test/tools/llvm-mca/X86/Barcelona/reg-move-elimination-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 -iterations=3 -timeline -register-file-stats < %s | FileCheck %s
   3
   4 # The register move from XMM0 to XMM1 can be eliminated at register renaming
   5 # stage. So, it should not consume pipeline resources.
   6
   7 xorps %xmm0, %xmm0
   8 movaps %xmm0, %xmm1
   9 addps %xmm1, %xmm1
  10
  11 # CHECK:      Iterations:        3
  12 # CHECK-NEXT: Instructions:      9
  13 # CHECK-NEXT: Total Cycles:      9
  14 # CHECK-NEXT: Total uOps:        9
  15
  16 # CHECK:      Dispatch Width:    4
  17 # CHECK-NEXT: uOps Per Cycle:    1.00
  18 # CHECK-NEXT: IPC:               1.00
  19 # CHECK-NEXT: Block RThroughput: 1.0
  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      0     0.25                        xorps   %xmm0, %xmm0
  31 # CHECK-NEXT:  1      1     1.00                        movaps  %xmm0, %xmm1
  32 # CHECK-NEXT:  1      3     1.00                        addps   %xmm1, %xmm1
  33
  34 # CHECK:      Register File statistics:
  35 # CHECK-NEXT: Total number of mappings created:    9
  36 # CHECK-NEXT: Max number of mappings used:         8
  37
  38 # CHECK:      Resources:
  39 # CHECK-NEXT: [0]   - SBDivider
  40 # CHECK-NEXT: [1]   - SBFPDivider
  41 # CHECK-NEXT: [2]   - SBPort0
  42 # CHECK-NEXT: [3]   - SBPort1
  43 # CHECK-NEXT: [4]   - SBPort4
  44 # CHECK-NEXT: [5]   - SBPort5
  45 # CHECK-NEXT: [6.0] - SBPort23
  46 # CHECK-NEXT: [6.1] - SBPort23
  47
  48 # CHECK:      Resource pressure per iteration:
  49 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6.0]  [6.1]
  50 # CHECK-NEXT:  -      -      -     1.00    -     1.00    -      -
  51
  52 # CHECK:      Resource pressure by instruction:
  53 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6.0]  [6.1]  Instructions:
  54 # CHECK-NEXT:  -      -      -      -      -      -      -      -     xorps     %xmm0, %xmm0
  55 # CHECK-NEXT:  -      -      -      -      -     1.00    -      -     movaps    %xmm0, %xmm1
  56 # CHECK-NEXT:  -      -      -     1.00    -      -      -      -     addps     %xmm1, %xmm1
  57
  58 # CHECK:      Timeline view:
  59 # CHECK-NEXT: Index     012345678
  60
  61 # CHECK:      [0,0]     DR   .  .   xorps       %xmm0, %xmm0
  62 # CHECK-NEXT: [0,1]     DeER .  .   movaps      %xmm0, %xmm1
  63 # CHECK-NEXT: [0,2]     D=eeeER .   addps       %xmm1, %xmm1
  64 # CHECK-NEXT: [1,0]     D-----R .   xorps       %xmm0, %xmm0
  65 # CHECK-NEXT: [1,1]     .DeE--R .   movaps      %xmm0, %xmm1
  66 # CHECK-NEXT: [1,2]     .D=eeeER.   addps       %xmm1, %xmm1
  67 # CHECK-NEXT: [2,0]     .D-----R.   xorps       %xmm0, %xmm0
  68 # CHECK-NEXT: [2,1]     .D=eE--R.   movaps      %xmm0, %xmm1
  69 # CHECK-NEXT: [2,2]     . D=eeeER   addps       %xmm1, %xmm1
  70
  71 # CHECK:      Average Wait times (based on the timeline view):
  72 # CHECK-NEXT: [0]: Executions
  73 # CHECK-NEXT: [1]: Average time spent waiting in a scheduler's queue
  74 # CHECK-NEXT: [2]: Average time spent waiting in a scheduler's queue while ready
  75 # CHECK-NEXT: [3]: Average time elapsed from WB until retire stage
  76
  77 # CHECK:            [0]    [1]    [2]    [3]
  78 # CHECK-NEXT: 0.     3     0.0    0.0    3.3       xorps        %xmm0, %xmm0
  79 # CHECK-NEXT: 1.     3     1.3    1.3    1.3       movaps       %xmm0, %xmm1
  80 # CHECK-NEXT: 2.     3     2.0    0.0    0.0       addps        %xmm1, %xmm1
  81 # CHECK-NEXT:        3     1.1    0.4    1.6       <total>