[llvm] r313022 - [X86][AVX2] Add additional fp-broadcast/subvector/shuffle scheduling tests
Simon Pilgrim via llvm-commits
llvm-commits at lists.llvm.org
Tue Sep 12 04:17:01 PDT 2017
Author: rksimon
Date: Tue Sep 12 04:17:01 2017
New Revision: 313022
URL: http://llvm.org/viewvc/llvm-project?rev=313022&view=rev
Log:
[X86][AVX2] Add additional fp-broadcast/subvector/shuffle scheduling tests
Modified:
llvm/trunk/test/CodeGen/X86/avx2-schedule.ll
Modified: llvm/trunk/test/CodeGen/X86/avx2-schedule.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/avx2-schedule.ll?rev=313022&r1=313021&r2=313022&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/avx2-schedule.ll (original)
+++ llvm/trunk/test/CodeGen/X86/avx2-schedule.ll Tue Sep 12 04:17:01 2017
@@ -4,6 +4,174 @@
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -print-schedule -mcpu=skylake | FileCheck %s --check-prefix=CHECK --check-prefix=SKYLAKE
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -print-schedule -mcpu=znver1 | FileCheck %s --check-prefix=CHECK --check-prefix=ZNVER1
+define <4 x double> @test_broadcastsd_ymm(<2 x double> %a0) {
+; GENERIC-LABEL: test_broadcastsd_ymm:
+; GENERIC: # BB#0:
+; GENERIC-NEXT: vbroadcastsd %xmm0, %ymm0 # sched: [1:1.00]
+; GENERIC-NEXT: vaddpd %ymm0, %ymm0, %ymm0 # sched: [3:1.00]
+; GENERIC-NEXT: retq # sched: [1:1.00]
+;
+; HASWELL-LABEL: test_broadcastsd_ymm:
+; HASWELL: # BB#0:
+; HASWELL-NEXT: vbroadcastsd %xmm0, %ymm0 # sched: [3:1.00]
+; HASWELL-NEXT: vaddpd %ymm0, %ymm0, %ymm0 # sched: [3:1.00]
+; HASWELL-NEXT: retq # sched: [2:1.00]
+;
+; SKYLAKE-LABEL: test_broadcastsd_ymm:
+; SKYLAKE: # BB#0:
+; SKYLAKE-NEXT: vbroadcastsd %xmm0, %ymm0 # sched: [3:1.00]
+; SKYLAKE-NEXT: vaddpd %ymm0, %ymm0, %ymm0 # sched: [3:1.00]
+; SKYLAKE-NEXT: retq # sched: [2:1.00]
+;
+; ZNVER1-LABEL: test_broadcastsd_ymm:
+; ZNVER1: # BB#0:
+; ZNVER1-NEXT: vbroadcastsd %xmm0, %ymm0 # sched: [100:0.25]
+; ZNVER1-NEXT: vaddpd %ymm0, %ymm0, %ymm0 # sched: [3:1.00]
+; ZNVER1-NEXT: retq # sched: [1:0.50]
+ %1 = shufflevector <2 x double> %a0, <2 x double> undef, <4 x i32> zeroinitializer
+ %2 = fadd <4 x double> %1, %1
+ ret <4 x double> %2
+}
+
+define <4 x float> @test_broadcastss(<4 x float> %a0) {
+; GENERIC-LABEL: test_broadcastss:
+; GENERIC: # BB#0:
+; GENERIC-NEXT: vbroadcastss %xmm0, %xmm0 # sched: [1:1.00]
+; GENERIC-NEXT: vaddps %xmm0, %xmm0, %xmm0 # sched: [3:1.00]
+; GENERIC-NEXT: retq # sched: [1:1.00]
+;
+; HASWELL-LABEL: test_broadcastss:
+; HASWELL: # BB#0:
+; HASWELL-NEXT: vbroadcastss %xmm0, %xmm0 # sched: [1:1.00]
+; HASWELL-NEXT: vaddps %xmm0, %xmm0, %xmm0 # sched: [3:1.00]
+; HASWELL-NEXT: retq # sched: [2:1.00]
+;
+; SKYLAKE-LABEL: test_broadcastss:
+; SKYLAKE: # BB#0:
+; SKYLAKE-NEXT: vbroadcastss %xmm0, %xmm0 # sched: [1:1.00]
+; SKYLAKE-NEXT: vaddps %xmm0, %xmm0, %xmm0 # sched: [3:1.00]
+; SKYLAKE-NEXT: retq # sched: [2:1.00]
+;
+; ZNVER1-LABEL: test_broadcastss:
+; ZNVER1: # BB#0:
+; ZNVER1-NEXT: vbroadcastss %xmm0, %xmm0 # sched: [1:0.50]
+; ZNVER1-NEXT: vaddps %xmm0, %xmm0, %xmm0 # sched: [3:1.00]
+; ZNVER1-NEXT: retq # sched: [1:0.50]
+ %1 = shufflevector <4 x float> %a0, <4 x float> undef, <4 x i32> zeroinitializer
+ %2 = fadd <4 x float> %1, %1
+ ret <4 x float> %2
+}
+
+define <8 x float> @test_broadcastss_ymm(<4 x float> %a0) {
+; GENERIC-LABEL: test_broadcastss_ymm:
+; GENERIC: # BB#0:
+; GENERIC-NEXT: vbroadcastss %xmm0, %ymm0 # sched: [1:1.00]
+; GENERIC-NEXT: vaddps %ymm0, %ymm0, %ymm0 # sched: [3:1.00]
+; GENERIC-NEXT: retq # sched: [1:1.00]
+;
+; HASWELL-LABEL: test_broadcastss_ymm:
+; HASWELL: # BB#0:
+; HASWELL-NEXT: vbroadcastss %xmm0, %ymm0 # sched: [3:1.00]
+; HASWELL-NEXT: vaddps %ymm0, %ymm0, %ymm0 # sched: [3:1.00]
+; HASWELL-NEXT: retq # sched: [2:1.00]
+;
+; SKYLAKE-LABEL: test_broadcastss_ymm:
+; SKYLAKE: # BB#0:
+; SKYLAKE-NEXT: vbroadcastss %xmm0, %ymm0 # sched: [3:1.00]
+; SKYLAKE-NEXT: vaddps %ymm0, %ymm0, %ymm0 # sched: [3:1.00]
+; SKYLAKE-NEXT: retq # sched: [2:1.00]
+;
+; ZNVER1-LABEL: test_broadcastss_ymm:
+; ZNVER1: # BB#0:
+; ZNVER1-NEXT: vbroadcastss %xmm0, %ymm0 # sched: [100:0.25]
+; ZNVER1-NEXT: vaddps %ymm0, %ymm0, %ymm0 # sched: [3:1.00]
+; ZNVER1-NEXT: retq # sched: [1:0.50]
+ %1 = shufflevector <4 x float> %a0, <4 x float> undef, <8 x i32> zeroinitializer
+ %2 = fadd <8 x float> %1, %1
+ ret <8 x float> %2
+}
+
+define <4 x i32> @test_extracti128(<8 x i32> %a0, <8 x i32> %a1, <4 x i32> *%a2) {
+; GENERIC-LABEL: test_extracti128:
+; GENERIC: # BB#0:
+; GENERIC-NEXT: vpaddd %ymm1, %ymm0, %ymm2 # sched: [3:1.00]
+; GENERIC-NEXT: vpsubd %ymm1, %ymm0, %ymm0 # sched: [3:1.00]
+; GENERIC-NEXT: vextracti128 $1, %ymm0, %xmm0 # sched: [1:1.00]
+; GENERIC-NEXT: vextracti128 $1, %ymm2, (%rdi) # sched: [1:1.00]
+; GENERIC-NEXT: vzeroupper
+; GENERIC-NEXT: retq # sched: [1:1.00]
+;
+; HASWELL-LABEL: test_extracti128:
+; HASWELL: # BB#0:
+; HASWELL-NEXT: vpaddd %ymm1, %ymm0, %ymm2 # sched: [1:0.50]
+; HASWELL-NEXT: vpsubd %ymm1, %ymm0, %ymm0 # sched: [1:0.50]
+; HASWELL-NEXT: vextracti128 $1, %ymm0, %xmm0 # sched: [3:1.00]
+; HASWELL-NEXT: vextracti128 $1, %ymm2, (%rdi) # sched: [1:1.00]
+; HASWELL-NEXT: vzeroupper # sched: [4:1.00]
+; HASWELL-NEXT: retq # sched: [2:1.00]
+;
+; SKYLAKE-LABEL: test_extracti128:
+; SKYLAKE: # BB#0:
+; SKYLAKE-NEXT: vpaddd %ymm1, %ymm0, %ymm2 # sched: [1:0.50]
+; SKYLAKE-NEXT: vpsubd %ymm1, %ymm0, %ymm0 # sched: [1:0.50]
+; SKYLAKE-NEXT: vextracti128 $1, %ymm0, %xmm0 # sched: [3:1.00]
+; SKYLAKE-NEXT: vextracti128 $1, %ymm2, (%rdi) # sched: [1:1.00]
+; SKYLAKE-NEXT: vzeroupper # sched: [4:1.00]
+; SKYLAKE-NEXT: retq # sched: [2:1.00]
+;
+; ZNVER1-LABEL: test_extracti128:
+; ZNVER1: # BB#0:
+; ZNVER1-NEXT: vpaddd %ymm1, %ymm0, %ymm2 # sched: [1:0.25]
+; ZNVER1-NEXT: vpsubd %ymm1, %ymm0, %ymm0 # sched: [1:0.25]
+; ZNVER1-NEXT: vextracti128 $1, %ymm0, %xmm0 # sched: [2:0.25]
+; ZNVER1-NEXT: vextracti128 $1, %ymm2, (%rdi) # sched: [1:0.50]
+; ZNVER1-NEXT: vzeroupper # sched: [100:?]
+; ZNVER1-NEXT: retq # sched: [1:0.50]
+ %1 = add <8 x i32> %a0, %a1
+ %2 = sub <8 x i32> %a0, %a1
+ %3 = shufflevector <8 x i32> %1, <8 x i32> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
+ %4 = shufflevector <8 x i32> %2, <8 x i32> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
+ store <4 x i32> %3, <4 x i32> *%a2
+ ret <4 x i32> %4
+}
+
+define <8 x i32> @test_inserti128(<8 x i32> %a0, <4 x i32> %a1, <4 x i32> *%a2) {
+; GENERIC-LABEL: test_inserti128:
+; GENERIC: # BB#0:
+; GENERIC-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm1 # sched: [1:1.00]
+; GENERIC-NEXT: vinserti128 $1, (%rdi), %ymm0, %ymm0 # sched: [5:1.00]
+; GENERIC-NEXT: vpaddd %ymm0, %ymm1, %ymm0 # sched: [3:1.00]
+; GENERIC-NEXT: retq # sched: [1:1.00]
+;
+; HASWELL-LABEL: test_inserti128:
+; HASWELL: # BB#0:
+; HASWELL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm1 # sched: [3:1.00]
+; HASWELL-NEXT: vinserti128 $1, (%rdi), %ymm0, %ymm0 # sched: [1:0.50]
+; HASWELL-NEXT: vpaddd %ymm0, %ymm1, %ymm0 # sched: [1:0.50]
+; HASWELL-NEXT: retq # sched: [2:1.00]
+;
+; SKYLAKE-LABEL: test_inserti128:
+; SKYLAKE: # BB#0:
+; SKYLAKE-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm1 # sched: [3:1.00]
+; SKYLAKE-NEXT: vinserti128 $1, (%rdi), %ymm0, %ymm0 # sched: [1:0.50]
+; SKYLAKE-NEXT: vpaddd %ymm0, %ymm1, %ymm0 # sched: [1:0.50]
+; SKYLAKE-NEXT: retq # sched: [2:1.00]
+;
+; ZNVER1-LABEL: test_inserti128:
+; ZNVER1: # BB#0:
+; ZNVER1-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm1 # sched: [2:0.25]
+; ZNVER1-NEXT: vinserti128 $1, (%rdi), %ymm0, %ymm0 # sched: [9:0.50]
+; ZNVER1-NEXT: vpaddd %ymm0, %ymm1, %ymm0 # sched: [1:0.25]
+; ZNVER1-NEXT: retq # sched: [1:0.50]
+ %1 = shufflevector <4 x i32> %a1, <4 x i32> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+ %2 = shufflevector <8 x i32> %a0, <8 x i32> %1, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
+ %3 = load <4 x i32>, <4 x i32> *%a2, align 16
+ %4 = shufflevector <4 x i32> %3, <4 x i32> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+ %5 = shufflevector <8 x i32> %a0, <8 x i32> %4, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
+ %6 = add <8 x i32> %2, %5
+ ret <8 x i32> %6
+}
+
define <32 x i8> @test_pabsb(<32 x i8> %a0, <32 x i8> *%a1) {
; GENERIC-LABEL: test_pabsb:
; GENERIC: # BB#0:
@@ -304,6 +472,183 @@ define <4 x i64> @test_pandn(<4 x i64> %
ret <4 x i64> %6
}
+define <4 x i64> @test_perm2i128(<4 x i64> %a0, <4 x i64> %a1, <4 x i64> *%a2) {
+; GENERIC-LABEL: test_perm2i128:
+; GENERIC: # BB#0:
+; GENERIC-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3],ymm1[0,1] sched: [1:1.00]
+; GENERIC-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],mem[0,1] sched: [5:1.00]
+; GENERIC-NEXT: vpaddq %ymm0, %ymm1, %ymm0 # sched: [3:1.00]
+; GENERIC-NEXT: retq # sched: [1:1.00]
+;
+; HASWELL-LABEL: test_perm2i128:
+; HASWELL: # BB#0:
+; HASWELL-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3],ymm1[0,1] sched: [3:1.00]
+; HASWELL-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],mem[0,1] sched: [3:1.00]
+; HASWELL-NEXT: vpaddq %ymm0, %ymm1, %ymm0 # sched: [1:0.50]
+; HASWELL-NEXT: retq # sched: [2:1.00]
+;
+; SKYLAKE-LABEL: test_perm2i128:
+; SKYLAKE: # BB#0:
+; SKYLAKE-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3],ymm1[0,1] sched: [3:1.00]
+; SKYLAKE-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],mem[0,1] sched: [3:1.00]
+; SKYLAKE-NEXT: vpaddq %ymm0, %ymm1, %ymm0 # sched: [1:0.50]
+; SKYLAKE-NEXT: retq # sched: [2:1.00]
+;
+; ZNVER1-LABEL: test_perm2i128:
+; ZNVER1: # BB#0:
+; ZNVER1-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm0[2,3],ymm1[0,1] sched: [2:0.25]
+; ZNVER1-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],mem[0,1] sched: [9:0.50]
+; ZNVER1-NEXT: vpaddq %ymm0, %ymm1, %ymm0 # sched: [1:0.25]
+; ZNVER1-NEXT: retq # sched: [1:0.50]
+ %1 = shufflevector <4 x i64> %a0, <4 x i64> %a1, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
+ %2 = load <4 x i64>, <4 x i64> *%a2, align 32
+ %3 = shufflevector <4 x i64> %a0, <4 x i64> %2, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
+ %4 = add <4 x i64> %1, %3
+ ret <4 x i64> %4
+}
+
+define <8 x i32> @test_permd(<8 x i32> %a0, <8 x i32> %a1, <8 x i32> *%a2) {
+; GENERIC-LABEL: test_permd:
+; GENERIC: # BB#0:
+; GENERIC-NEXT: vpermd %ymm1, %ymm0, %ymm1 # sched: [1:1.00]
+; GENERIC-NEXT: vpermd (%rdi), %ymm0, %ymm0 # sched: [5:1.00]
+; GENERIC-NEXT: vpaddd %ymm0, %ymm1, %ymm0 # sched: [3:1.00]
+; GENERIC-NEXT: retq # sched: [1:1.00]
+;
+; HASWELL-LABEL: test_permd:
+; HASWELL: # BB#0:
+; HASWELL-NEXT: vpermd %ymm1, %ymm0, %ymm1 # sched: [3:1.00]
+; HASWELL-NEXT: vpermd (%rdi), %ymm0, %ymm0 # sched: [3:1.00]
+; HASWELL-NEXT: vpaddd %ymm0, %ymm1, %ymm0 # sched: [1:0.50]
+; HASWELL-NEXT: retq # sched: [2:1.00]
+;
+; SKYLAKE-LABEL: test_permd:
+; SKYLAKE: # BB#0:
+; SKYLAKE-NEXT: vpermd %ymm1, %ymm0, %ymm1 # sched: [3:1.00]
+; SKYLAKE-NEXT: vpermd (%rdi), %ymm0, %ymm0 # sched: [3:1.00]
+; SKYLAKE-NEXT: vpaddd %ymm0, %ymm1, %ymm0 # sched: [1:0.50]
+; SKYLAKE-NEXT: retq # sched: [2:1.00]
+;
+; ZNVER1-LABEL: test_permd:
+; ZNVER1: # BB#0:
+; ZNVER1-NEXT: vpermd %ymm1, %ymm0, %ymm1 # sched: [2:0.25]
+; ZNVER1-NEXT: vpermd (%rdi), %ymm0, %ymm0 # sched: [9:0.50]
+; ZNVER1-NEXT: vpaddd %ymm0, %ymm1, %ymm0 # sched: [1:0.25]
+; ZNVER1-NEXT: retq # sched: [1:0.50]
+ %1 = call <8 x i32> @llvm.x86.avx2.permd(<8 x i32> %a1, <8 x i32> %a0)
+ %2 = load <8 x i32>, <8 x i32> *%a2, align 32
+ %3 = call <8 x i32> @llvm.x86.avx2.permd(<8 x i32> %2, <8 x i32> %a0)
+ %4 = add <8 x i32> %1, %3
+ ret <8 x i32> %4
+}
+declare <8 x i32> @llvm.x86.avx2.permd(<8 x i32>, <8 x i32>) nounwind readonly
+
+define <4 x double> @test_permpd(<4 x double> %a0, <4 x double> *%a1) {
+; GENERIC-LABEL: test_permpd:
+; GENERIC: # BB#0:
+; GENERIC-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,2,2,3] sched: [1:1.00]
+; GENERIC-NEXT: vpermpd {{.*#+}} ymm1 = mem[0,2,2,3] sched: [5:1.00]
+; GENERIC-NEXT: vaddpd %ymm1, %ymm0, %ymm0 # sched: [3:1.00]
+; GENERIC-NEXT: retq # sched: [1:1.00]
+;
+; HASWELL-LABEL: test_permpd:
+; HASWELL: # BB#0:
+; HASWELL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,2,2,3] sched: [3:1.00]
+; HASWELL-NEXT: vpermpd {{.*#+}} ymm1 = mem[0,2,2,3] sched: [3:1.00]
+; HASWELL-NEXT: vaddpd %ymm1, %ymm0, %ymm0 # sched: [3:1.00]
+; HASWELL-NEXT: retq # sched: [2:1.00]
+;
+; SKYLAKE-LABEL: test_permpd:
+; SKYLAKE: # BB#0:
+; SKYLAKE-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,2,2,3] sched: [3:1.00]
+; SKYLAKE-NEXT: vpermpd {{.*#+}} ymm1 = mem[0,2,2,3] sched: [3:1.00]
+; SKYLAKE-NEXT: vaddpd %ymm1, %ymm0, %ymm0 # sched: [3:1.00]
+; SKYLAKE-NEXT: retq # sched: [2:1.00]
+;
+; ZNVER1-LABEL: test_permpd:
+; ZNVER1: # BB#0:
+; ZNVER1-NEXT: vpermpd {{.*#+}} ymm1 = mem[0,2,2,3] sched: [107:0.50]
+; ZNVER1-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,2,2,3] sched: [100:0.25]
+; ZNVER1-NEXT: vaddpd %ymm1, %ymm0, %ymm0 # sched: [3:1.00]
+; ZNVER1-NEXT: retq # sched: [1:0.50]
+ %1 = shufflevector <4 x double> %a0, <4 x double> undef, <4 x i32> <i32 3, i32 2, i32 2, i32 3>
+ %2 = load <4 x double>, <4 x double> *%a1, align 32
+ %3 = shufflevector <4 x double> %2, <4 x double> undef, <4 x i32> <i32 0, i32 2, i32 2, i32 3>
+ %4 = fadd <4 x double> %1, %3
+ ret <4 x double> %4
+}
+
+define <8 x float> @test_permps(<8 x i32> %a0, <8 x float> %a1, <8 x float> *%a2) {
+; GENERIC-LABEL: test_permps:
+; GENERIC: # BB#0:
+; GENERIC-NEXT: vpermps %ymm1, %ymm0, %ymm1 # sched: [1:1.00]
+; GENERIC-NEXT: vpermps (%rdi), %ymm0, %ymm0 # sched: [5:1.00]
+; GENERIC-NEXT: vaddps %ymm0, %ymm1, %ymm0 # sched: [3:1.00]
+; GENERIC-NEXT: retq # sched: [1:1.00]
+;
+; HASWELL-LABEL: test_permps:
+; HASWELL: # BB#0:
+; HASWELL-NEXT: vpermps %ymm1, %ymm0, %ymm1 # sched: [3:1.00]
+; HASWELL-NEXT: vpermps (%rdi), %ymm0, %ymm0 # sched: [3:1.00]
+; HASWELL-NEXT: vaddps %ymm0, %ymm1, %ymm0 # sched: [3:1.00]
+; HASWELL-NEXT: retq # sched: [2:1.00]
+;
+; SKYLAKE-LABEL: test_permps:
+; SKYLAKE: # BB#0:
+; SKYLAKE-NEXT: vpermps %ymm1, %ymm0, %ymm1 # sched: [3:1.00]
+; SKYLAKE-NEXT: vpermps (%rdi), %ymm0, %ymm0 # sched: [3:1.00]
+; SKYLAKE-NEXT: vaddps %ymm0, %ymm1, %ymm0 # sched: [3:1.00]
+; SKYLAKE-NEXT: retq # sched: [2:1.00]
+;
+; ZNVER1-LABEL: test_permps:
+; ZNVER1: # BB#0:
+; ZNVER1-NEXT: vpermps %ymm1, %ymm0, %ymm1 # sched: [100:0.25]
+; ZNVER1-NEXT: vpermps (%rdi), %ymm0, %ymm0 # sched: [107:0.50]
+; ZNVER1-NEXT: vaddps %ymm0, %ymm1, %ymm0 # sched: [3:1.00]
+; ZNVER1-NEXT: retq # sched: [1:0.50]
+ %1 = call <8 x float> @llvm.x86.avx2.permps(<8 x float> %a1, <8 x i32> %a0)
+ %2 = load <8 x float>, <8 x float> *%a2, align 32
+ %3 = call <8 x float> @llvm.x86.avx2.permps(<8 x float> %2, <8 x i32> %a0)
+ %4 = fadd <8 x float> %1, %3
+ ret <8 x float> %4
+}
+declare <8 x float> @llvm.x86.avx2.permps(<8 x float>, <8 x i32>) nounwind readonly
+
+define <4 x i64> @test_permq(<4 x i64> %a0, <4 x i64> *%a1) {
+; GENERIC-LABEL: test_permq:
+; GENERIC: # BB#0:
+; GENERIC-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,2,2,3] sched: [1:1.00]
+; GENERIC-NEXT: vpermq {{.*#+}} ymm1 = mem[0,2,2,3] sched: [5:1.00]
+; GENERIC-NEXT: vpaddq %ymm1, %ymm0, %ymm0 # sched: [3:1.00]
+; GENERIC-NEXT: retq # sched: [1:1.00]
+;
+; HASWELL-LABEL: test_permq:
+; HASWELL: # BB#0:
+; HASWELL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,2,2,3] sched: [3:1.00]
+; HASWELL-NEXT: vpermq {{.*#+}} ymm1 = mem[0,2,2,3] sched: [3:1.00]
+; HASWELL-NEXT: vpaddq %ymm1, %ymm0, %ymm0 # sched: [1:0.50]
+; HASWELL-NEXT: retq # sched: [2:1.00]
+;
+; SKYLAKE-LABEL: test_permq:
+; SKYLAKE: # BB#0:
+; SKYLAKE-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,2,2,3] sched: [3:1.00]
+; SKYLAKE-NEXT: vpermq {{.*#+}} ymm1 = mem[0,2,2,3] sched: [3:1.00]
+; SKYLAKE-NEXT: vpaddq %ymm1, %ymm0, %ymm0 # sched: [1:0.50]
+; SKYLAKE-NEXT: retq # sched: [2:1.00]
+;
+; ZNVER1-LABEL: test_permq:
+; ZNVER1: # BB#0:
+; ZNVER1-NEXT: vpermq {{.*#+}} ymm1 = mem[0,2,2,3] sched: [9:0.50]
+; ZNVER1-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,2,2,3] sched: [2:0.25]
+; ZNVER1-NEXT: vpaddq %ymm1, %ymm0, %ymm0 # sched: [1:0.25]
+; ZNVER1-NEXT: retq # sched: [1:0.50]
+ %1 = shufflevector <4 x i64> %a0, <4 x i64> undef, <4 x i32> <i32 3, i32 2, i32 2, i32 3>
+ %2 = load <4 x i64>, <4 x i64> *%a1, align 32
+ %3 = shufflevector <4 x i64> %2, <4 x i64> undef, <4 x i32> <i32 0, i32 2, i32 2, i32 3>
+ %4 = add <4 x i64> %1, %3
+ ret <4 x i64> %4
+}
+
define <8 x i32> @test_pmulld(<8 x i32> %a0, <8 x i32> %a1, <8 x i32> *%a2) {
; GENERIC-LABEL: test_pmulld:
; GENERIC: # BB#0:
@@ -399,6 +744,142 @@ define <4 x i64> @test_por(<4 x i64> %a0
ret <4 x i64> %4
}
+define <32 x i8> @test_pshufb(<32 x i8> %a0, <32 x i8> %a1, <32 x i8> *%a2) {
+; GENERIC-LABEL: test_pshufb:
+; GENERIC: # BB#0:
+; GENERIC-NEXT: vpshufb %ymm1, %ymm0, %ymm0 # sched: [1:1.00]
+; GENERIC-NEXT: vpshufb (%rdi), %ymm0, %ymm0 # sched: [5:1.00]
+; GENERIC-NEXT: retq # sched: [1:1.00]
+;
+; HASWELL-LABEL: test_pshufb:
+; HASWELL: # BB#0:
+; HASWELL-NEXT: vpshufb %ymm1, %ymm0, %ymm0 # sched: [1:1.00]
+; HASWELL-NEXT: vpshufb (%rdi), %ymm0, %ymm0 # sched: [1:1.00]
+; HASWELL-NEXT: retq # sched: [2:1.00]
+;
+; SKYLAKE-LABEL: test_pshufb:
+; SKYLAKE: # BB#0:
+; SKYLAKE-NEXT: vpshufb %ymm1, %ymm0, %ymm0 # sched: [1:1.00]
+; SKYLAKE-NEXT: vpshufb (%rdi), %ymm0, %ymm0 # sched: [1:1.00]
+; SKYLAKE-NEXT: retq # sched: [2:1.00]
+;
+; ZNVER1-LABEL: test_pshufb:
+; ZNVER1: # BB#0:
+; ZNVER1-NEXT: vpshufb %ymm1, %ymm0, %ymm0 # sched: [1:0.25]
+; ZNVER1-NEXT: vpshufb (%rdi), %ymm0, %ymm0 # sched: [8:0.50]
+; ZNVER1-NEXT: retq # sched: [1:0.50]
+ %1 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %a0, <32 x i8> %a1)
+ %2 = load <32 x i8>, <32 x i8> *%a2, align 32
+ %3 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %1, <32 x i8> %2)
+ ret <32 x i8> %3
+}
+declare <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8>, <32 x i8>) nounwind readnone
+
+define <8 x i32> @test_pshufd(<8 x i32> %a0, <8 x i32> *%a1) {
+; GENERIC-LABEL: test_pshufd:
+; GENERIC: # BB#0:
+; GENERIC-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4] sched: [1:1.00]
+; GENERIC-NEXT: vpshufd {{.*#+}} ymm1 = mem[1,0,3,2,5,4,7,6] sched: [5:1.00]
+; GENERIC-NEXT: vpor %ymm1, %ymm0, %ymm0 # sched: [1:1.00]
+; GENERIC-NEXT: retq # sched: [1:1.00]
+;
+; HASWELL-LABEL: test_pshufd:
+; HASWELL: # BB#0:
+; HASWELL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4] sched: [1:1.00]
+; HASWELL-NEXT: vpshufd {{.*#+}} ymm1 = mem[1,0,3,2,5,4,7,6] sched: [1:1.00]
+; HASWELL-NEXT: vpor %ymm1, %ymm0, %ymm0 # sched: [1:0.33]
+; HASWELL-NEXT: retq # sched: [2:1.00]
+;
+; SKYLAKE-LABEL: test_pshufd:
+; SKYLAKE: # BB#0:
+; SKYLAKE-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4] sched: [1:1.00]
+; SKYLAKE-NEXT: vpshufd {{.*#+}} ymm1 = mem[1,0,3,2,5,4,7,6] sched: [1:1.00]
+; SKYLAKE-NEXT: vpor %ymm1, %ymm0, %ymm0 # sched: [1:0.33]
+; SKYLAKE-NEXT: retq # sched: [2:1.00]
+;
+; ZNVER1-LABEL: test_pshufd:
+; ZNVER1: # BB#0:
+; ZNVER1-NEXT: vpshufd {{.*#+}} ymm1 = mem[1,0,3,2,5,4,7,6] sched: [8:0.50]
+; ZNVER1-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4] sched: [1:0.25]
+; ZNVER1-NEXT: vpor %ymm1, %ymm0, %ymm0 # sched: [1:0.25]
+; ZNVER1-NEXT: retq # sched: [1:0.50]
+ %1 = shufflevector <8 x i32> %a0, <8 x i32> undef, <8 x i32> <i32 3, i32 2, i32 1, i32 0, i32 7, i32 6, i32 5, i32 4>
+ %2 = load <8 x i32>, <8 x i32> *%a1, align 32
+ %3 = shufflevector <8 x i32> %2, <8 x i32> undef, <8 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6>
+ %4 = or <8 x i32> %1, %3
+ ret <8 x i32> %4
+}
+
+define <16 x i16> @test_pshufhw(<16 x i16> %a0, <16 x i16> *%a1) {
+; GENERIC-LABEL: test_pshufhw:
+; GENERIC: # BB#0:
+; GENERIC-NEXT: vpshufhw {{.*#+}} ymm0 = ymm0[0,1,2,3,7,6,5,4,8,9,10,11,15,14,13,12] sched: [1:1.00]
+; GENERIC-NEXT: vpshufhw {{.*#+}} ymm1 = mem[0,1,2,3,5,4,7,6,8,9,10,11,13,12,15,14] sched: [5:1.00]
+; GENERIC-NEXT: vpor %ymm1, %ymm0, %ymm0 # sched: [1:1.00]
+; GENERIC-NEXT: retq # sched: [1:1.00]
+;
+; HASWELL-LABEL: test_pshufhw:
+; HASWELL: # BB#0:
+; HASWELL-NEXT: vpshufhw {{.*#+}} ymm0 = ymm0[0,1,2,3,7,6,5,4,8,9,10,11,15,14,13,12] sched: [1:1.00]
+; HASWELL-NEXT: vpshufhw {{.*#+}} ymm1 = mem[0,1,2,3,5,4,7,6,8,9,10,11,13,12,15,14] sched: [1:1.00]
+; HASWELL-NEXT: vpor %ymm1, %ymm0, %ymm0 # sched: [1:0.33]
+; HASWELL-NEXT: retq # sched: [2:1.00]
+;
+; SKYLAKE-LABEL: test_pshufhw:
+; SKYLAKE: # BB#0:
+; SKYLAKE-NEXT: vpshufhw {{.*#+}} ymm0 = ymm0[0,1,2,3,7,6,5,4,8,9,10,11,15,14,13,12] sched: [1:1.00]
+; SKYLAKE-NEXT: vpshufhw {{.*#+}} ymm1 = mem[0,1,2,3,5,4,7,6,8,9,10,11,13,12,15,14] sched: [1:1.00]
+; SKYLAKE-NEXT: vpor %ymm1, %ymm0, %ymm0 # sched: [1:0.33]
+; SKYLAKE-NEXT: retq # sched: [2:1.00]
+;
+; ZNVER1-LABEL: test_pshufhw:
+; ZNVER1: # BB#0:
+; ZNVER1-NEXT: vpshufhw {{.*#+}} ymm1 = mem[0,1,2,3,5,4,7,6,8,9,10,11,13,12,15,14] sched: [8:0.50]
+; ZNVER1-NEXT: vpshufhw {{.*#+}} ymm0 = ymm0[0,1,2,3,7,6,5,4,8,9,10,11,15,14,13,12] sched: [1:0.25]
+; ZNVER1-NEXT: vpor %ymm1, %ymm0, %ymm0 # sched: [1:0.25]
+; ZNVER1-NEXT: retq # sched: [1:0.50]
+ %1 = shufflevector <16 x i16> %a0, <16 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 7, i32 6, i32 5, i32 4, i32 8, i32 9, i32 10, i32 11, i32 15, i32 14, i32 13, i32 12>
+ %2 = load <16 x i16>, <16 x i16> *%a1, align 32
+ %3 = shufflevector <16 x i16> %2, <16 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 5, i32 4, i32 7, i32 6, i32 8, i32 9, i32 10, i32 11, i32 13, i32 12, i32 15, i32 14>
+ %4 = or <16 x i16> %1, %3
+ ret <16 x i16> %4
+}
+
+define <16 x i16> @test_pshuflw(<16 x i16> %a0, <16 x i16> *%a1) {
+; GENERIC-LABEL: test_pshuflw:
+; GENERIC: # BB#0:
+; GENERIC-NEXT: vpshuflw {{.*#+}} ymm0 = ymm0[3,2,1,0,4,5,6,7,11,10,9,8,12,13,14,15] sched: [1:1.00]
+; GENERIC-NEXT: vpshuflw {{.*#+}} ymm1 = mem[1,0,3,2,4,5,6,7,9,8,11,10,12,13,14,15] sched: [5:1.00]
+; GENERIC-NEXT: vpor %ymm1, %ymm0, %ymm0 # sched: [1:1.00]
+; GENERIC-NEXT: retq # sched: [1:1.00]
+;
+; HASWELL-LABEL: test_pshuflw:
+; HASWELL: # BB#0:
+; HASWELL-NEXT: vpshuflw {{.*#+}} ymm0 = ymm0[3,2,1,0,4,5,6,7,11,10,9,8,12,13,14,15] sched: [1:1.00]
+; HASWELL-NEXT: vpshuflw {{.*#+}} ymm1 = mem[1,0,3,2,4,5,6,7,9,8,11,10,12,13,14,15] sched: [1:1.00]
+; HASWELL-NEXT: vpor %ymm1, %ymm0, %ymm0 # sched: [1:0.33]
+; HASWELL-NEXT: retq # sched: [2:1.00]
+;
+; SKYLAKE-LABEL: test_pshuflw:
+; SKYLAKE: # BB#0:
+; SKYLAKE-NEXT: vpshuflw {{.*#+}} ymm0 = ymm0[3,2,1,0,4,5,6,7,11,10,9,8,12,13,14,15] sched: [1:1.00]
+; SKYLAKE-NEXT: vpshuflw {{.*#+}} ymm1 = mem[1,0,3,2,4,5,6,7,9,8,11,10,12,13,14,15] sched: [1:1.00]
+; SKYLAKE-NEXT: vpor %ymm1, %ymm0, %ymm0 # sched: [1:0.33]
+; SKYLAKE-NEXT: retq # sched: [2:1.00]
+;
+; ZNVER1-LABEL: test_pshuflw:
+; ZNVER1: # BB#0:
+; ZNVER1-NEXT: vpshuflw {{.*#+}} ymm1 = mem[1,0,3,2,4,5,6,7,9,8,11,10,12,13,14,15] sched: [8:0.50]
+; ZNVER1-NEXT: vpshuflw {{.*#+}} ymm0 = ymm0[3,2,1,0,4,5,6,7,11,10,9,8,12,13,14,15] sched: [1:0.25]
+; ZNVER1-NEXT: vpor %ymm1, %ymm0, %ymm0 # sched: [1:0.25]
+; ZNVER1-NEXT: retq # sched: [1:0.50]
+ %1 = shufflevector <16 x i16> %a0, <16 x i16> undef, <16 x i32> <i32 3, i32 2, i32 1, i32 0, i32 4, i32 5, i32 6, i32 7, i32 11, i32 10, i32 9, i32 8, i32 12, i32 13, i32 14, i32 15>
+ %2 = load <16 x i16>, <16 x i16> *%a1, align 32
+ %3 = shufflevector <16 x i16> %2, <16 x i16> undef, <16 x i32> <i32 1, i32 0, i32 3, i32 2, i32 4, i32 5, i32 6, i32 7, i32 9, i32 8, i32 11, i32 10, i32 12, i32 13, i32 14, i32 15>
+ %4 = or <16 x i16> %1, %3
+ ret <16 x i16> %4
+}
+
define <32 x i8> @test_psubb(<32 x i8> %a0, <32 x i8> %a1, <32 x i8> *%a2) {
; GENERIC-LABEL: test_psubb:
; GENERIC: # BB#0:
More information about the llvm-commits
mailing list