88 lines · plain
1# NOTE: Assertions have been autogenerated by utils/update_mca_test_checks.py2# RUN: llvm-mca -mtriple=amdgcn -mcpu=gfx942 --timeline --iterations=1 --timeline-max-cycles=0 < %s | FileCheck %s3 4v_pk_mov_b32 v[0:1], v[2:3], v[4:5]5v_pk_add_f32 v[0:1], v[0:1], v[0:1]6v_pk_mul_f32 v[0:1], v[0:1], v[0:1]7v_add_co_u32 v5, s[0:1], v1, v28v_sub_co_u32 v5, s[0:1], v1, v29v_add_u32 v5, v1, v210v_sub_u32 v5, v1, v211 12# CHECK: Iterations: 113# CHECK-NEXT: Instructions: 714# CHECK-NEXT: Total Cycles: 1015# CHECK-NEXT: Total uOps: 916 17# CHECK: Dispatch Width: 118# CHECK-NEXT: uOps Per Cycle: 0.9019# CHECK-NEXT: IPC: 0.7020# CHECK-NEXT: Block RThroughput: 9.021 22# CHECK: Instruction Info:23# CHECK-NEXT: [1]: #uOps24# CHECK-NEXT: [2]: Latency25# CHECK-NEXT: [3]: RThroughput26# CHECK-NEXT: [4]: MayLoad27# CHECK-NEXT: [5]: MayStore28# CHECK-NEXT: [6]: HasSideEffects (U)29 30# CHECK: [1] [2] [3] [4] [5] [6] Instructions:31# CHECK-NEXT: 1 1 1.00 U v_pk_mov_b32 v[0:1], v[2:3], v[4:5]32# CHECK-NEXT: 1 1 1.00 U v_pk_add_f32 v[0:1], v[0:1], v[0:1]33# CHECK-NEXT: 1 1 1.00 U v_pk_mul_f32 v[0:1], v[0:1], v[0:1]34# CHECK-NEXT: 2 1 1.00 U v_add_co_u32_e64 v5, s[0:1], v1, v235# CHECK-NEXT: 2 1 1.00 U v_sub_co_u32_e64 v5, s[0:1], v1, v236# CHECK-NEXT: 1 1 1.00 U v_add_u32_e32 v5, v1, v237# CHECK-NEXT: 1 1 1.00 U v_sub_u32_e32 v5, v1, v238 39# CHECK: Resources:40# CHECK-NEXT: [0] - HWBranch41# CHECK-NEXT: [1] - HWExport42# CHECK-NEXT: [2] - HWLGKM43# CHECK-NEXT: [3] - HWSALU44# CHECK-NEXT: [4] - HWVALU45# CHECK-NEXT: [5] - HWVMEM46# CHECK-NEXT: [6] - HWXDL47 48# CHECK: Resource pressure per iteration:49# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6]50# CHECK-NEXT: - - - 2.00 7.00 - -51 52# CHECK: Resource pressure by instruction:53# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] Instructions:54# CHECK-NEXT: - - - - 1.00 - - v_pk_mov_b32 v[0:1], v[2:3], v[4:5]55# CHECK-NEXT: - - - - 1.00 - - v_pk_add_f32 v[0:1], v[0:1], v[0:1]56# CHECK-NEXT: - - - - 1.00 - - v_pk_mul_f32 v[0:1], v[0:1], v[0:1]57# CHECK-NEXT: - - - 1.00 1.00 - - v_add_co_u32_e64 v5, s[0:1], v1, v258# CHECK-NEXT: - - - 1.00 1.00 - - v_sub_co_u32_e64 v5, s[0:1], v1, v259# CHECK-NEXT: - - - - 1.00 - - v_add_u32_e32 v5, v1, v260# CHECK-NEXT: - - - - 1.00 - - v_sub_u32_e32 v5, v1, v261 62# CHECK: Timeline view:63# CHECK-NEXT: Index 012345678964 65# CHECK: [0,0] DE . . v_pk_mov_b32 v[0:1], v[2:3], v[4:5]66# CHECK-NEXT: [0,1] .DE . . v_pk_add_f32 v[0:1], v[0:1], v[0:1]67# CHECK-NEXT: [0,2] . DE . . v_pk_mul_f32 v[0:1], v[0:1], v[0:1]68# CHECK-NEXT: [0,3] . DE. . v_add_co_u32_e64 v5, s[0:1], v1, v269# CHECK-NEXT: [0,4] . DeE . v_sub_co_u32_e64 v5, s[0:1], v1, v270# CHECK-NEXT: [0,5] . . DE. v_add_u32_e32 v5, v1, v271# CHECK-NEXT: [0,6] . . DE v_sub_u32_e32 v5, v1, v272 73# CHECK: Average Wait times (based on the timeline view):74# CHECK-NEXT: [0]: Executions75# CHECK-NEXT: [1]: Average time spent waiting in a scheduler's queue76# CHECK-NEXT: [2]: Average time spent waiting in a scheduler's queue while ready77# CHECK-NEXT: [3]: Average time elapsed from WB until retire stage78 79# CHECK: [0] [1] [2] [3]80# CHECK-NEXT: 0. 1 0.0 0.0 0.0 v_pk_mov_b32 v[0:1], v[2:3], v[4:5]81# CHECK-NEXT: 1. 1 0.0 0.0 0.0 v_pk_add_f32 v[0:1], v[0:1], v[0:1]82# CHECK-NEXT: 2. 1 0.0 0.0 0.0 v_pk_mul_f32 v[0:1], v[0:1], v[0:1]83# CHECK-NEXT: 3. 1 0.0 0.0 0.0 v_add_co_u32_e64 v5, s[0:1], v1, v284# CHECK-NEXT: 4. 1 0.0 0.0 0.0 v_sub_co_u32_e64 v5, s[0:1], v1, v285# CHECK-NEXT: 5. 1 0.0 0.0 0.0 v_add_u32_e32 v5, v1, v286# CHECK-NEXT: 6. 1 0.0 0.0 0.0 v_sub_u32_e32 v5, v1, v287# CHECK-NEXT: 1 0.0 0.0 0.0 <total>88