From f38dcbecb643e30931b56bbcf37254477eac3977 Mon Sep 17 00:00:00 2001 From: Roman Lebedev Date: Fri, 14 May 2021 13:37:22 +0300 Subject: [PATCH] [X86] AMD Zen 3: same-reg SSE XMM ANDNPS is a 1-cycle(!) dep-breaking zero-idiom Same as SSE XMM XORPS/XORPD, it is not zero-cycle, even though it breaks the deps. As confirmed by the exegesis measurements, and ref docs. --- llvm/lib/Target/X86/X86ScheduleZnver3.td | 3 +- .../llvm-mca/X86/Znver3/zero-idioms-sse-xmm.s | 34 +++++++++++----------- 2 files changed, 19 insertions(+), 18 deletions(-) diff --git a/llvm/lib/Target/X86/X86ScheduleZnver3.td b/llvm/lib/Target/X86/X86ScheduleZnver3.td index 0116163..96234e0 100644 --- a/llvm/lib/Target/X86/X86ScheduleZnver3.td +++ b/llvm/lib/Target/X86/X86ScheduleZnver3.td @@ -1551,7 +1551,8 @@ def : IsZeroIdiomFunction<[ SUB64rr, SUB64rr_REV ], ZeroIdiomPredicate>, // SSE XMM Zero-idioms. - DepBreakingClass<[ XORPSrr, XORPDrr ], ZeroIdiomPredicate>, + DepBreakingClass<[ XORPSrr, XORPDrr, + ANDNPSrr ], ZeroIdiomPredicate>, // AVX XMM Zero-idioms. DepBreakingClass<[ VXORPSrr, VXORPDrr ], ZeroIdiomPredicate>, diff --git a/llvm/test/tools/llvm-mca/X86/Znver3/zero-idioms-sse-xmm.s b/llvm/test/tools/llvm-mca/X86/Znver3/zero-idioms-sse-xmm.s index 835e6c2..cf20c92 100644 --- a/llvm/test/tools/llvm-mca/X86/Znver3/zero-idioms-sse-xmm.s +++ b/llvm/test/tools/llvm-mca/X86/Znver3/zero-idioms-sse-xmm.s @@ -202,12 +202,12 @@ andnps %xmm0, %xmm1 # CHECK: Iterations: 10000 # CHECK-NEXT: Instructions: 20000 -# CHECK-NEXT: Total Cycles: 20003 +# CHECK-NEXT: Total Cycles: 5004 # CHECK-NEXT: Total uOps: 20000 # CHECK: Dispatch Width: 6 -# CHECK-NEXT: uOps Per Cycle: 1.00 -# CHECK-NEXT: IPC: 1.00 +# CHECK-NEXT: uOps Per Cycle: 4.00 +# CHECK-NEXT: IPC: 4.00 # CHECK-NEXT: Block RThroughput: 0.5 # CHECK: Instruction Info: @@ -223,13 +223,13 @@ andnps %xmm0, %xmm1 # CHECK-NEXT: 1 1 0.25 andnps %xmm0, %xmm1 # CHECK: Register File statistics: -# CHECK-NEXT: Total number of mappings created: 20000 -# CHECK-NEXT: Max number of mappings used: 66 +# CHECK-NEXT: Total number of mappings created: 10000 +# CHECK-NEXT: Max number of mappings used: 37 # CHECK: * Register File #1 -- Zn3FpPRF: # CHECK-NEXT: Number of physical registers: 160 -# CHECK-NEXT: Total number of mappings created: 20000 -# CHECK-NEXT: Max number of mappings used: 66 +# CHECK-NEXT: Total number of mappings created: 10000 +# CHECK-NEXT: Max number of mappings used: 37 # CHECK: * Register File #2 -- Zn3IntegerPRF: # CHECK-NEXT: Number of physical registers: 192 @@ -267,16 +267,16 @@ andnps %xmm0, %xmm1 # CHECK: Resource pressure by instruction: # CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12.0] [12.1] [13] [14.0] [14.1] [14.2] [15.0] [15.1] [15.2] [16.0] [16.1] Instructions: -# CHECK-NEXT: - - - - - - - - - 0.50 - 0.50 - - - - - - - - - - - andnps %xmm1, %xmm1 -# CHECK-NEXT: - - - - - - - - 0.50 - 0.50 - - - - - - - - - - - - andnps %xmm0, %xmm1 +# CHECK-NEXT: - - - - - - - - - 0.50 0.25 0.25 - - - - - - - - - - - andnps %xmm1, %xmm1 +# CHECK-NEXT: - - - - - - - - 0.50 - 0.25 0.25 - - - - - - - - - - - andnps %xmm0, %xmm1 # CHECK: Timeline view: -# CHECK-NEXT: Index 0123456 +# CHECK-NEXT: Index 01234 -# CHECK: [0,0] DeER .. andnps %xmm1, %xmm1 -# CHECK-NEXT: [0,1] D=eER.. andnps %xmm0, %xmm1 -# CHECK-NEXT: [1,0] D==eER. andnps %xmm1, %xmm1 -# CHECK-NEXT: [1,1] D===eER andnps %xmm0, %xmm1 +# CHECK: [0,0] DeER. andnps %xmm1, %xmm1 +# CHECK-NEXT: [0,1] D=eER andnps %xmm0, %xmm1 +# CHECK-NEXT: [1,0] DeE-R andnps %xmm1, %xmm1 +# CHECK-NEXT: [1,1] D=eER andnps %xmm0, %xmm1 # CHECK: Average Wait times (based on the timeline view): # CHECK-NEXT: [0]: Executions @@ -285,6 +285,6 @@ andnps %xmm0, %xmm1 # CHECK-NEXT: [3]: Average time elapsed from WB until retire stage # CHECK: [0] [1] [2] [3] -# CHECK-NEXT: 0. 2 2.0 0.5 0.0 andnps %xmm1, %xmm1 -# CHECK-NEXT: 1. 2 3.0 0.0 0.0 andnps %xmm0, %xmm1 -# CHECK-NEXT: 2 2.5 0.3 0.0 +# CHECK-NEXT: 0. 2 1.0 1.0 0.5 andnps %xmm1, %xmm1 +# CHECK-NEXT: 1. 2 2.0 0.0 0.0 andnps %xmm0, %xmm1 +# CHECK-NEXT: 2 1.5 0.5 0.3 -- 2.7.4