test/tools/llvm-mca/X86/BtVer2/partial-reg-update-6.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=1500 -timeline -timeline-max-iterations=3 < %s | FileCheck %s
   3
   4 # perf stat reports a throughput of 0.60 IPC for this code snippet.
   5 # Each lzcnt has a false dependency on %ecx; the first lzcnt has to wait on the
   6 # imul. However, the folded load can start immediately.
   7 # The last lzcnt has a false dependency on %cx. However, even in this case, the
   8 # folded load can start immediately.
   9
  10 imul %edx, %ecx
  11 lzcnt (%rsp), %cx
  12 lzcnt 2(%rsp), %cx
  13
  14 # CHECK:      Iterations:        1500
  15 # CHECK-NEXT: Instructions:      4500
  16 # CHECK-NEXT: Total Cycles:      7503
  17 # CHECK-NEXT: Total uOps:        4500
  18
  19 # CHECK:      Dispatch Width:    2
  20 # CHECK-NEXT: uOps Per Cycle:    0.60
  21 # CHECK-NEXT: IPC:               0.60
  22 # CHECK-NEXT: Block RThroughput: 2.0
  23
  24 # CHECK:      Instruction Info:
  25 # CHECK-NEXT: [1]: #uOps
  26 # CHECK-NEXT: [2]: Latency
  27 # CHECK-NEXT: [3]: RThroughput
  28 # CHECK-NEXT: [4]: MayLoad
  29 # CHECK-NEXT: [5]: MayStore
  30 # CHECK-NEXT: [6]: HasSideEffects (U)
  31
  32 # CHECK:      [1]    [2]    [3]    [4]    [5]    [6]    Instructions:
  33 # CHECK-NEXT:  1      3     1.00                        imull   %edx, %ecx
  34 # CHECK-NEXT:  1      4     1.00    *                   lzcntw  (%rsp), %cx
  35 # CHECK-NEXT:  1      4     1.00    *                   lzcntw  2(%rsp), %cx
  36
  37 # CHECK:      Resources:
  38 # CHECK-NEXT: [0]   - JALU0
  39 # CHECK-NEXT: [1]   - JALU1
  40 # CHECK-NEXT: [2]   - JDiv
  41 # CHECK-NEXT: [3]   - JFPA
  42 # CHECK-NEXT: [4]   - JFPM
  43 # CHECK-NEXT: [5]   - JFPU0
  44 # CHECK-NEXT: [6]   - JFPU1
  45 # CHECK-NEXT: [7]   - JLAGU
  46 # CHECK-NEXT: [8]   - JMul
  47 # CHECK-NEXT: [9]   - JSAGU
  48 # CHECK-NEXT: [10]  - JSTC
  49 # CHECK-NEXT: [11]  - JVALU0
  50 # CHECK-NEXT: [12]  - JVALU1
  51 # CHECK-NEXT: [13]  - JVIMUL
  52
  53 # CHECK:      Resource pressure per iteration:
  54 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]    [10]   [11]   [12]   [13]
  55 # CHECK-NEXT: 1.50   1.50    -      -      -      -      -     2.00   1.00    -      -      -      -      -
  56
  57 # CHECK:      Resource pressure by instruction:
  58 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]    [10]   [11]   [12]   [13]   Instructions:
  59 # CHECK-NEXT:  -     1.00    -      -      -      -      -      -     1.00    -      -      -      -      -     imull   %edx, %ecx
  60 # CHECK-NEXT: 1.00    -      -      -      -      -      -     1.00    -      -      -      -      -      -     lzcntw  (%rsp), %cx
  61 # CHECK-NEXT: 0.50   0.50    -      -      -      -      -     1.00    -      -      -      -      -      -     lzcntw  2(%rsp), %cx
  62
  63 # CHECK:      Timeline view:
  64 # CHECK-NEXT:                     01234567
  65 # CHECK-NEXT: Index     0123456789
  66
  67 # CHECK:      [0,0]     DeeeER    .    . .   imull      %edx, %ecx
  68 # CHECK-NEXT: [0,1]     DeeeeER   .    . .   lzcntw     (%rsp), %cx
  69 # CHECK-NEXT: [0,2]     .DeeeeER  .    . .   lzcntw     2(%rsp), %cx
  70 # CHECK-NEXT: [1,0]     .D====eeeER    . .   imull      %edx, %ecx
  71 # CHECK-NEXT: [1,1]     . D===eeeeER   . .   lzcntw     (%rsp), %cx
  72 # CHECK-NEXT: [1,2]     . D====eeeeER  . .   lzcntw     2(%rsp), %cx
  73 # CHECK-NEXT: [2,0]     .  D=======eeeER .   imull      %edx, %ecx
  74 # CHECK-NEXT: [2,1]     .  D=======eeeeER.   lzcntw     (%rsp), %cx
  75 # CHECK-NEXT: [2,2]     .   D=======eeeeER   lzcntw     2(%rsp), %cx
  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.     3     4.7    0.3    0.0       imull        %edx, %ecx
  85 # CHECK-NEXT: 1.     3     4.3    0.0    0.0       lzcntw       (%rsp), %cx
  86 # CHECK-NEXT: 2.     3     4.7    0.0    0.0       lzcntw       2(%rsp), %cx
  87 # CHECK-NEXT:        3     4.6    0.1    0.0       <total>