymm2, %ymm3, %ymm4
  73 # CHECK-NEXT: [1,3]     . D=======eeeE------------R   vaddps    %xmm4, %xmm5, %xmm6
  74 # CHECK-NEXT: [1,4]     . D==========eeeeeE-------R   vmulps    %xmm6, %xmm3, %xmm4
  75 # CHECK-NEXT: [1,5]     . D===============eeeE----R   vaddps    %xmm4, %xmm5, %xmm0
  76
  77 # CHECK:      Average Wait times (based on the timeline view):
  78 # CHECK-NEXT: [0]: Executions
  79 # CHECK-NEXT: [1]: Average time spent waiting in a scheduler's queue
  80 # CHECK-NEXT: [2]: Average time spent waiting in a scheduler's queue while ready
  81 # CHECK-NEXT: [3]: Average time elapsed from WB until retire stage
  82
  83 # CHECK:            [0]    [1]    [2]    [3]
  84 # CHECK-NEXT: 0.     2     10.0   0.5    0.0       vmulps       %zmm0, %zmm1, %zmm2
  85 # CHECK-NEXT: 1.     2     1.0    1.0    11.0      vaddps       %xmm1, %xmm1, %xmm2
  86 # CHECK-NEXT: 2.     2     3.5    0.0    7.5       vmulps       %ymm2, %ymm3, %ymm4
  87 # CHECK-NEXT: 3.     2     8.5    0.0    6.0       vaddps       %xmm4, %xmm5, %xmm6
  88 # CHECK-NEXT: 4.     2     11.0   0.0    3.5       vmulps       %xmm6, %xmm3, %xmm4
  89 # CHECK-NEXT: 5.     2     16.0   0.0    2.0       vaddps       %xmm4, %xmm5, %xmm0