126 lines · plain
1# NOTE: Assertions have been autogenerated by utils/update_mca_test_checks.py2# RUN: llvm-mca -mtriple=x86_64-unknown-unknown -mcpu=btver2 -iterations=100 -resource-pressure=false -timeline -timeline-max-iterations=2 < %s | FileCheck %s3 4# In this test, the VDIVPS takes 38 cycles to write to register YMM3. The first5# VADDPS does not depend on the VDIVPS (the WAW dependency is eliminated at6# register renaming stage). So the first VADDPS can be executed in parallel to7# the VDIVPS. That VADDPS also writes to register XMM3, and the upper half of8# YMM3 is implicitly cleared. As a consequence, the definition of YMM3 from the9# VDIVPS is killed, and the subsequent VADDPS instructions don't need to wait10# for the VDIVPS to complete.11# The block reciprocal throughput is limited by the VDIVPS reciprocal throughput12# (which is 38 cycles). The sequence of VADDPS can be executed in parallel on13# the FPA unit; their latency is "hidden" by the long latency of the VDIVPS.14 15vdivps %ymm0, %ymm1, %ymm316vaddps %xmm0, %xmm1, %xmm317vaddps %ymm3, %ymm1, %ymm418vaddps %ymm3, %ymm1, %ymm419vaddps %ymm3, %ymm1, %ymm420vaddps %ymm3, %ymm1, %ymm421vaddps %ymm3, %ymm1, %ymm422vaddps %ymm3, %ymm1, %ymm423vaddps %ymm3, %ymm1, %ymm424vaddps %ymm3, %ymm1, %ymm425vaddps %ymm3, %ymm1, %ymm426vaddps %ymm3, %ymm1, %ymm427vaddps %ymm3, %ymm1, %ymm428vaddps %ymm3, %ymm1, %ymm429vaddps %ymm3, %ymm1, %ymm430vaddps %ymm3, %ymm1, %ymm431vaddps %ymm3, %ymm1, %ymm432vandps %xmm4, %xmm1, %xmm033 34# CHECK: Iterations: 10035# CHECK-NEXT: Instructions: 180036# CHECK-NEXT: Total Cycles: 381137# CHECK-NEXT: Total uOps: 340038 39# CHECK: Dispatch Width: 240# CHECK-NEXT: uOps Per Cycle: 0.8941# CHECK-NEXT: IPC: 0.4742# CHECK-NEXT: Block RThroughput: 38.043 44# CHECK: Instruction Info:45# CHECK-NEXT: [1]: #uOps46# CHECK-NEXT: [2]: Latency47# CHECK-NEXT: [3]: RThroughput48# CHECK-NEXT: [4]: MayLoad49# CHECK-NEXT: [5]: MayStore50# CHECK-NEXT: [6]: HasSideEffects (U)51 52# CHECK: [1] [2] [3] [4] [5] [6] Instructions:53# CHECK-NEXT: 2 38 38.00 vdivps %ymm0, %ymm1, %ymm354# CHECK-NEXT: 1 3 1.00 vaddps %xmm0, %xmm1, %xmm355# CHECK-NEXT: 2 3 2.00 vaddps %ymm3, %ymm1, %ymm456# CHECK-NEXT: 2 3 2.00 vaddps %ymm3, %ymm1, %ymm457# CHECK-NEXT: 2 3 2.00 vaddps %ymm3, %ymm1, %ymm458# CHECK-NEXT: 2 3 2.00 vaddps %ymm3, %ymm1, %ymm459# CHECK-NEXT: 2 3 2.00 vaddps %ymm3, %ymm1, %ymm460# CHECK-NEXT: 2 3 2.00 vaddps %ymm3, %ymm1, %ymm461# CHECK-NEXT: 2 3 2.00 vaddps %ymm3, %ymm1, %ymm462# CHECK-NEXT: 2 3 2.00 vaddps %ymm3, %ymm1, %ymm463# CHECK-NEXT: 2 3 2.00 vaddps %ymm3, %ymm1, %ymm464# CHECK-NEXT: 2 3 2.00 vaddps %ymm3, %ymm1, %ymm465# CHECK-NEXT: 2 3 2.00 vaddps %ymm3, %ymm1, %ymm466# CHECK-NEXT: 2 3 2.00 vaddps %ymm3, %ymm1, %ymm467# CHECK-NEXT: 2 3 2.00 vaddps %ymm3, %ymm1, %ymm468# CHECK-NEXT: 2 3 2.00 vaddps %ymm3, %ymm1, %ymm469# CHECK-NEXT: 2 3 2.00 vaddps %ymm3, %ymm1, %ymm470# CHECK-NEXT: 1 1 0.50 vandps %xmm4, %xmm1, %xmm071 72# CHECK: Timeline view:73# CHECK-NEXT: 0123456789 0123456789 0123456789 012345678974# CHECK-NEXT: Index 0123456789 0123456789 0123456789 012345678975 76# CHECK: [0,0] DeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeER . . . . . . . . vdivps %ymm0, %ymm1, %ymm377# CHECK-NEXT: [0,1] .DeeeE----------------------------------R . . . . . . . . vaddps %xmm0, %xmm1, %xmm378# CHECK-NEXT: [0,2] . D==eeeE--------------------------------R . . . . . . . . vaddps %ymm3, %ymm1, %ymm479# CHECK-NEXT: [0,3] . D===eeeE------------------------------R . . . . . . . . vaddps %ymm3, %ymm1, %ymm480# CHECK-NEXT: [0,4] . D====eeeE-----------------------------R . . . . . . . . vaddps %ymm3, %ymm1, %ymm481# CHECK-NEXT: [0,5] . D=====eeeE---------------------------R . . . . . . . . vaddps %ymm3, %ymm1, %ymm482# CHECK-NEXT: [0,6] . .D======eeeE--------------------------R . . . . . . . . vaddps %ymm3, %ymm1, %ymm483# CHECK-NEXT: [0,7] . . D=======eeeE------------------------R . . . . . . . . vaddps %ymm3, %ymm1, %ymm484# CHECK-NEXT: [0,8] . . D========eeeE-----------------------R. . . . . . . . vaddps %ymm3, %ymm1, %ymm485# CHECK-NEXT: [0,9] . . D=========eeeE---------------------R. . . . . . . . vaddps %ymm3, %ymm1, %ymm486# CHECK-NEXT: [0,10] . . D==========eeeE--------------------R . . . . . . . vaddps %ymm3, %ymm1, %ymm487# CHECK-NEXT: [0,11] . . .D===========eeeE------------------R . . . . . . . vaddps %ymm3, %ymm1, %ymm488# CHECK-NEXT: [0,12] . . . D============eeeE-----------------R . . . . . . . vaddps %ymm3, %ymm1, %ymm489# CHECK-NEXT: [0,13] . . . D=============eeeE---------------R . . . . . . . vaddps %ymm3, %ymm1, %ymm490# CHECK-NEXT: [0,14] . . . D==============eeeE--------------R . . . . . . . vaddps %ymm3, %ymm1, %ymm491# CHECK-NEXT: [0,15] . . . D===============eeeE------------R . . . . . . . vaddps %ymm3, %ymm1, %ymm492# CHECK-NEXT: [0,16] . . . .D================eeeE-----------R . . . . . . . vaddps %ymm3, %ymm1, %ymm493# CHECK-NEXT: [0,17] . . . . D==================eE----------R . . . . . . . vandps %xmm4, %xmm1, %xmm094# CHECK-NEXT: [1,0] . . . . D====================eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeER. vdivps %ymm0, %ymm1, %ymm395# CHECK-NEXT: [1,1] . . . . D=================eeeE-------------------------------------R. vaddps %xmm0, %xmm1, %xmm396# CHECK-NEXT: [1,2] . . . . D===================eeeE-----------------------------------R vaddps %ymm3, %ymm1, %ymm497# CHECK-NEXT: [1,3] . . . . .D====================eeeE---------------------------------R vaddps %ymm3, %ymm1, %ymm498# CHECK-NEXT: Truncated display due to cycle limit99 100# CHECK: Average Wait times (based on the timeline view):101# CHECK-NEXT: [0]: Executions102# CHECK-NEXT: [1]: Average time spent waiting in a scheduler's queue103# CHECK-NEXT: [2]: Average time spent waiting in a scheduler's queue while ready104# CHECK-NEXT: [3]: Average time elapsed from WB until retire stage105 106# CHECK: [0] [1] [2] [3]107# CHECK-NEXT: 0. 2 11.0 1.5 0.0 vdivps %ymm0, %ymm1, %ymm3108# CHECK-NEXT: 1. 2 9.5 0.5 35.5 vaddps %xmm0, %xmm1, %xmm3109# CHECK-NEXT: 2. 2 11.5 0.0 33.5 vaddps %ymm3, %ymm1, %ymm4110# CHECK-NEXT: 3. 2 12.5 2.0 31.5 vaddps %ymm3, %ymm1, %ymm4111# CHECK-NEXT: 4. 2 13.5 4.0 30.5 vaddps %ymm3, %ymm1, %ymm4112# CHECK-NEXT: 5. 2 14.5 6.0 28.5 vaddps %ymm3, %ymm1, %ymm4113# CHECK-NEXT: 6. 2 15.5 7.5 27.5 vaddps %ymm3, %ymm1, %ymm4114# CHECK-NEXT: 7. 2 16.5 9.0 25.5 vaddps %ymm3, %ymm1, %ymm4115# CHECK-NEXT: 8. 2 17.5 10.5 24.5 vaddps %ymm3, %ymm1, %ymm4116# CHECK-NEXT: 9. 2 18.5 12.0 22.5 vaddps %ymm3, %ymm1, %ymm4117# CHECK-NEXT: 10. 2 19.5 13.5 21.5 vaddps %ymm3, %ymm1, %ymm4118# CHECK-NEXT: 11. 2 20.5 15.0 19.5 vaddps %ymm3, %ymm1, %ymm4119# CHECK-NEXT: 12. 2 21.5 16.5 18.5 vaddps %ymm3, %ymm1, %ymm4120# CHECK-NEXT: 13. 2 22.5 18.0 16.5 vaddps %ymm3, %ymm1, %ymm4121# CHECK-NEXT: 14. 2 23.5 19.5 15.5 vaddps %ymm3, %ymm1, %ymm4122# CHECK-NEXT: 15. 2 21.0 21.0 13.5 vaddps %ymm3, %ymm1, %ymm4123# CHECK-NEXT: 16. 2 22.0 22.0 12.5 vaddps %ymm3, %ymm1, %ymm4124# CHECK-NEXT: 17. 2 24.0 0.0 11.5 vandps %xmm4, %xmm1, %xmm0125# CHECK-NEXT: 2 17.5 9.9 21.6 <total>126