test/tools/llvm-mca/X86/BtVer2/memcpy-like-test.s

   1 # NOTE: Assertions have been autogenerated by utils/update_mca_test_checks.py
   2 # RUN: llvm-mca -mtriple=x86_64-unknown-unknown -mcpu=btver2 -iterations=100 -timeline -timeline-max-iterations=1 < %s | FileCheck %s
   3
   4 vmovaps (%rsi), %xmm0
   5 vmovaps %xmm0, (%rdi)
   6 vmovaps 16(%rsi), %xmm0
   7 vmovaps %xmm0, 16(%rdi)
   8 vmovaps 32(%rsi), %xmm0
   9 vmovaps %xmm0, 32(%rdi)
  10 vmovaps 48(%rsi), %xmm0
  11 vmovaps %xmm0, 48(%rdi)
  12
  13 # CHECK:      Iterations:        100
  14 # CHECK-NEXT: Instructions:      800
  15 # CHECK-NEXT: Total Cycles:      408
  16 # CHECK-NEXT: Total uOps:        800
  17
  18 # CHECK:      Dispatch Width:    2
  19 # CHECK-NEXT: uOps Per Cycle:    1.96
  20 # CHECK-NEXT: IPC:               1.96
  21 # CHECK-NEXT: Block RThroughput: 4.0
  22
  23 # CHECK:      Instruction Info:
  24 # CHECK-NEXT: [1]: #uOps
  25 # CHECK-NEXT: [2]: Latency
  26 # CHECK-NEXT: [3]: RThroughput
  27 # CHECK-NEXT: [4]: MayLoad
  28 # CHECK-NEXT: [5]: MayStore
  29 # CHECK-NEXT: [6]: HasSideEffects (U)
  30
  31 # CHECK:      [1]    [2]    [3]    [4]    [5]    [6]    Instructions:
  32 # CHECK-NEXT:  1      5     1.00    *                   vmovaps (%rsi), %xmm0
  33 # CHECK-NEXT:  1      1     1.00           *            vmovaps %xmm0, (%rdi)
  34 # CHECK-NEXT:  1      5     1.00    *                   vmovaps 16(%rsi), %xmm0
  35 # CHECK-NEXT:  1      1     1.00           *            vmovaps %xmm0, 16(%rdi)
  36 # CHECK-NEXT:  1      5     1.00    *                   vmovaps 32(%rsi), %xmm0
  37 # CHECK-NEXT:  1      1     1.00           *            vmovaps %xmm0, 32(%rdi)
  38 # CHECK-NEXT:  1      5     1.00    *                   vmovaps 48(%rsi), %xmm0
  39 # CHECK-NEXT:  1      1     1.00           *            vmovaps %xmm0, 48(%rdi)
  40
  41 # CHECK:      Resources:
  42 # CHECK-NEXT: [0]   - JALU0
  43 # CHECK-NEXT: [1]   - JALU1
  44 # CHECK-NEXT: [2]   - JDiv
  45 # CHECK-NEXT: [3]   - JFPA
  46 # CHECK-NEXT: [4]   - JFPM
  47 # CHECK-NEXT: [5]   - JFPU0
  48 # CHECK-NEXT: [6]   - JFPU1
  49 # CHECK-NEXT: [7]   - JLAGU
  50 # CHECK-NEXT: [8]   - JMul
  51 # CHECK-NEXT: [9]   - JSAGU
  52 # CHECK-NEXT: [10]  - JSTC
  53 # CHECK-NEXT: [11]  - JVALU0
  54 # CHECK-NEXT: [12]  - JVALU1
  55 # CHECK-NEXT: [13]  - JVIMUL
  56
  57 # CHECK:      Resource pressure per iteration:
  58 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]    [10]   [11]   [12]   [13]
  59 # CHECK-NEXT:  -      -      -     2.00   2.00   3.97   4.03   4.00    -     4.00   4.00    -      -      -
  60
  61 # CHECK:      Resource pressure by instruction:
  62 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]    [10]   [11]   [12]   [13]   Instructions:
  63 # CHECK-NEXT:  -      -      -      -     1.00   0.98   0.02   1.00    -      -      -      -      -      -     vmovaps (%rsi), %xmm0
  64 # CHECK-NEXT:  -      -      -      -      -      -     1.00    -      -     1.00   1.00    -      -      -     vmovaps %xmm0, (%rdi)
  65 # CHECK-NEXT:  -      -      -     1.00    -     1.00    -     1.00    -      -      -      -      -      -     vmovaps 16(%rsi), %xmm0
  66 # CHECK-NEXT:  -      -      -      -      -      -     1.00    -      -     1.00   1.00    -      -      -     vmovaps %xmm0, 16(%rdi)
  67 # CHECK-NEXT:  -      -      -      -     1.00   0.99   0.01   1.00    -      -      -      -      -      -     vmovaps 32(%rsi), %xmm0
  68 # CHECK-NEXT:  -      -      -      -      -      -     1.00    -      -     1.00   1.00    -      -      -     vmovaps %xmm0, 32(%rdi)
  69 # CHECK-NEXT:  -      -      -     1.00    -     1.00    -     1.00    -      -      -      -      -      -     vmovaps 48(%rsi), %xmm0
  70 # CHECK-NEXT:  -      -      -      -      -      -     1.00    -      -     1.00   1.00    -      -      -     vmovaps %xmm0, 48(%rdi)
  71
  72 # CHECK:      Timeline view:
  73 # CHECK-NEXT:                     01
  74 # CHECK-NEXT: Index     0123456789
  75
  76 # CHECK:      [0,0]     DeeeeeER  ..   vmovaps  (%rsi), %xmm0
  77 # CHECK-NEXT: [0,1]     D=====eER ..   vmovaps  %xmm0, (%rdi)
  78 # CHECK-NEXT: [0,2]     .DeeeeeER ..   vmovaps  16(%rsi), %xmm0
  79 # CHECK-NEXT: [0,3]     .D=====eER..   vmovaps  %xmm0, 16(%rdi)
  80 # CHECK-NEXT: [0,4]     . DeeeeeER..   vmovaps  32(%rsi), %xmm0
  81 # CHECK-NEXT: [0,5]     . D=====eER.   vmovaps  %xmm0, 32(%rdi)
  82 # CHECK-NEXT: [0,6]     .  DeeeeeER.   vmovaps  48(%rsi), %xmm0
  83 # CHECK-NEXT: [0,7]     .  D=====eER   vmovaps  %xmm0, 48(%rdi)
  84
  85 # CHECK:      Average Wait times (based on the timeline view):
  86 # CHECK-NEXT: [0]: Executions
  87 # CHECK-NEXT: [1]: Average time spent waiting in a scheduler's queue
  88 # CHECK-NEXT: [2]: Average time spent waiting in a scheduler's queue while ready
  89 # CHECK-NEXT: [3]: Average time elapsed from WB until retire stage
  90
  91 # CHECK:            [0]    [1]    [2]    [3]
  92 # CHECK-NEXT: 0.     1     1.0    1.0    0.0       vmovaps      (%rsi), %xmm0
  93 # CHECK-NEXT: 1.     1     6.0    0.0    0.0       vmovaps      %xmm0, (%rdi)
  94 # CHECK-NEXT: 2.     1     1.0    1.0    0.0       vmovaps      16(%rsi), %xmm0
  95 # CHECK-NEXT: 3.     1     6.0    0.0    0.0       vmovaps      %xmm0, 16(%rdi)
  96 # CHECK-NEXT: 4.     1     1.0    1.0    0.0       vmovaps      32(%rsi), %xmm0
  97 # CHECK-NEXT: 5.     1     6.0    0.0    0.0       vmovaps      %xmm0, 32(%rdi)
  98 # CHECK-NEXT: 6.     1     1.0    1.0    0.0       vmovaps      48(%rsi), %xmm0
  99 # CHECK-NEXT: 7.     1     6.0    0.0    0.0       vmovaps      %xmm0, 48(%rdi)