[X86][PCLMUL] Add scheduling latency/throughput test for PCLMULQDQ instruction

Added it to the SSE42 tests as targets seem to always have both

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@311166 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/test/CodeGen/X86/sse42-schedule.ll b/test/CodeGen/X86/sse42-schedule.ll
index 9bf89a9a69efb6c586d50d1e3eb8b87dfa2b6734..d174ac3e534cf9bf4cb6b86ed4d172781f84cf30 100644 (file)
--- a/test/CodeGen/X86/sse42-schedule.ll
+++ b/test/CodeGen/X86/sse42-schedule.ll
@@ -1,5 +1,5 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc < %s -mtriple=x86_64-unknown-unknown -print-schedule -mcpu=x86-64 -mattr=+sse4.2 | FileCheck %s --check-prefix=CHECK --check-prefix=GENERIC
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -print-schedule -mcpu=x86-64 -mattr=+sse4.2,+pclmul | FileCheck %s --check-prefix=CHECK --check-prefix=GENERIC
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -print-schedule -mcpu=slm | FileCheck %s --check-prefix=CHECK --check-prefix=SLM
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -print-schedule -mcpu=sandybridge | FileCheck %s --check-prefix=CHECK --check-prefix=SANDY
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -print-schedule -mcpu=ivybridge | FileCheck %s --check-prefix=CHECK --check-prefix=SANDY
@@ -554,3 +554,46 @@ define <2 x i64> @test_pcmpgtq(<2 x i64> %a0, <2 x i64> %a1, <2 x i64> *%a2) {
   %5 = sext <2 x i1> %4 to <2 x i64>
   ret <2 x i64> %5
 }
+
+define <2 x i64> @test_pclmulqdq(<2 x i64> %a0, <2 x i64> %a1, <2 x i64> *%a2) {
+; GENERIC-LABEL: test_pclmulqdq:
+; GENERIC:       # BB#0:
+; GENERIC-NEXT:    pclmulqdq $0, %xmm1, %xmm0 # sched: [14:6.00]
+; GENERIC-NEXT:    pclmulqdq $0, (%rdi), %xmm0 # sched: [14:5.67]
+; GENERIC-NEXT:    retq # sched: [1:1.00]
+;
+; SLM-LABEL: test_pclmulqdq:
+; SLM:       # BB#0:
+; SLM-NEXT:    pclmulqdq $0, %xmm1, %xmm0 # sched: [10:10.00]
+; SLM-NEXT:    pclmulqdq $0, (%rdi), %xmm0 # sched: [10:10.00]
+; SLM-NEXT:    retq # sched: [4:1.00]
+;
+; SANDY-LABEL: test_pclmulqdq:
+; SANDY:       # BB#0:
+; SANDY-NEXT:    vpclmulqdq $0, %xmm1, %xmm0, %xmm0 # sched: [14:6.00]
+; SANDY-NEXT:    vpclmulqdq $0, (%rdi), %xmm0, %xmm0 # sched: [14:5.67]
+; SANDY-NEXT:    retq # sched: [1:1.00]
+;
+; HASWELL-LABEL: test_pclmulqdq:
+; HASWELL:       # BB#0:
+; HASWELL-NEXT:    vpclmulqdq $0, %xmm1, %xmm0, %xmm0 # sched: [7:2.00]
+; HASWELL-NEXT:    vpclmulqdq $0, (%rdi), %xmm0, %xmm0 # sched: [7:2.00]
+; HASWELL-NEXT:    retq # sched: [1:1.00]
+;
+; BTVER2-LABEL: test_pclmulqdq:
+; BTVER2:       # BB#0:
+; BTVER2-NEXT:    vpclmulqdq $0, %xmm1, %xmm0, %xmm0 # sched: [2:1.00]
+; BTVER2-NEXT:    vpclmulqdq $0, (%rdi), %xmm0, %xmm0 # sched: [7:1.00]
+; BTVER2-NEXT:    retq # sched: [4:1.00]
+;
+; ZNVER1-LABEL: test_pclmulqdq:
+; ZNVER1:       # BB#0:
+; ZNVER1-NEXT:    vpclmulqdq $0, %xmm1, %xmm0, %xmm0 # sched: [100:?]
+; ZNVER1-NEXT:    vpclmulqdq $0, (%rdi), %xmm0, %xmm0 # sched: [100:?]
+; ZNVER1-NEXT:    retq # sched: [5:0.50]
+  %1 = load <2 x i64>, <2 x i64> *%a2, align 16
+  %2 = call <2 x i64> @llvm.x86.pclmulqdq(<2 x i64> %a0, <2 x i64> %a1, i8 0)
+  %3 = call <2 x i64> @llvm.x86.pclmulqdq(<2 x i64> %1, <2 x i64> %2, i8 0)
+  ret <2 x i64> %3
+}
+declare <2 x i64> @llvm.x86.pclmulqdq(<2 x i64>, <2 x i64>, i8)