104 lines · plain
1# NOTE: Assertions have been autogenerated by utils/update_mca_test_checks.py2# RUN: llvm-mca -mtriple=x86_64-unknown-unknown -mcpu=bdver2 -timeline -timeline-max-iterations=3 -iterations=1500 < %s | FileCheck %s3 4# perf stat reports an IPC of 2.00 for this block of code.5 6# All of the vector packed compares from this test are dependency breaking7# instructions. That means, there is no RAW dependency between any of the8# instructions, and the code can be fully parallelized in hardware.9 10vpcmpeqb %xmm0, %xmm0, %xmm111vpcmpeqw %xmm1, %xmm1, %xmm212vpcmpeqd %xmm2, %xmm2, %xmm313vpcmpeqq %xmm3, %xmm3, %xmm014 15# CHECK: Iterations: 150016# CHECK-NEXT: Instructions: 600017# CHECK-NEXT: Total Cycles: 300518# CHECK-NEXT: Total uOps: 600019 20# CHECK: Dispatch Width: 421# CHECK-NEXT: uOps Per Cycle: 2.0022# CHECK-NEXT: IPC: 2.0023# CHECK-NEXT: Block RThroughput: 2.024 25# CHECK: Instruction Info:26# CHECK-NEXT: [1]: #uOps27# CHECK-NEXT: [2]: Latency28# CHECK-NEXT: [3]: RThroughput29# CHECK-NEXT: [4]: MayLoad30# CHECK-NEXT: [5]: MayStore31# CHECK-NEXT: [6]: HasSideEffects (U)32 33# CHECK: [1] [2] [3] [4] [5] [6] Instructions:34# CHECK-NEXT: 1 2 0.50 vpcmpeqb %xmm0, %xmm0, %xmm135# CHECK-NEXT: 1 2 0.50 vpcmpeqw %xmm1, %xmm1, %xmm236# CHECK-NEXT: 1 2 0.50 vpcmpeqd %xmm2, %xmm2, %xmm337# CHECK-NEXT: 1 2 0.50 vpcmpeqq %xmm3, %xmm3, %xmm038 39# CHECK: Resources:40# CHECK-NEXT: [0.0] - PdAGLU0141# CHECK-NEXT: [0.1] - PdAGLU0142# CHECK-NEXT: [1] - PdBranch43# CHECK-NEXT: [2] - PdCount44# CHECK-NEXT: [3] - PdDiv45# CHECK-NEXT: [4] - PdEX046# CHECK-NEXT: [5] - PdEX147# CHECK-NEXT: [6] - PdFPCVT48# CHECK-NEXT: [7.0] - PdFPFMA49# CHECK-NEXT: [7.1] - PdFPFMA50# CHECK-NEXT: [8.0] - PdFPMAL51# CHECK-NEXT: [8.1] - PdFPMAL52# CHECK-NEXT: [9] - PdFPMMA53# CHECK-NEXT: [10] - PdFPSTO54# CHECK-NEXT: [11] - PdFPU055# CHECK-NEXT: [12] - PdFPU156# CHECK-NEXT: [13] - PdFPU257# CHECK-NEXT: [14] - PdFPU358# CHECK-NEXT: [15] - PdFPXBR59# CHECK-NEXT: [16.0] - PdLoad60# CHECK-NEXT: [16.1] - PdLoad61# CHECK-NEXT: [17] - PdMul62# CHECK-NEXT: [18] - PdStore63 64# CHECK: Resource pressure per iteration:65# 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]66# CHECK-NEXT: - - - - - - - - - - 2.00 2.00 - - - - 2.00 2.00 - - - - -67 68# CHECK: Resource pressure by instruction:69# 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:70# CHECK-NEXT: - - - - - - - - - - 1.00 - - - - - 1.00 - - - - - - vpcmpeqb %xmm0, %xmm0, %xmm171# CHECK-NEXT: - - - - - - - - - - - 1.00 - - - - - 1.00 - - - - - vpcmpeqw %xmm1, %xmm1, %xmm272# CHECK-NEXT: - - - - - - - - - - 1.00 - - - - - 1.00 - - - - - - vpcmpeqd %xmm2, %xmm2, %xmm373# CHECK-NEXT: - - - - - - - - - - - 1.00 - - - - - 1.00 - - - - - vpcmpeqq %xmm3, %xmm3, %xmm074 75# CHECK: Timeline view:76# CHECK-NEXT: 077# CHECK-NEXT: Index 012345678978 79# CHECK: [0,0] DeeER. . vpcmpeqb %xmm0, %xmm0, %xmm180# CHECK-NEXT: [0,1] D=eeER . vpcmpeqw %xmm1, %xmm1, %xmm281# CHECK-NEXT: [0,2] DeeE-R . vpcmpeqd %xmm2, %xmm2, %xmm382# CHECK-NEXT: [0,3] D==eeER . vpcmpeqq %xmm3, %xmm3, %xmm083# CHECK-NEXT: [1,0] .DeeE-R . vpcmpeqb %xmm0, %xmm0, %xmm184# CHECK-NEXT: [1,1] .D==eeER . vpcmpeqw %xmm1, %xmm1, %xmm285# CHECK-NEXT: [1,2] .D=eeE-R . vpcmpeqd %xmm2, %xmm2, %xmm386# CHECK-NEXT: [1,3] .D===eeER . vpcmpeqq %xmm3, %xmm3, %xmm087# CHECK-NEXT: [2,0] . D=eeE-R . vpcmpeqb %xmm0, %xmm0, %xmm188# CHECK-NEXT: [2,1] . D===eeER. vpcmpeqw %xmm1, %xmm1, %xmm289# CHECK-NEXT: [2,2] . D==eeE-R. vpcmpeqd %xmm2, %xmm2, %xmm390# CHECK-NEXT: [2,3] . D====eeER vpcmpeqq %xmm3, %xmm3, %xmm091 92# CHECK: Average Wait times (based on the timeline view):93# CHECK-NEXT: [0]: Executions94# CHECK-NEXT: [1]: Average time spent waiting in a scheduler's queue95# CHECK-NEXT: [2]: Average time spent waiting in a scheduler's queue while ready96# CHECK-NEXT: [3]: Average time elapsed from WB until retire stage97 98# CHECK: [0] [1] [2] [3]99# CHECK-NEXT: 0. 3 1.3 1.3 0.7 vpcmpeqb %xmm0, %xmm0, %xmm1100# CHECK-NEXT: 1. 3 3.0 3.0 0.0 vpcmpeqw %xmm1, %xmm1, %xmm2101# CHECK-NEXT: 2. 3 2.0 2.0 1.0 vpcmpeqd %xmm2, %xmm2, %xmm3102# CHECK-NEXT: 3. 3 4.0 0.0 0.0 vpcmpeqq %xmm3, %xmm3, %xmm0103# CHECK-NEXT: 3 2.6 1.6 0.4 <total>104