104 lines
5.1 KiB
ArmAsm
104 lines
5.1 KiB
ArmAsm
|
# NOTE: Assertions have been autogenerated by utils/update_mca_test_checks.py
|
||
|
# RUN: llvm-mca -mtriple=amdgcn -mcpu=gfx1200 --timeline --iterations=1 --timeline-max-cycles=0 < %s | FileCheck %s
|
||
|
|
||
|
v_s_exp_f32 s0, s0
|
||
|
v_s_log_f32 s0, s0
|
||
|
v_s_rcp_f32 s1, s1
|
||
|
v_s_rsq_f32 s1, s0
|
||
|
v_s_sqrt_f32 s2, s1
|
||
|
v_s_exp_f16 s3, s1
|
||
|
v_s_log_f16 s4, s1
|
||
|
v_s_rcp_f16 s5, s2
|
||
|
v_s_rsq_f16 s5, s4
|
||
|
v_s_sqrt_f16 s5, s5
|
||
|
|
||
|
# CHECK: Iterations: 1
|
||
|
# CHECK-NEXT: Instructions: 10
|
||
|
# CHECK-NEXT: Total Cycles: 45
|
||
|
# CHECK-NEXT: Total uOps: 10
|
||
|
|
||
|
# CHECK: Dispatch Width: 1
|
||
|
# CHECK-NEXT: uOps Per Cycle: 0.22
|
||
|
# CHECK-NEXT: IPC: 0.22
|
||
|
# CHECK-NEXT: Block RThroughput: 10.0
|
||
|
|
||
|
# CHECK: Instruction Info:
|
||
|
# CHECK-NEXT: [1]: #uOps
|
||
|
# CHECK-NEXT: [2]: Latency
|
||
|
# CHECK-NEXT: [3]: RThroughput
|
||
|
# CHECK-NEXT: [4]: MayLoad
|
||
|
# CHECK-NEXT: [5]: MayStore
|
||
|
# CHECK-NEXT: [6]: HasSideEffects (U)
|
||
|
|
||
|
# CHECK: [1] [2] [3] [4] [5] [6] Instructions:
|
||
|
# CHECK-NEXT: 1 7 1.00 U v_s_exp_f32 s0, s0
|
||
|
# CHECK-NEXT: 1 7 1.00 U v_s_log_f32 s0, s0
|
||
|
# CHECK-NEXT: 1 7 1.00 U v_s_rcp_f32 s1, s1
|
||
|
# CHECK-NEXT: 1 7 1.00 U v_s_rsq_f32 s1, s0
|
||
|
# CHECK-NEXT: 1 7 1.00 U v_s_sqrt_f32 s2, s1
|
||
|
# CHECK-NEXT: 1 7 1.00 U v_s_exp_f16 s3, s1
|
||
|
# CHECK-NEXT: 1 7 1.00 U v_s_log_f16 s4, s1
|
||
|
# CHECK-NEXT: 1 7 1.00 U v_s_rcp_f16 s5, s2
|
||
|
# CHECK-NEXT: 1 7 1.00 U v_s_rsq_f16 s5, s4
|
||
|
# CHECK-NEXT: 1 7 1.00 U v_s_sqrt_f16 s5, s5
|
||
|
|
||
|
# CHECK: Resources:
|
||
|
# CHECK-NEXT: [0] - HWBranch
|
||
|
# CHECK-NEXT: [1] - HWExport
|
||
|
# CHECK-NEXT: [2] - HWLGKM
|
||
|
# CHECK-NEXT: [3] - HWRC
|
||
|
# CHECK-NEXT: [4] - HWSALU
|
||
|
# CHECK-NEXT: [5] - HWVALU
|
||
|
# CHECK-NEXT: [6] - HWVMEM
|
||
|
|
||
|
# CHECK: Resource pressure per iteration:
|
||
|
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6]
|
||
|
# CHECK-NEXT: - - - 10.00 - 10.00 -
|
||
|
|
||
|
# CHECK: Resource pressure by instruction:
|
||
|
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] Instructions:
|
||
|
# CHECK-NEXT: - - - 1.00 - 1.00 - v_s_exp_f32 s0, s0
|
||
|
# CHECK-NEXT: - - - 1.00 - 1.00 - v_s_log_f32 s0, s0
|
||
|
# CHECK-NEXT: - - - 1.00 - 1.00 - v_s_rcp_f32 s1, s1
|
||
|
# CHECK-NEXT: - - - 1.00 - 1.00 - v_s_rsq_f32 s1, s0
|
||
|
# CHECK-NEXT: - - - 1.00 - 1.00 - v_s_sqrt_f32 s2, s1
|
||
|
# CHECK-NEXT: - - - 1.00 - 1.00 - v_s_exp_f16 s3, s1
|
||
|
# CHECK-NEXT: - - - 1.00 - 1.00 - v_s_log_f16 s4, s1
|
||
|
# CHECK-NEXT: - - - 1.00 - 1.00 - v_s_rcp_f16 s5, s2
|
||
|
# CHECK-NEXT: - - - 1.00 - 1.00 - v_s_rsq_f16 s5, s4
|
||
|
# CHECK-NEXT: - - - 1.00 - 1.00 - v_s_sqrt_f16 s5, s5
|
||
|
|
||
|
# CHECK: Timeline view:
|
||
|
# CHECK-NEXT: 0123456789 0123456789
|
||
|
# CHECK-NEXT: Index 0123456789 0123456789 01234
|
||
|
|
||
|
# CHECK: [0,0] DeeeeeeE . . . . . . . . v_s_exp_f32 s0, s0
|
||
|
# CHECK-NEXT: [0,1] . . DeeeeeeE. . . . . . . v_s_log_f32 s0, s0
|
||
|
# CHECK-NEXT: [0,2] . . DeeeeeeE . . . . . . v_s_rcp_f32 s1, s1
|
||
|
# CHECK-NEXT: [0,3] . . . DeeeeeeE . . . . . v_s_rsq_f32 s1, s0
|
||
|
# CHECK-NEXT: [0,4] . . . . .DeeeeeeE . . . . v_s_sqrt_f32 s2, s1
|
||
|
# CHECK-NEXT: [0,5] . . . . . DeeeeeeE. . . . v_s_exp_f16 s3, s1
|
||
|
# CHECK-NEXT: [0,6] . . . . . DeeeeeeE . . . v_s_log_f16 s4, s1
|
||
|
# CHECK-NEXT: [0,7] . . . . . . DeeeeeeE . . v_s_rcp_f16 s5, s2
|
||
|
# CHECK-NEXT: [0,8] . . . . . . DeeeeeeE . . v_s_rsq_f16 s5, s4
|
||
|
# CHECK-NEXT: [0,9] . . . . . . . . DeeeeeeE v_s_sqrt_f16 s5, s5
|
||
|
|
||
|
# CHECK: Average Wait times (based on the timeline view):
|
||
|
# CHECK-NEXT: [0]: Executions
|
||
|
# CHECK-NEXT: [1]: Average time spent waiting in a scheduler's queue
|
||
|
# CHECK-NEXT: [2]: Average time spent waiting in a scheduler's queue while ready
|
||
|
# CHECK-NEXT: [3]: Average time elapsed from WB until retire stage
|
||
|
|
||
|
# CHECK: [0] [1] [2] [3]
|
||
|
# CHECK-NEXT: 0. 1 0.0 0.0 0.0 v_s_exp_f32 s0, s0
|
||
|
# CHECK-NEXT: 1. 1 0.0 0.0 0.0 v_s_log_f32 s0, s0
|
||
|
# CHECK-NEXT: 2. 1 0.0 0.0 0.0 v_s_rcp_f32 s1, s1
|
||
|
# CHECK-NEXT: 3. 1 0.0 0.0 0.0 v_s_rsq_f32 s1, s0
|
||
|
# CHECK-NEXT: 4. 1 0.0 0.0 0.0 v_s_sqrt_f32 s2, s1
|
||
|
# CHECK-NEXT: 5. 1 0.0 0.0 0.0 v_s_exp_f16 s3, s1
|
||
|
# CHECK-NEXT: 6. 1 0.0 0.0 0.0 v_s_log_f16 s4, s1
|
||
|
# CHECK-NEXT: 7. 1 0.0 0.0 0.0 v_s_rcp_f16 s5, s2
|
||
|
# CHECK-NEXT: 8. 1 0.0 0.0 0.0 v_s_rsq_f16 s5, s4
|
||
|
# CHECK-NEXT: 9. 1 0.0 0.0 0.0 v_s_sqrt_f16 s5, s5
|
||
|
# CHECK-NEXT: 1 0.0 0.0 0.0 <total>
|