test/tools/llvm-mca/X86/BdVer2/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=bdver2 -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 vxorps %xmm0, %xmm0, %xmm0
   8 vmovaps %xmm0, %xmm1
   9 vaddps %xmm1, %xmm1, %xmm2
  10
  11 # CHECK:      Iterations:        3
  12 # CHECK-NEXT: Instructions:      9
  13 # CHECK-NEXT: Total Cycles:      13
  14 # CHECK-NEXT: Total uOps:        9
  15
  16 # CHECK:      Dispatch Width:    4
  17 # CHECK-NEXT: uOps Per Cycle:    0.69
  18 # CHECK-NEXT: IPC:               0.69
  19 # CHECK-NEXT: Block RThroughput: 1.5
  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                        vxorps  %xmm0, %xmm0, %xmm0
  31 # CHECK-NEXT:  1      1     1.00                        vmovaps %xmm0, %xmm1
  32 # CHECK-NEXT:  1      5     1.00                        vaddps  %xmm1, %xmm1, %xmm2
  33
  34 # CHECK:      Register File statistics:
  35 # CHECK-NEXT: Total number of mappings created:    6
  36 # CHECK-NEXT: Max number of mappings used:         6
  37
  38 # CHECK:      *  Register File #1 -- PdFpuPRF:
  39 # CHECK-NEXT:    Number of physical registers:     160
  40 # CHECK-NEXT:    Total number of mappings created: 6
  41 # CHECK-NEXT:    Max number of mappings used:      6
  42
  43 # CHECK:      *  Register File #2 -- PdIntegerPRF:
  44 # CHECK-NEXT:    Number of physical registers:     96
  45 # CHECK-NEXT:    Total number of mappings created: 0
  46 # CHECK-NEXT:    Max number of mappings used:      0
  47
  48 # CHECK:      Resources:
  49 # CHECK-NEXT: [0.0] - PdAGLU01
  50 # CHECK-NEXT: [0.1] - PdAGLU01
  51 # CHECK-NEXT: [1]   - PdBranch
  52 # CHECK-NEXT: [2]   - PdCount
  53 # CHECK-NEXT: [3]   - PdDiv
  54 # CHECK-NEXT: [4]   - PdEX0
  55 # CHECK-NEXT: [5]   - PdEX1
  56 # CHECK-NEXT: [6]   - PdFPCVT
  57 # CHECK-NEXT: [7.0] - PdFPFMA
  58 # CHECK-NEXT: [7.1] - PdFPFMA
  59 # CHECK-NEXT: [8.0] - PdFPMAL
  60 # CHECK-NEXT: [8.1] - PdFPMAL
  61 # CHECK-NEXT: [9]   - PdFPMMA
  62 # CHECK-NEXT: [10]  - PdFPSTO
  63 # CHECK-NEXT: [11]  - PdFPU0
  64 # CHECK-NEXT: [12]  - PdFPU1
  65 # CHECK-NEXT: [13]  - PdFPU2
  66 # CHECK-NEXT: [14]  - PdFPU3
  67 # CHECK-NEXT: [15]  - PdFPXBR
  68 # CHECK-NEXT: [16.0] - PdLoad
  69 # CHECK-NEXT: [16.1] - PdLoad
  70 # CHECK-NEXT: [17]  - PdMul
  71 # CHECK-NEXT: [18]  - PdStore
  72
  73 # CHECK:      Resource pressure per iteration:
  74 # CHECK-NEXT: [0.0]  [0.1]  [1]    [2]    [3]    [4]    [5]    [6]    [7.0]  [7.1]  [8.0]  [8.1]  [9]    [10]   [11]   [12]   [13]   [14]   [15]   [16.0] [16.1] [17]   [18]
  75 # CHECK-NEXT:  -      -      -      -      -      -      -      -     1.33   1.67    -      -      -      -     1.33   0.67    -      -      -      -      -      -      -
  76
  77 # CHECK:      Resource pressure by instruction:
  78 # CHECK-NEXT: [0.0]  [0.1]  [1]    [2]    [3]    [4]    [5]    [6]    [7.0]  [7.1]  [8.0]  [8.1]  [9]    [10]   [11]   [12]   [13]   [14]   [15]   [16.0] [16.1] [17]   [18]   Instructions:
  79 # CHECK-NEXT:  -      -      -      -      -      -      -      -      -      -      -      -      -      -      -      -      -      -      -      -      -      -      -     vxorps   %xmm0, %xmm0, %xmm0
  80 # CHECK-NEXT:  -      -      -      -      -      -      -      -     0.67   1.33    -      -      -      -     0.33   0.67    -      -      -      -      -      -      -     vmovaps  %xmm0, %xmm1
  81 # CHECK-NEXT:  -      -      -      -      -      -      -      -     0.67   0.33    -      -      -      -     1.00    -      -      -      -      -      -      -      -     vaddps   %xmm1, %xmm1, %xmm2
  82
  83 # CHECK:      Timeline view:
  84 # CHECK-NEXT:                     012
  85 # CHECK-NEXT: Index     0123456789
  86
  87 # CHECK:      [0,0]     DR   .    . .   vxorps  %xmm0, %xmm0, %xmm0
  88 # CHECK-NEXT: [0,1]     DeER .    . .   vmovaps %xmm0, %xmm1
  89 # CHECK-NEXT: [0,2]     D=eeeeeER . .   vaddps  %xmm1, %xmm1, %xmm2
  90 # CHECK-NEXT: [1,0]     D-------R . .   vxorps  %xmm0, %xmm0, %xmm0
  91 # CHECK-NEXT: [1,1]     .D=eE---R . .   vmovaps %xmm0, %xmm1
  92 # CHECK-NEXT: [1,2]     .D===eeeeeER.   vaddps  %xmm1, %xmm1, %xmm2
  93 # CHECK-NEXT: [2,0]     .D---------R.   vxorps  %xmm0, %xmm0, %xmm0
  94 # CHECK-NEXT: [2,1]     .D=eE------R.   vmovaps %xmm0, %xmm1
  95 # CHECK-NEXT: [2,2]     . D===eeeeeER   vaddps  %xmm1, %xmm1, %xmm2
  96
  97 # CHECK:      Average Wait times (based on the timeline view):
  98 # CHECK-NEXT: [0]: Executions
  99 # CHECK-NEXT: [1]: Average time spent waiting in a scheduler's queue
 100 # CHECK-NEXT: [2]: Average time spent waiting in a scheduler's queue while ready
 101 # CHECK-NEXT: [3]: Average time elapsed from WB until retire stage
 102
 103 # CHECK:            [0]    [1]    [2]    [3]
 104 # CHECK-NEXT: 0.     3     0.0    0.0    5.3       vxorps       %xmm0, %xmm0, %xmm0
 105 # CHECK-NEXT: 1.     3     1.7    1.7    3.0       vmovaps      %xmm0, %xmm1
 106 # CHECK-NEXT: 2.     3     3.3    1.0    0.0       vaddps       %xmm1, %xmm1, %xmm2
 107 # CHECK-NEXT:        3     1.7    0.9    2.8       <total>