| 1 |
# NOTE: Assertions have been autogenerated by utils/update_mca_test_checks.py |
| 2 |
# RUN: llvm-mca -mtriple=aarch64 -mcpu=tsv110 --instruction-info=0 --resource-pressure=0 --timeline --iterations=1 < %s | FileCheck %s |
| 3 |
|
| 4 |
# LLVM-MCA-BEGIN madd nobypass |
| 5 |
mul x0, x1, x2 |
| 6 |
add x0, x0, x1 |
| 7 |
add x0, x0, x1 |
| 8 |
add x0, x0, x1 |
| 9 |
# LLVM-MCA-END |
| 10 |
|
| 11 |
# LLVM-MCA-BEGIN madd bypass |
| 12 |
mul x0, x1, x2 |
| 13 |
madd x0, x1, x2, x0 |
| 14 |
madd x0, x1, x2, x0 |
| 15 |
madd x0, x0, x0, x0 |
| 16 |
# LLVM-MCA-END |
| 17 |
|
| 18 |
# CHECK: [0] Code Region - madd nobypass |
| 19 |
|
| 20 |
# CHECK: Iterations: 1 |
| 21 |
# CHECK-NEXT: Instructions: 4 |
| 22 |
# CHECK-NEXT: Total Cycles: 10 |
| 23 |
# CHECK-NEXT: Total uOps: 4 |
| 24 |
|
| 25 |
# CHECK: Dispatch Width: 4 |
| 26 |
# CHECK-NEXT: uOps Per Cycle: 0.40 |
| 27 |
# CHECK-NEXT: IPC: 0.40 |
| 28 |
# CHECK-NEXT: Block RThroughput: 1.0 |
| 29 |
|
| 30 |
# CHECK: Timeline view: |
| 31 |
# CHECK-NEXT: Index 0123456789 |
| 32 |
|
| 33 |
# CHECK: [0,0] DeeeeER . mul x0, x1, x2 |
| 34 |
# CHECK-NEXT: [0,1] D====eER . add x0, x0, x1 |
| 35 |
# CHECK-NEXT: [0,2] D=====eER. add x0, x0, x1 |
| 36 |
# CHECK-NEXT: [0,3] D======eER add x0, x0, x1 |
| 37 |
|
| 38 |
# CHECK: Average Wait times (based on the timeline view): |
| 39 |
# CHECK-NEXT: [0]: Executions |
| 40 |
# CHECK-NEXT: [1]: Average time spent waiting in a scheduler's queue |
| 41 |
# CHECK-NEXT: [2]: Average time spent waiting in a scheduler's queue while ready |
| 42 |
# CHECK-NEXT: [3]: Average time elapsed from WB until retire stage |
| 43 |
|
| 44 |
# CHECK: [0] [1] [2] [3] |
| 45 |
# CHECK-NEXT: 0. 1 1.0 1.0 0.0 mul x0, x1, x2 |
| 46 |
# CHECK-NEXT: 1. 1 5.0 0.0 0.0 add x0, x0, x1 |
| 47 |
# CHECK-NEXT: 2. 1 6.0 0.0 0.0 add x0, x0, x1 |
| 48 |
# CHECK-NEXT: 3. 1 7.0 0.0 0.0 add x0, x0, x1 |
| 49 |
# CHECK-NEXT: 1 4.8 0.3 0.0 <total> |
| 50 |
|
| 51 |
# CHECK: [1] Code Region - madd bypass |
| 52 |
|
| 53 |
# CHECK: Iterations: 1 |
| 54 |
# CHECK-NEXT: Instructions: 4 |
| 55 |
# CHECK-NEXT: Total Cycles: 12 |
| 56 |
# CHECK-NEXT: Total uOps: 4 |
| 57 |
|
| 58 |
# CHECK: Dispatch Width: 4 |
| 59 |
# CHECK-NEXT: uOps Per Cycle: 0.33 |
| 60 |
# CHECK-NEXT: IPC: 0.33 |
| 61 |
# CHECK-NEXT: Block RThroughput: 4.0 |
| 62 |
|
| 63 |
# CHECK: Timeline view: |
| 64 |
# CHECK-NEXT: 01 |
| 65 |
# CHECK-NEXT: Index 0123456789 |
| 66 |
|
| 67 |
# CHECK: [0,0] DeeeeER .. mul x0, x1, x2 |
| 68 |
# CHECK-NEXT: [0,1] D==eeeeER .. madd x0, x1, x2, x0 |
| 69 |
# CHECK-NEXT: [0,2] D=eeeeE-R .. madd x0, x1, x2, x0 |
| 70 |
# CHECK-NEXT: [0,3] D=====eeeeER madd x0, x0, x0, x0 |
| 71 |
|
| 72 |
# CHECK: Average Wait times (based on the timeline view): |
| 73 |
# CHECK-NEXT: [0]: Executions |
| 74 |
# CHECK-NEXT: [1]: Average time spent waiting in a scheduler's queue |
| 75 |
# CHECK-NEXT: [2]: Average time spent waiting in a scheduler's queue while ready |
| 76 |
# CHECK-NEXT: [3]: Average time elapsed from WB until retire stage |
| 77 |
|
| 78 |
# CHECK: [0] [1] [2] [3] |
| 79 |
# CHECK-NEXT: 0. 1 1.0 1.0 0.0 mul x0, x1, x2 |
| 80 |
# CHECK-NEXT: 1. 1 3.0 3.0 0.0 madd x0, x1, x2, x0 |
| 81 |
# CHECK-NEXT: 2. 1 2.0 2.0 1.0 madd x0, x1, x2, x0 |
| 82 |
# CHECK-NEXT: 3. 1 6.0 0.0 0.0 madd x0, x0, x0, x0 |
| 83 |
# CHECK-NEXT: 1 3.0 1.5 0.3 <total> |