mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-11-23 11:13:28 +01:00
52abc65d5e
Allows narrowInsertExtractVectorBinOp to reduce vector size llvm-svn: 364431
2459 lines
94 KiB
LLVM
2459 lines
94 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse2 | FileCheck %s --check-prefix=ALL --check-prefix=SSE --check-prefix=SSE2
|
|
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse4.1 | FileCheck %s --check-prefix=ALL --check-prefix=SSE --check-prefix=SSE41
|
|
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX1
|
|
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX2
|
|
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f,+avx512bw | FileCheck %s --check-prefix=ALL --check-prefix=AVX512 --check-prefix=AVX512BW
|
|
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f,+avx512bw,+avx512vl | FileCheck %s --check-prefix=ALL --check-prefix=AVX512 --check-prefix=AVX512VL
|
|
|
|
;
|
|
; vXi64
|
|
;
|
|
|
|
define i64 @test_v2i64(<2 x i64> %a0) {
|
|
; SSE2-LABEL: test_v2i64:
|
|
; SSE2: # %bb.0:
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; SSE2-NEXT: movdqa {{.*#+}} xmm2 = [9223372039002259456,9223372039002259456]
|
|
; SSE2-NEXT: movdqa %xmm0, %xmm3
|
|
; SSE2-NEXT: pxor %xmm2, %xmm3
|
|
; SSE2-NEXT: pxor %xmm1, %xmm2
|
|
; SSE2-NEXT: movdqa %xmm3, %xmm4
|
|
; SSE2-NEXT: pcmpgtd %xmm2, %xmm4
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
|
|
; SSE2-NEXT: pcmpeqd %xmm3, %xmm2
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
|
|
; SSE2-NEXT: pand %xmm5, %xmm2
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
|
|
; SSE2-NEXT: por %xmm2, %xmm3
|
|
; SSE2-NEXT: pand %xmm3, %xmm0
|
|
; SSE2-NEXT: pandn %xmm1, %xmm3
|
|
; SSE2-NEXT: por %xmm0, %xmm3
|
|
; SSE2-NEXT: movq %xmm3, %rax
|
|
; SSE2-NEXT: retq
|
|
;
|
|
; SSE41-LABEL: test_v2i64:
|
|
; SSE41: # %bb.0:
|
|
; SSE41-NEXT: movdqa %xmm0, %xmm1
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm2 = xmm0[2,3,0,1]
|
|
; SSE41-NEXT: movdqa {{.*#+}} xmm3 = [9223372039002259456,9223372039002259456]
|
|
; SSE41-NEXT: movdqa %xmm0, %xmm4
|
|
; SSE41-NEXT: pxor %xmm3, %xmm4
|
|
; SSE41-NEXT: pxor %xmm2, %xmm3
|
|
; SSE41-NEXT: movdqa %xmm4, %xmm0
|
|
; SSE41-NEXT: pcmpgtd %xmm3, %xmm0
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm5 = xmm0[0,0,2,2]
|
|
; SSE41-NEXT: pcmpeqd %xmm4, %xmm3
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm3 = xmm3[1,1,3,3]
|
|
; SSE41-NEXT: pand %xmm5, %xmm3
|
|
; SSE41-NEXT: por %xmm3, %xmm0
|
|
; SSE41-NEXT: blendvpd %xmm0, %xmm1, %xmm2
|
|
; SSE41-NEXT: movq %xmm2, %rax
|
|
; SSE41-NEXT: retq
|
|
;
|
|
; AVX-LABEL: test_v2i64:
|
|
; AVX: # %bb.0:
|
|
; AVX-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; AVX-NEXT: vmovdqa {{.*#+}} xmm2 = [9223372036854775808,9223372036854775808]
|
|
; AVX-NEXT: vpxor %xmm2, %xmm0, %xmm3
|
|
; AVX-NEXT: vpxor %xmm2, %xmm1, %xmm2
|
|
; AVX-NEXT: vpcmpgtq %xmm2, %xmm3, %xmm2
|
|
; AVX-NEXT: vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
|
|
; AVX-NEXT: vmovq %xmm0, %rax
|
|
; AVX-NEXT: retq
|
|
;
|
|
; AVX512BW-LABEL: test_v2i64:
|
|
; AVX512BW: # %bb.0:
|
|
; AVX512BW-NEXT: # kill: def $xmm0 killed $xmm0 def $zmm0
|
|
; AVX512BW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; AVX512BW-NEXT: vpmaxuq %zmm1, %zmm0, %zmm0
|
|
; AVX512BW-NEXT: vmovq %xmm0, %rax
|
|
; AVX512BW-NEXT: vzeroupper
|
|
; AVX512BW-NEXT: retq
|
|
;
|
|
; AVX512VL-LABEL: test_v2i64:
|
|
; AVX512VL: # %bb.0:
|
|
; AVX512VL-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; AVX512VL-NEXT: vpmaxuq %xmm1, %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vmovq %xmm0, %rax
|
|
; AVX512VL-NEXT: retq
|
|
%1 = call i64 @llvm.experimental.vector.reduce.umax.v2i64(<2 x i64> %a0)
|
|
ret i64 %1
|
|
}
|
|
|
|
define i64 @test_v4i64(<4 x i64> %a0) {
|
|
; SSE2-LABEL: test_v4i64:
|
|
; SSE2: # %bb.0:
|
|
; SSE2-NEXT: movdqa {{.*#+}} xmm2 = [9223372039002259456,9223372039002259456]
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm3
|
|
; SSE2-NEXT: pxor %xmm2, %xmm3
|
|
; SSE2-NEXT: movdqa %xmm0, %xmm4
|
|
; SSE2-NEXT: pxor %xmm2, %xmm4
|
|
; SSE2-NEXT: movdqa %xmm4, %xmm5
|
|
; SSE2-NEXT: pcmpgtd %xmm3, %xmm5
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm6 = xmm5[0,0,2,2]
|
|
; SSE2-NEXT: pcmpeqd %xmm3, %xmm4
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
|
|
; SSE2-NEXT: pand %xmm6, %xmm3
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm5[1,1,3,3]
|
|
; SSE2-NEXT: por %xmm3, %xmm4
|
|
; SSE2-NEXT: pand %xmm4, %xmm0
|
|
; SSE2-NEXT: pandn %xmm1, %xmm4
|
|
; SSE2-NEXT: por %xmm0, %xmm4
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm4[2,3,0,1]
|
|
; SSE2-NEXT: movdqa %xmm4, %xmm1
|
|
; SSE2-NEXT: pxor %xmm2, %xmm1
|
|
; SSE2-NEXT: pxor %xmm0, %xmm2
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm3
|
|
; SSE2-NEXT: pcmpgtd %xmm2, %xmm3
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm3[0,0,2,2]
|
|
; SSE2-NEXT: pcmpeqd %xmm1, %xmm2
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
|
|
; SSE2-NEXT: pand %xmm5, %xmm1
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm3[1,1,3,3]
|
|
; SSE2-NEXT: por %xmm1, %xmm2
|
|
; SSE2-NEXT: pand %xmm2, %xmm4
|
|
; SSE2-NEXT: pandn %xmm0, %xmm2
|
|
; SSE2-NEXT: por %xmm4, %xmm2
|
|
; SSE2-NEXT: movq %xmm2, %rax
|
|
; SSE2-NEXT: retq
|
|
;
|
|
; SSE41-LABEL: test_v4i64:
|
|
; SSE41: # %bb.0:
|
|
; SSE41-NEXT: movdqa %xmm0, %xmm2
|
|
; SSE41-NEXT: movdqa {{.*#+}} xmm3 = [9223372039002259456,9223372039002259456]
|
|
; SSE41-NEXT: movdqa %xmm1, %xmm4
|
|
; SSE41-NEXT: pxor %xmm3, %xmm4
|
|
; SSE41-NEXT: movdqa %xmm0, %xmm5
|
|
; SSE41-NEXT: pxor %xmm3, %xmm5
|
|
; SSE41-NEXT: movdqa %xmm5, %xmm0
|
|
; SSE41-NEXT: pcmpgtd %xmm4, %xmm0
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm6 = xmm0[0,0,2,2]
|
|
; SSE41-NEXT: pcmpeqd %xmm4, %xmm5
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm4 = xmm5[1,1,3,3]
|
|
; SSE41-NEXT: pand %xmm6, %xmm4
|
|
; SSE41-NEXT: por %xmm4, %xmm0
|
|
; SSE41-NEXT: blendvpd %xmm0, %xmm2, %xmm1
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm2 = xmm1[2,3,0,1]
|
|
; SSE41-NEXT: movdqa %xmm1, %xmm0
|
|
; SSE41-NEXT: pxor %xmm3, %xmm0
|
|
; SSE41-NEXT: pxor %xmm2, %xmm3
|
|
; SSE41-NEXT: movdqa %xmm0, %xmm4
|
|
; SSE41-NEXT: pcmpgtd %xmm3, %xmm4
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
|
|
; SSE41-NEXT: pcmpeqd %xmm0, %xmm3
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm3[1,1,3,3]
|
|
; SSE41-NEXT: pand %xmm5, %xmm0
|
|
; SSE41-NEXT: por %xmm4, %xmm0
|
|
; SSE41-NEXT: blendvpd %xmm0, %xmm1, %xmm2
|
|
; SSE41-NEXT: movq %xmm2, %rax
|
|
; SSE41-NEXT: retq
|
|
;
|
|
; AVX1-LABEL: test_v4i64:
|
|
; AVX1: # %bb.0:
|
|
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
|
|
; AVX1-NEXT: vmovdqa {{.*#+}} xmm2 = [9223372036854775808,9223372036854775808]
|
|
; AVX1-NEXT: vpxor %xmm2, %xmm1, %xmm3
|
|
; AVX1-NEXT: vpxor %xmm2, %xmm0, %xmm4
|
|
; AVX1-NEXT: vpcmpgtq %xmm3, %xmm4, %xmm3
|
|
; AVX1-NEXT: vblendvpd %xmm3, %xmm0, %xmm1, %xmm0
|
|
; AVX1-NEXT: vpermilps {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; AVX1-NEXT: vxorpd %xmm2, %xmm0, %xmm3
|
|
; AVX1-NEXT: vxorpd %xmm2, %xmm1, %xmm2
|
|
; AVX1-NEXT: vpcmpgtq %xmm2, %xmm3, %xmm2
|
|
; AVX1-NEXT: vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
|
|
; AVX1-NEXT: vmovq %xmm0, %rax
|
|
; AVX1-NEXT: vzeroupper
|
|
; AVX1-NEXT: retq
|
|
;
|
|
; AVX2-LABEL: test_v4i64:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX2-NEXT: vpbroadcastq {{.*#+}} ymm2 = [9223372036854775808,9223372036854775808,9223372036854775808,9223372036854775808]
|
|
; AVX2-NEXT: vpxor %xmm2, %xmm1, %xmm3
|
|
; AVX2-NEXT: vpxor %xmm2, %xmm0, %xmm4
|
|
; AVX2-NEXT: vpcmpgtq %xmm3, %xmm4, %xmm3
|
|
; AVX2-NEXT: vblendvpd %xmm3, %xmm0, %xmm1, %xmm0
|
|
; AVX2-NEXT: vpermilps {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; AVX2-NEXT: vxorpd %xmm2, %xmm0, %xmm3
|
|
; AVX2-NEXT: vxorpd %xmm2, %xmm1, %xmm2
|
|
; AVX2-NEXT: vpcmpgtq %xmm2, %xmm3, %xmm2
|
|
; AVX2-NEXT: vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
|
|
; AVX2-NEXT: vmovq %xmm0, %rax
|
|
; AVX2-NEXT: vzeroupper
|
|
; AVX2-NEXT: retq
|
|
;
|
|
; AVX512BW-LABEL: test_v4i64:
|
|
; AVX512BW: # %bb.0:
|
|
; AVX512BW-NEXT: # kill: def $ymm0 killed $ymm0 def $zmm0
|
|
; AVX512BW-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX512BW-NEXT: vpmaxuq %zmm1, %zmm0, %zmm0
|
|
; AVX512BW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; AVX512BW-NEXT: vpmaxuq %zmm1, %zmm0, %zmm0
|
|
; AVX512BW-NEXT: vmovq %xmm0, %rax
|
|
; AVX512BW-NEXT: vzeroupper
|
|
; AVX512BW-NEXT: retq
|
|
;
|
|
; AVX512VL-LABEL: test_v4i64:
|
|
; AVX512VL: # %bb.0:
|
|
; AVX512VL-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX512VL-NEXT: vpmaxuq %xmm1, %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; AVX512VL-NEXT: vpmaxuq %xmm1, %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vmovq %xmm0, %rax
|
|
; AVX512VL-NEXT: vzeroupper
|
|
; AVX512VL-NEXT: retq
|
|
%1 = call i64 @llvm.experimental.vector.reduce.umax.v4i64(<4 x i64> %a0)
|
|
ret i64 %1
|
|
}
|
|
|
|
define i64 @test_v8i64(<8 x i64> %a0) {
|
|
; SSE2-LABEL: test_v8i64:
|
|
; SSE2: # %bb.0:
|
|
; SSE2-NEXT: movdqa {{.*#+}} xmm4 = [9223372039002259456,9223372039002259456]
|
|
; SSE2-NEXT: movdqa %xmm2, %xmm5
|
|
; SSE2-NEXT: pxor %xmm4, %xmm5
|
|
; SSE2-NEXT: movdqa %xmm0, %xmm6
|
|
; SSE2-NEXT: pxor %xmm4, %xmm6
|
|
; SSE2-NEXT: movdqa %xmm6, %xmm7
|
|
; SSE2-NEXT: pcmpgtd %xmm5, %xmm7
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm8 = xmm7[0,0,2,2]
|
|
; SSE2-NEXT: pcmpeqd %xmm5, %xmm6
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm6 = xmm6[1,1,3,3]
|
|
; SSE2-NEXT: pand %xmm8, %xmm6
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm7[1,1,3,3]
|
|
; SSE2-NEXT: por %xmm6, %xmm5
|
|
; SSE2-NEXT: pand %xmm5, %xmm0
|
|
; SSE2-NEXT: pandn %xmm2, %xmm5
|
|
; SSE2-NEXT: por %xmm0, %xmm5
|
|
; SSE2-NEXT: movdqa %xmm3, %xmm0
|
|
; SSE2-NEXT: pxor %xmm4, %xmm0
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm2
|
|
; SSE2-NEXT: pxor %xmm4, %xmm2
|
|
; SSE2-NEXT: movdqa %xmm2, %xmm6
|
|
; SSE2-NEXT: pcmpgtd %xmm0, %xmm6
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm7 = xmm6[0,0,2,2]
|
|
; SSE2-NEXT: pcmpeqd %xmm0, %xmm2
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm2[1,1,3,3]
|
|
; SSE2-NEXT: pand %xmm7, %xmm0
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm6[1,1,3,3]
|
|
; SSE2-NEXT: por %xmm0, %xmm2
|
|
; SSE2-NEXT: pand %xmm2, %xmm1
|
|
; SSE2-NEXT: pandn %xmm3, %xmm2
|
|
; SSE2-NEXT: por %xmm1, %xmm2
|
|
; SSE2-NEXT: movdqa %xmm2, %xmm0
|
|
; SSE2-NEXT: pxor %xmm4, %xmm0
|
|
; SSE2-NEXT: movdqa %xmm5, %xmm1
|
|
; SSE2-NEXT: pxor %xmm4, %xmm1
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm3
|
|
; SSE2-NEXT: pcmpgtd %xmm0, %xmm3
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm6 = xmm3[0,0,2,2]
|
|
; SSE2-NEXT: pcmpeqd %xmm0, %xmm1
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3]
|
|
; SSE2-NEXT: pand %xmm6, %xmm0
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm3[1,1,3,3]
|
|
; SSE2-NEXT: por %xmm0, %xmm1
|
|
; SSE2-NEXT: pand %xmm1, %xmm5
|
|
; SSE2-NEXT: pandn %xmm2, %xmm1
|
|
; SSE2-NEXT: por %xmm5, %xmm1
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm1[2,3,0,1]
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm2
|
|
; SSE2-NEXT: pxor %xmm4, %xmm2
|
|
; SSE2-NEXT: pxor %xmm0, %xmm4
|
|
; SSE2-NEXT: movdqa %xmm2, %xmm3
|
|
; SSE2-NEXT: pcmpgtd %xmm4, %xmm3
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm3[0,0,2,2]
|
|
; SSE2-NEXT: pcmpeqd %xmm2, %xmm4
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm4[1,1,3,3]
|
|
; SSE2-NEXT: pand %xmm5, %xmm2
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm3[1,1,3,3]
|
|
; SSE2-NEXT: por %xmm2, %xmm3
|
|
; SSE2-NEXT: pand %xmm3, %xmm1
|
|
; SSE2-NEXT: pandn %xmm0, %xmm3
|
|
; SSE2-NEXT: por %xmm1, %xmm3
|
|
; SSE2-NEXT: movq %xmm3, %rax
|
|
; SSE2-NEXT: retq
|
|
;
|
|
; SSE41-LABEL: test_v8i64:
|
|
; SSE41: # %bb.0:
|
|
; SSE41-NEXT: movdqa %xmm0, %xmm4
|
|
; SSE41-NEXT: movdqa {{.*#+}} xmm5 = [9223372039002259456,9223372039002259456]
|
|
; SSE41-NEXT: movdqa %xmm2, %xmm6
|
|
; SSE41-NEXT: pxor %xmm5, %xmm6
|
|
; SSE41-NEXT: movdqa %xmm0, %xmm7
|
|
; SSE41-NEXT: pxor %xmm5, %xmm7
|
|
; SSE41-NEXT: movdqa %xmm7, %xmm0
|
|
; SSE41-NEXT: pcmpgtd %xmm6, %xmm0
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm8 = xmm0[0,0,2,2]
|
|
; SSE41-NEXT: pcmpeqd %xmm6, %xmm7
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm6 = xmm7[1,1,3,3]
|
|
; SSE41-NEXT: pand %xmm8, %xmm6
|
|
; SSE41-NEXT: por %xmm6, %xmm0
|
|
; SSE41-NEXT: blendvpd %xmm0, %xmm4, %xmm2
|
|
; SSE41-NEXT: movdqa %xmm3, %xmm0
|
|
; SSE41-NEXT: pxor %xmm5, %xmm0
|
|
; SSE41-NEXT: movdqa %xmm1, %xmm4
|
|
; SSE41-NEXT: pxor %xmm5, %xmm4
|
|
; SSE41-NEXT: movdqa %xmm4, %xmm6
|
|
; SSE41-NEXT: pcmpgtd %xmm0, %xmm6
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm7 = xmm6[0,0,2,2]
|
|
; SSE41-NEXT: pcmpeqd %xmm0, %xmm4
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm4[1,1,3,3]
|
|
; SSE41-NEXT: pand %xmm7, %xmm0
|
|
; SSE41-NEXT: por %xmm6, %xmm0
|
|
; SSE41-NEXT: blendvpd %xmm0, %xmm1, %xmm3
|
|
; SSE41-NEXT: movapd %xmm3, %xmm0
|
|
; SSE41-NEXT: xorpd %xmm5, %xmm0
|
|
; SSE41-NEXT: movapd %xmm2, %xmm1
|
|
; SSE41-NEXT: xorpd %xmm5, %xmm1
|
|
; SSE41-NEXT: movapd %xmm1, %xmm4
|
|
; SSE41-NEXT: pcmpgtd %xmm0, %xmm4
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm6 = xmm4[0,0,2,2]
|
|
; SSE41-NEXT: pcmpeqd %xmm0, %xmm1
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3]
|
|
; SSE41-NEXT: pand %xmm6, %xmm0
|
|
; SSE41-NEXT: por %xmm4, %xmm0
|
|
; SSE41-NEXT: blendvpd %xmm0, %xmm2, %xmm3
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm3[2,3,0,1]
|
|
; SSE41-NEXT: movdqa %xmm3, %xmm0
|
|
; SSE41-NEXT: pxor %xmm5, %xmm0
|
|
; SSE41-NEXT: pxor %xmm1, %xmm5
|
|
; SSE41-NEXT: movdqa %xmm0, %xmm2
|
|
; SSE41-NEXT: pcmpgtd %xmm5, %xmm2
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm4 = xmm2[0,0,2,2]
|
|
; SSE41-NEXT: pcmpeqd %xmm0, %xmm5
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm5[1,1,3,3]
|
|
; SSE41-NEXT: pand %xmm4, %xmm0
|
|
; SSE41-NEXT: por %xmm2, %xmm0
|
|
; SSE41-NEXT: blendvpd %xmm0, %xmm3, %xmm1
|
|
; SSE41-NEXT: movq %xmm1, %rax
|
|
; SSE41-NEXT: retq
|
|
;
|
|
; AVX1-LABEL: test_v8i64:
|
|
; AVX1: # %bb.0:
|
|
; AVX1-NEXT: vmovdqa {{.*#+}} xmm2 = [9223372036854775808,9223372036854775808]
|
|
; AVX1-NEXT: vpxor %xmm2, %xmm1, %xmm3
|
|
; AVX1-NEXT: vpxor %xmm2, %xmm0, %xmm4
|
|
; AVX1-NEXT: vpcmpgtq %xmm3, %xmm4, %xmm3
|
|
; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm4
|
|
; AVX1-NEXT: vpxor %xmm2, %xmm4, %xmm5
|
|
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm6
|
|
; AVX1-NEXT: vpxor %xmm2, %xmm6, %xmm7
|
|
; AVX1-NEXT: vpcmpgtq %xmm5, %xmm7, %xmm5
|
|
; AVX1-NEXT: vblendvpd %xmm5, %xmm6, %xmm4, %xmm4
|
|
; AVX1-NEXT: vxorpd %xmm2, %xmm4, %xmm5
|
|
; AVX1-NEXT: vblendvpd %xmm3, %xmm0, %xmm1, %xmm0
|
|
; AVX1-NEXT: vxorpd %xmm2, %xmm0, %xmm1
|
|
; AVX1-NEXT: vpcmpgtq %xmm5, %xmm1, %xmm1
|
|
; AVX1-NEXT: vblendvpd %xmm1, %xmm0, %xmm4, %xmm0
|
|
; AVX1-NEXT: vpermilps {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; AVX1-NEXT: vxorpd %xmm2, %xmm0, %xmm3
|
|
; AVX1-NEXT: vxorpd %xmm2, %xmm1, %xmm2
|
|
; AVX1-NEXT: vpcmpgtq %xmm2, %xmm3, %xmm2
|
|
; AVX1-NEXT: vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
|
|
; AVX1-NEXT: vmovq %xmm0, %rax
|
|
; AVX1-NEXT: vzeroupper
|
|
; AVX1-NEXT: retq
|
|
;
|
|
; AVX2-LABEL: test_v8i64:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vpbroadcastq {{.*#+}} ymm2 = [9223372036854775808,9223372036854775808,9223372036854775808,9223372036854775808]
|
|
; AVX2-NEXT: vpxor %ymm2, %ymm1, %ymm3
|
|
; AVX2-NEXT: vpxor %ymm2, %ymm0, %ymm4
|
|
; AVX2-NEXT: vpcmpgtq %ymm3, %ymm4, %ymm3
|
|
; AVX2-NEXT: vblendvpd %ymm3, %ymm0, %ymm1, %ymm0
|
|
; AVX2-NEXT: vextractf128 $1, %ymm0, %xmm1
|
|
; AVX2-NEXT: vxorpd %xmm2, %xmm1, %xmm3
|
|
; AVX2-NEXT: vxorpd %xmm2, %xmm0, %xmm4
|
|
; AVX2-NEXT: vpcmpgtq %xmm3, %xmm4, %xmm3
|
|
; AVX2-NEXT: vblendvpd %xmm3, %xmm0, %xmm1, %xmm0
|
|
; AVX2-NEXT: vpermilps {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; AVX2-NEXT: vxorpd %xmm2, %xmm0, %xmm3
|
|
; AVX2-NEXT: vxorpd %xmm2, %xmm1, %xmm2
|
|
; AVX2-NEXT: vpcmpgtq %xmm2, %xmm3, %xmm2
|
|
; AVX2-NEXT: vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
|
|
; AVX2-NEXT: vmovq %xmm0, %rax
|
|
; AVX2-NEXT: vzeroupper
|
|
; AVX2-NEXT: retq
|
|
;
|
|
; AVX512BW-LABEL: test_v8i64:
|
|
; AVX512BW: # %bb.0:
|
|
; AVX512BW-NEXT: vextracti64x4 $1, %zmm0, %ymm1
|
|
; AVX512BW-NEXT: vpmaxuq %zmm1, %zmm0, %zmm0
|
|
; AVX512BW-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX512BW-NEXT: vpmaxuq %zmm1, %zmm0, %zmm0
|
|
; AVX512BW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; AVX512BW-NEXT: vpmaxuq %zmm1, %zmm0, %zmm0
|
|
; AVX512BW-NEXT: vmovq %xmm0, %rax
|
|
; AVX512BW-NEXT: vzeroupper
|
|
; AVX512BW-NEXT: retq
|
|
;
|
|
; AVX512VL-LABEL: test_v8i64:
|
|
; AVX512VL: # %bb.0:
|
|
; AVX512VL-NEXT: vextracti64x4 $1, %zmm0, %ymm1
|
|
; AVX512VL-NEXT: vpmaxuq %zmm1, %zmm0, %zmm0
|
|
; AVX512VL-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX512VL-NEXT: vpmaxuq %xmm1, %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; AVX512VL-NEXT: vpmaxuq %xmm1, %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vmovq %xmm0, %rax
|
|
; AVX512VL-NEXT: vzeroupper
|
|
; AVX512VL-NEXT: retq
|
|
%1 = call i64 @llvm.experimental.vector.reduce.umax.v8i64(<8 x i64> %a0)
|
|
ret i64 %1
|
|
}
|
|
|
|
define i64 @test_v16i64(<16 x i64> %a0) {
|
|
; SSE2-LABEL: test_v16i64:
|
|
; SSE2: # %bb.0:
|
|
; SSE2-NEXT: movdqa {{.*#+}} xmm8 = [9223372039002259456,9223372039002259456]
|
|
; SSE2-NEXT: movdqa %xmm5, %xmm9
|
|
; SSE2-NEXT: pxor %xmm8, %xmm9
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm10
|
|
; SSE2-NEXT: pxor %xmm8, %xmm10
|
|
; SSE2-NEXT: movdqa %xmm10, %xmm11
|
|
; SSE2-NEXT: pcmpgtd %xmm9, %xmm11
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm12 = xmm11[0,0,2,2]
|
|
; SSE2-NEXT: pcmpeqd %xmm9, %xmm10
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm10 = xmm10[1,1,3,3]
|
|
; SSE2-NEXT: pand %xmm12, %xmm10
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm9 = xmm11[1,1,3,3]
|
|
; SSE2-NEXT: por %xmm10, %xmm9
|
|
; SSE2-NEXT: pand %xmm9, %xmm1
|
|
; SSE2-NEXT: pandn %xmm5, %xmm9
|
|
; SSE2-NEXT: por %xmm1, %xmm9
|
|
; SSE2-NEXT: movdqa %xmm7, %xmm1
|
|
; SSE2-NEXT: pxor %xmm8, %xmm1
|
|
; SSE2-NEXT: movdqa %xmm3, %xmm5
|
|
; SSE2-NEXT: pxor %xmm8, %xmm5
|
|
; SSE2-NEXT: movdqa %xmm5, %xmm10
|
|
; SSE2-NEXT: pcmpgtd %xmm1, %xmm10
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm11 = xmm10[0,0,2,2]
|
|
; SSE2-NEXT: pcmpeqd %xmm1, %xmm5
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3]
|
|
; SSE2-NEXT: pand %xmm11, %xmm5
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm10[1,1,3,3]
|
|
; SSE2-NEXT: por %xmm5, %xmm1
|
|
; SSE2-NEXT: pand %xmm1, %xmm3
|
|
; SSE2-NEXT: pandn %xmm7, %xmm1
|
|
; SSE2-NEXT: por %xmm3, %xmm1
|
|
; SSE2-NEXT: movdqa %xmm4, %xmm3
|
|
; SSE2-NEXT: pxor %xmm8, %xmm3
|
|
; SSE2-NEXT: movdqa %xmm0, %xmm5
|
|
; SSE2-NEXT: pxor %xmm8, %xmm5
|
|
; SSE2-NEXT: movdqa %xmm5, %xmm7
|
|
; SSE2-NEXT: pcmpgtd %xmm3, %xmm7
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm10 = xmm7[0,0,2,2]
|
|
; SSE2-NEXT: pcmpeqd %xmm3, %xmm5
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3]
|
|
; SSE2-NEXT: pand %xmm10, %xmm5
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm7[1,1,3,3]
|
|
; SSE2-NEXT: por %xmm5, %xmm3
|
|
; SSE2-NEXT: pand %xmm3, %xmm0
|
|
; SSE2-NEXT: pandn %xmm4, %xmm3
|
|
; SSE2-NEXT: por %xmm0, %xmm3
|
|
; SSE2-NEXT: movdqa %xmm6, %xmm0
|
|
; SSE2-NEXT: pxor %xmm8, %xmm0
|
|
; SSE2-NEXT: movdqa %xmm2, %xmm4
|
|
; SSE2-NEXT: pxor %xmm8, %xmm4
|
|
; SSE2-NEXT: movdqa %xmm4, %xmm5
|
|
; SSE2-NEXT: pcmpgtd %xmm0, %xmm5
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm7 = xmm5[0,0,2,2]
|
|
; SSE2-NEXT: pcmpeqd %xmm0, %xmm4
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm4[1,1,3,3]
|
|
; SSE2-NEXT: pand %xmm7, %xmm0
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm5[1,1,3,3]
|
|
; SSE2-NEXT: por %xmm0, %xmm4
|
|
; SSE2-NEXT: pand %xmm4, %xmm2
|
|
; SSE2-NEXT: pandn %xmm6, %xmm4
|
|
; SSE2-NEXT: por %xmm2, %xmm4
|
|
; SSE2-NEXT: movdqa %xmm4, %xmm0
|
|
; SSE2-NEXT: pxor %xmm8, %xmm0
|
|
; SSE2-NEXT: movdqa %xmm3, %xmm2
|
|
; SSE2-NEXT: pxor %xmm8, %xmm2
|
|
; SSE2-NEXT: movdqa %xmm2, %xmm5
|
|
; SSE2-NEXT: pcmpgtd %xmm0, %xmm5
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm6 = xmm5[0,0,2,2]
|
|
; SSE2-NEXT: pcmpeqd %xmm0, %xmm2
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
|
|
; SSE2-NEXT: pand %xmm6, %xmm2
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm5[1,1,3,3]
|
|
; SSE2-NEXT: por %xmm2, %xmm0
|
|
; SSE2-NEXT: pand %xmm0, %xmm3
|
|
; SSE2-NEXT: pandn %xmm4, %xmm0
|
|
; SSE2-NEXT: por %xmm3, %xmm0
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm2
|
|
; SSE2-NEXT: pxor %xmm8, %xmm2
|
|
; SSE2-NEXT: movdqa %xmm9, %xmm3
|
|
; SSE2-NEXT: pxor %xmm8, %xmm3
|
|
; SSE2-NEXT: movdqa %xmm3, %xmm4
|
|
; SSE2-NEXT: pcmpgtd %xmm2, %xmm4
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
|
|
; SSE2-NEXT: pcmpeqd %xmm2, %xmm3
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm3[1,1,3,3]
|
|
; SSE2-NEXT: pand %xmm5, %xmm2
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
|
|
; SSE2-NEXT: por %xmm2, %xmm3
|
|
; SSE2-NEXT: pand %xmm3, %xmm9
|
|
; SSE2-NEXT: pandn %xmm1, %xmm3
|
|
; SSE2-NEXT: por %xmm9, %xmm3
|
|
; SSE2-NEXT: movdqa %xmm3, %xmm1
|
|
; SSE2-NEXT: pxor %xmm8, %xmm1
|
|
; SSE2-NEXT: movdqa %xmm0, %xmm2
|
|
; SSE2-NEXT: pxor %xmm8, %xmm2
|
|
; SSE2-NEXT: movdqa %xmm2, %xmm4
|
|
; SSE2-NEXT: pcmpgtd %xmm1, %xmm4
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
|
|
; SSE2-NEXT: pcmpeqd %xmm1, %xmm2
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
|
|
; SSE2-NEXT: pand %xmm5, %xmm1
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm4[1,1,3,3]
|
|
; SSE2-NEXT: por %xmm1, %xmm2
|
|
; SSE2-NEXT: pand %xmm2, %xmm0
|
|
; SSE2-NEXT: pandn %xmm3, %xmm2
|
|
; SSE2-NEXT: por %xmm0, %xmm2
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm2[2,3,0,1]
|
|
; SSE2-NEXT: movdqa %xmm2, %xmm1
|
|
; SSE2-NEXT: pxor %xmm8, %xmm1
|
|
; SSE2-NEXT: pxor %xmm0, %xmm8
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm3
|
|
; SSE2-NEXT: pcmpgtd %xmm8, %xmm3
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2]
|
|
; SSE2-NEXT: pcmpeqd %xmm1, %xmm8
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm8[1,1,3,3]
|
|
; SSE2-NEXT: pand %xmm4, %xmm1
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm3[1,1,3,3]
|
|
; SSE2-NEXT: por %xmm1, %xmm3
|
|
; SSE2-NEXT: pand %xmm3, %xmm2
|
|
; SSE2-NEXT: pandn %xmm0, %xmm3
|
|
; SSE2-NEXT: por %xmm2, %xmm3
|
|
; SSE2-NEXT: movq %xmm3, %rax
|
|
; SSE2-NEXT: retq
|
|
;
|
|
; SSE41-LABEL: test_v16i64:
|
|
; SSE41: # %bb.0:
|
|
; SSE41-NEXT: movdqa %xmm0, %xmm8
|
|
; SSE41-NEXT: movdqa {{.*#+}} xmm9 = [9223372039002259456,9223372039002259456]
|
|
; SSE41-NEXT: movdqa %xmm5, %xmm10
|
|
; SSE41-NEXT: pxor %xmm9, %xmm10
|
|
; SSE41-NEXT: movdqa %xmm1, %xmm0
|
|
; SSE41-NEXT: pxor %xmm9, %xmm0
|
|
; SSE41-NEXT: movdqa %xmm0, %xmm11
|
|
; SSE41-NEXT: pcmpgtd %xmm10, %xmm11
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm12 = xmm11[0,0,2,2]
|
|
; SSE41-NEXT: pcmpeqd %xmm10, %xmm0
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
|
|
; SSE41-NEXT: pand %xmm12, %xmm0
|
|
; SSE41-NEXT: por %xmm11, %xmm0
|
|
; SSE41-NEXT: blendvpd %xmm0, %xmm1, %xmm5
|
|
; SSE41-NEXT: movdqa %xmm7, %xmm0
|
|
; SSE41-NEXT: pxor %xmm9, %xmm0
|
|
; SSE41-NEXT: movdqa %xmm3, %xmm1
|
|
; SSE41-NEXT: pxor %xmm9, %xmm1
|
|
; SSE41-NEXT: movdqa %xmm1, %xmm10
|
|
; SSE41-NEXT: pcmpgtd %xmm0, %xmm10
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm11 = xmm10[0,0,2,2]
|
|
; SSE41-NEXT: pcmpeqd %xmm0, %xmm1
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3]
|
|
; SSE41-NEXT: pand %xmm11, %xmm0
|
|
; SSE41-NEXT: por %xmm10, %xmm0
|
|
; SSE41-NEXT: blendvpd %xmm0, %xmm3, %xmm7
|
|
; SSE41-NEXT: movdqa %xmm4, %xmm1
|
|
; SSE41-NEXT: pxor %xmm9, %xmm1
|
|
; SSE41-NEXT: movdqa %xmm8, %xmm3
|
|
; SSE41-NEXT: pxor %xmm9, %xmm3
|
|
; SSE41-NEXT: movdqa %xmm3, %xmm0
|
|
; SSE41-NEXT: pcmpgtd %xmm1, %xmm0
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm10 = xmm0[0,0,2,2]
|
|
; SSE41-NEXT: pcmpeqd %xmm1, %xmm3
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm3[1,1,3,3]
|
|
; SSE41-NEXT: pand %xmm10, %xmm1
|
|
; SSE41-NEXT: por %xmm1, %xmm0
|
|
; SSE41-NEXT: blendvpd %xmm0, %xmm8, %xmm4
|
|
; SSE41-NEXT: movdqa %xmm6, %xmm0
|
|
; SSE41-NEXT: pxor %xmm9, %xmm0
|
|
; SSE41-NEXT: movdqa %xmm2, %xmm1
|
|
; SSE41-NEXT: pxor %xmm9, %xmm1
|
|
; SSE41-NEXT: movdqa %xmm1, %xmm3
|
|
; SSE41-NEXT: pcmpgtd %xmm0, %xmm3
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm8 = xmm3[0,0,2,2]
|
|
; SSE41-NEXT: pcmpeqd %xmm0, %xmm1
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3]
|
|
; SSE41-NEXT: pand %xmm8, %xmm0
|
|
; SSE41-NEXT: por %xmm3, %xmm0
|
|
; SSE41-NEXT: blendvpd %xmm0, %xmm2, %xmm6
|
|
; SSE41-NEXT: movapd %xmm6, %xmm0
|
|
; SSE41-NEXT: xorpd %xmm9, %xmm0
|
|
; SSE41-NEXT: movapd %xmm4, %xmm1
|
|
; SSE41-NEXT: xorpd %xmm9, %xmm1
|
|
; SSE41-NEXT: movapd %xmm1, %xmm2
|
|
; SSE41-NEXT: pcmpgtd %xmm0, %xmm2
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm3 = xmm2[0,0,2,2]
|
|
; SSE41-NEXT: pcmpeqd %xmm0, %xmm1
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3]
|
|
; SSE41-NEXT: pand %xmm3, %xmm0
|
|
; SSE41-NEXT: por %xmm2, %xmm0
|
|
; SSE41-NEXT: blendvpd %xmm0, %xmm4, %xmm6
|
|
; SSE41-NEXT: movapd %xmm7, %xmm0
|
|
; SSE41-NEXT: xorpd %xmm9, %xmm0
|
|
; SSE41-NEXT: movapd %xmm5, %xmm1
|
|
; SSE41-NEXT: xorpd %xmm9, %xmm1
|
|
; SSE41-NEXT: movapd %xmm1, %xmm2
|
|
; SSE41-NEXT: pcmpgtd %xmm0, %xmm2
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm3 = xmm2[0,0,2,2]
|
|
; SSE41-NEXT: pcmpeqd %xmm0, %xmm1
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3]
|
|
; SSE41-NEXT: pand %xmm3, %xmm0
|
|
; SSE41-NEXT: por %xmm2, %xmm0
|
|
; SSE41-NEXT: blendvpd %xmm0, %xmm5, %xmm7
|
|
; SSE41-NEXT: movapd %xmm7, %xmm0
|
|
; SSE41-NEXT: xorpd %xmm9, %xmm0
|
|
; SSE41-NEXT: movapd %xmm6, %xmm1
|
|
; SSE41-NEXT: xorpd %xmm9, %xmm1
|
|
; SSE41-NEXT: movapd %xmm1, %xmm2
|
|
; SSE41-NEXT: pcmpgtd %xmm0, %xmm2
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm3 = xmm2[0,0,2,2]
|
|
; SSE41-NEXT: pcmpeqd %xmm0, %xmm1
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3]
|
|
; SSE41-NEXT: pand %xmm3, %xmm0
|
|
; SSE41-NEXT: por %xmm2, %xmm0
|
|
; SSE41-NEXT: blendvpd %xmm0, %xmm6, %xmm7
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm7[2,3,0,1]
|
|
; SSE41-NEXT: movdqa %xmm7, %xmm0
|
|
; SSE41-NEXT: pxor %xmm9, %xmm0
|
|
; SSE41-NEXT: pxor %xmm1, %xmm9
|
|
; SSE41-NEXT: movdqa %xmm0, %xmm2
|
|
; SSE41-NEXT: pcmpgtd %xmm9, %xmm2
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm3 = xmm2[0,0,2,2]
|
|
; SSE41-NEXT: pcmpeqd %xmm0, %xmm9
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm9[1,1,3,3]
|
|
; SSE41-NEXT: pand %xmm3, %xmm0
|
|
; SSE41-NEXT: por %xmm2, %xmm0
|
|
; SSE41-NEXT: blendvpd %xmm0, %xmm7, %xmm1
|
|
; SSE41-NEXT: movq %xmm1, %rax
|
|
; SSE41-NEXT: retq
|
|
;
|
|
; AVX1-LABEL: test_v16i64:
|
|
; AVX1: # %bb.0:
|
|
; AVX1-NEXT: vextractf128 $1, %ymm2, %xmm8
|
|
; AVX1-NEXT: vmovdqa {{.*#+}} xmm4 = [9223372036854775808,9223372036854775808]
|
|
; AVX1-NEXT: vpxor %xmm4, %xmm8, %xmm6
|
|
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm10
|
|
; AVX1-NEXT: vpxor %xmm4, %xmm10, %xmm5
|
|
; AVX1-NEXT: vpcmpgtq %xmm6, %xmm5, %xmm9
|
|
; AVX1-NEXT: vextractf128 $1, %ymm3, %xmm11
|
|
; AVX1-NEXT: vpxor %xmm4, %xmm11, %xmm5
|
|
; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm7
|
|
; AVX1-NEXT: vpxor %xmm4, %xmm7, %xmm6
|
|
; AVX1-NEXT: vpcmpgtq %xmm5, %xmm6, %xmm12
|
|
; AVX1-NEXT: vpxor %xmm4, %xmm2, %xmm6
|
|
; AVX1-NEXT: vpxor %xmm4, %xmm0, %xmm5
|
|
; AVX1-NEXT: vpcmpgtq %xmm6, %xmm5, %xmm13
|
|
; AVX1-NEXT: vpxor %xmm4, %xmm3, %xmm6
|
|
; AVX1-NEXT: vpxor %xmm4, %xmm1, %xmm5
|
|
; AVX1-NEXT: vpcmpgtq %xmm6, %xmm5, %xmm5
|
|
; AVX1-NEXT: vblendvpd %xmm5, %xmm1, %xmm3, %xmm1
|
|
; AVX1-NEXT: vxorpd %xmm4, %xmm1, %xmm3
|
|
; AVX1-NEXT: vblendvpd %xmm13, %xmm0, %xmm2, %xmm0
|
|
; AVX1-NEXT: vxorpd %xmm4, %xmm0, %xmm2
|
|
; AVX1-NEXT: vpcmpgtq %xmm3, %xmm2, %xmm2
|
|
; AVX1-NEXT: vblendvpd %xmm12, %xmm7, %xmm11, %xmm3
|
|
; AVX1-NEXT: vxorpd %xmm4, %xmm3, %xmm5
|
|
; AVX1-NEXT: vblendvpd %xmm9, %xmm10, %xmm8, %xmm6
|
|
; AVX1-NEXT: vxorpd %xmm4, %xmm6, %xmm7
|
|
; AVX1-NEXT: vpcmpgtq %xmm5, %xmm7, %xmm5
|
|
; AVX1-NEXT: vblendvpd %xmm5, %xmm6, %xmm3, %xmm3
|
|
; AVX1-NEXT: vxorpd %xmm4, %xmm3, %xmm5
|
|
; AVX1-NEXT: vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
|
|
; AVX1-NEXT: vxorpd %xmm4, %xmm0, %xmm1
|
|
; AVX1-NEXT: vpcmpgtq %xmm5, %xmm1, %xmm1
|
|
; AVX1-NEXT: vblendvpd %xmm1, %xmm0, %xmm3, %xmm0
|
|
; AVX1-NEXT: vpermilps {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; AVX1-NEXT: vxorpd %xmm4, %xmm0, %xmm2
|
|
; AVX1-NEXT: vxorpd %xmm4, %xmm1, %xmm3
|
|
; AVX1-NEXT: vpcmpgtq %xmm3, %xmm2, %xmm2
|
|
; AVX1-NEXT: vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
|
|
; AVX1-NEXT: vmovq %xmm0, %rax
|
|
; AVX1-NEXT: vzeroupper
|
|
; AVX1-NEXT: retq
|
|
;
|
|
; AVX2-LABEL: test_v16i64:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vpbroadcastq {{.*#+}} ymm4 = [9223372036854775808,9223372036854775808,9223372036854775808,9223372036854775808]
|
|
; AVX2-NEXT: vpxor %ymm4, %ymm2, %ymm5
|
|
; AVX2-NEXT: vpxor %ymm4, %ymm0, %ymm6
|
|
; AVX2-NEXT: vpcmpgtq %ymm5, %ymm6, %ymm5
|
|
; AVX2-NEXT: vblendvpd %ymm5, %ymm0, %ymm2, %ymm0
|
|
; AVX2-NEXT: vpxor %ymm4, %ymm3, %ymm2
|
|
; AVX2-NEXT: vpxor %ymm4, %ymm1, %ymm5
|
|
; AVX2-NEXT: vpcmpgtq %ymm2, %ymm5, %ymm2
|
|
; AVX2-NEXT: vblendvpd %ymm2, %ymm1, %ymm3, %ymm1
|
|
; AVX2-NEXT: vxorpd %ymm4, %ymm1, %ymm2
|
|
; AVX2-NEXT: vxorpd %ymm4, %ymm0, %ymm3
|
|
; AVX2-NEXT: vpcmpgtq %ymm2, %ymm3, %ymm2
|
|
; AVX2-NEXT: vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
|
|
; AVX2-NEXT: vextractf128 $1, %ymm0, %xmm1
|
|
; AVX2-NEXT: vxorpd %xmm4, %xmm1, %xmm2
|
|
; AVX2-NEXT: vxorpd %xmm4, %xmm0, %xmm3
|
|
; AVX2-NEXT: vpcmpgtq %xmm2, %xmm3, %xmm2
|
|
; AVX2-NEXT: vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
|
|
; AVX2-NEXT: vpermilps {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; AVX2-NEXT: vxorpd %xmm4, %xmm0, %xmm2
|
|
; AVX2-NEXT: vxorpd %xmm4, %xmm1, %xmm3
|
|
; AVX2-NEXT: vpcmpgtq %xmm3, %xmm2, %xmm2
|
|
; AVX2-NEXT: vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
|
|
; AVX2-NEXT: vmovq %xmm0, %rax
|
|
; AVX2-NEXT: vzeroupper
|
|
; AVX2-NEXT: retq
|
|
;
|
|
; AVX512BW-LABEL: test_v16i64:
|
|
; AVX512BW: # %bb.0:
|
|
; AVX512BW-NEXT: vpmaxuq %zmm1, %zmm0, %zmm0
|
|
; AVX512BW-NEXT: vextracti64x4 $1, %zmm0, %ymm1
|
|
; AVX512BW-NEXT: vpmaxuq %zmm1, %zmm0, %zmm0
|
|
; AVX512BW-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX512BW-NEXT: vpmaxuq %zmm1, %zmm0, %zmm0
|
|
; AVX512BW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; AVX512BW-NEXT: vpmaxuq %zmm1, %zmm0, %zmm0
|
|
; AVX512BW-NEXT: vmovq %xmm0, %rax
|
|
; AVX512BW-NEXT: vzeroupper
|
|
; AVX512BW-NEXT: retq
|
|
;
|
|
; AVX512VL-LABEL: test_v16i64:
|
|
; AVX512VL: # %bb.0:
|
|
; AVX512VL-NEXT: vpmaxuq %zmm1, %zmm0, %zmm0
|
|
; AVX512VL-NEXT: vextracti64x4 $1, %zmm0, %ymm1
|
|
; AVX512VL-NEXT: vpmaxuq %zmm1, %zmm0, %zmm0
|
|
; AVX512VL-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX512VL-NEXT: vpmaxuq %xmm1, %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; AVX512VL-NEXT: vpmaxuq %xmm1, %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vmovq %xmm0, %rax
|
|
; AVX512VL-NEXT: vzeroupper
|
|
; AVX512VL-NEXT: retq
|
|
%1 = call i64 @llvm.experimental.vector.reduce.umax.v16i64(<16 x i64> %a0)
|
|
ret i64 %1
|
|
}
|
|
|
|
;
|
|
; vXi32
|
|
;
|
|
|
|
define i32 @test_v2i32(<2 x i32> %a0) {
|
|
; SSE2-LABEL: test_v2i32:
|
|
; SSE2: # %bb.0:
|
|
; SSE2-NEXT: movdqa {{.*#+}} xmm1 = [4294967295,0,4294967295,0]
|
|
; SSE2-NEXT: pand %xmm0, %xmm1
|
|
; SSE2-NEXT: pxor %xmm2, %xmm2
|
|
; SSE2-NEXT: punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm2[2],xmm0[3],xmm2[3]
|
|
; SSE2-NEXT: movdqa {{.*#+}} xmm2 = [2147483648,2147483648]
|
|
; SSE2-NEXT: movdqa %xmm0, %xmm3
|
|
; SSE2-NEXT: pxor %xmm2, %xmm3
|
|
; SSE2-NEXT: pxor %xmm1, %xmm2
|
|
; SSE2-NEXT: movdqa %xmm2, %xmm4
|
|
; SSE2-NEXT: pcmpgtd %xmm3, %xmm4
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
|
|
; SSE2-NEXT: pcmpeqd %xmm3, %xmm2
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
|
|
; SSE2-NEXT: pand %xmm5, %xmm2
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
|
|
; SSE2-NEXT: por %xmm2, %xmm3
|
|
; SSE2-NEXT: pand %xmm3, %xmm1
|
|
; SSE2-NEXT: pandn %xmm0, %xmm3
|
|
; SSE2-NEXT: por %xmm1, %xmm3
|
|
; SSE2-NEXT: movd %xmm3, %eax
|
|
; SSE2-NEXT: retq
|
|
;
|
|
; SSE41-LABEL: test_v2i32:
|
|
; SSE41: # %bb.0:
|
|
; SSE41-NEXT: pxor %xmm1, %xmm1
|
|
; SSE41-NEXT: pblendw {{.*#+}} xmm1 = xmm0[0,1],xmm1[2,3],xmm0[4,5],xmm1[6,7]
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
|
|
; SSE41-NEXT: pmovzxdq {{.*#+}} xmm2 = xmm0[0],zero,xmm0[1],zero
|
|
; SSE41-NEXT: movdqa {{.*#+}} xmm0 = [2147483648,2147483648]
|
|
; SSE41-NEXT: movdqa %xmm1, %xmm3
|
|
; SSE41-NEXT: pxor %xmm0, %xmm3
|
|
; SSE41-NEXT: pxor %xmm2, %xmm0
|
|
; SSE41-NEXT: movdqa %xmm3, %xmm4
|
|
; SSE41-NEXT: pcmpeqd %xmm0, %xmm4
|
|
; SSE41-NEXT: pcmpgtd %xmm0, %xmm3
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm3[0,0,2,2]
|
|
; SSE41-NEXT: pand %xmm4, %xmm0
|
|
; SSE41-NEXT: por %xmm3, %xmm0
|
|
; SSE41-NEXT: blendvpd %xmm0, %xmm1, %xmm2
|
|
; SSE41-NEXT: movd %xmm2, %eax
|
|
; SSE41-NEXT: retq
|
|
;
|
|
; AVX1-LABEL: test_v2i32:
|
|
; AVX1: # %bb.0:
|
|
; AVX1-NEXT: vpxor %xmm1, %xmm1, %xmm1
|
|
; AVX1-NEXT: vpblendw {{.*#+}} xmm2 = xmm0[0,1],xmm1[2,3],xmm0[4,5],xmm1[6,7]
|
|
; AVX1-NEXT: vpunpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
|
|
; AVX1-NEXT: vpcmpgtq %xmm0, %xmm2, %xmm1
|
|
; AVX1-NEXT: vblendvpd %xmm1, %xmm2, %xmm0, %xmm0
|
|
; AVX1-NEXT: vmovd %xmm0, %eax
|
|
; AVX1-NEXT: retq
|
|
;
|
|
; AVX2-LABEL: test_v2i32:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vpxor %xmm1, %xmm1, %xmm1
|
|
; AVX2-NEXT: vpblendd {{.*#+}} xmm2 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
|
|
; AVX2-NEXT: vpunpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
|
|
; AVX2-NEXT: vpcmpgtq %xmm0, %xmm2, %xmm1
|
|
; AVX2-NEXT: vblendvpd %xmm1, %xmm2, %xmm0, %xmm0
|
|
; AVX2-NEXT: vmovd %xmm0, %eax
|
|
; AVX2-NEXT: retq
|
|
;
|
|
; AVX512BW-LABEL: test_v2i32:
|
|
; AVX512BW: # %bb.0:
|
|
; AVX512BW-NEXT: vpxor %xmm1, %xmm1, %xmm1
|
|
; AVX512BW-NEXT: vpblendd {{.*#+}} xmm2 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
|
|
; AVX512BW-NEXT: vpunpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
|
|
; AVX512BW-NEXT: vpmaxuq %zmm0, %zmm2, %zmm0
|
|
; AVX512BW-NEXT: vmovd %xmm0, %eax
|
|
; AVX512BW-NEXT: vzeroupper
|
|
; AVX512BW-NEXT: retq
|
|
;
|
|
; AVX512VL-LABEL: test_v2i32:
|
|
; AVX512VL: # %bb.0:
|
|
; AVX512VL-NEXT: vpxor %xmm1, %xmm1, %xmm1
|
|
; AVX512VL-NEXT: vpblendd {{.*#+}} xmm1 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
|
|
; AVX512VL-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
|
|
; AVX512VL-NEXT: vpmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero
|
|
; AVX512VL-NEXT: vpmaxuq %xmm0, %xmm1, %xmm0
|
|
; AVX512VL-NEXT: vmovd %xmm0, %eax
|
|
; AVX512VL-NEXT: retq
|
|
%1 = call i32 @llvm.experimental.vector.reduce.umax.v2i32(<2 x i32> %a0)
|
|
ret i32 %1
|
|
}
|
|
|
|
define i32 @test_v4i32(<4 x i32> %a0) {
|
|
; SSE2-LABEL: test_v4i32:
|
|
; SSE2: # %bb.0:
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; SSE2-NEXT: movdqa {{.*#+}} xmm2 = [2147483648,2147483648,2147483648,2147483648]
|
|
; SSE2-NEXT: movdqa %xmm0, %xmm3
|
|
; SSE2-NEXT: pxor %xmm2, %xmm3
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm4
|
|
; SSE2-NEXT: pxor %xmm2, %xmm4
|
|
; SSE2-NEXT: pcmpgtd %xmm4, %xmm3
|
|
; SSE2-NEXT: pand %xmm3, %xmm0
|
|
; SSE2-NEXT: pandn %xmm1, %xmm3
|
|
; SSE2-NEXT: por %xmm0, %xmm3
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm3[1,1,2,3]
|
|
; SSE2-NEXT: movdqa %xmm3, %xmm1
|
|
; SSE2-NEXT: pxor %xmm2, %xmm1
|
|
; SSE2-NEXT: pxor %xmm0, %xmm2
|
|
; SSE2-NEXT: pcmpgtd %xmm2, %xmm1
|
|
; SSE2-NEXT: pand %xmm1, %xmm3
|
|
; SSE2-NEXT: pandn %xmm0, %xmm1
|
|
; SSE2-NEXT: por %xmm3, %xmm1
|
|
; SSE2-NEXT: movd %xmm1, %eax
|
|
; SSE2-NEXT: retq
|
|
;
|
|
; SSE41-LABEL: test_v4i32:
|
|
; SSE41: # %bb.0:
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; SSE41-NEXT: pmaxud %xmm0, %xmm1
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,2,3]
|
|
; SSE41-NEXT: pmaxud %xmm1, %xmm0
|
|
; SSE41-NEXT: movd %xmm0, %eax
|
|
; SSE41-NEXT: retq
|
|
;
|
|
; AVX-LABEL: test_v4i32:
|
|
; AVX: # %bb.0:
|
|
; AVX-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; AVX-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
|
|
; AVX-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX-NEXT: vmovd %xmm0, %eax
|
|
; AVX-NEXT: retq
|
|
;
|
|
; AVX512-LABEL: test_v4i32:
|
|
; AVX512: # %bb.0:
|
|
; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; AVX512-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
|
|
; AVX512-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX512-NEXT: vmovd %xmm0, %eax
|
|
; AVX512-NEXT: retq
|
|
%1 = call i32 @llvm.experimental.vector.reduce.umax.v4i32(<4 x i32> %a0)
|
|
ret i32 %1
|
|
}
|
|
|
|
define i32 @test_v8i32(<8 x i32> %a0) {
|
|
; SSE2-LABEL: test_v8i32:
|
|
; SSE2: # %bb.0:
|
|
; SSE2-NEXT: movdqa {{.*#+}} xmm2 = [2147483648,2147483648,2147483648,2147483648]
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm3
|
|
; SSE2-NEXT: pxor %xmm2, %xmm3
|
|
; SSE2-NEXT: movdqa %xmm0, %xmm4
|
|
; SSE2-NEXT: pxor %xmm2, %xmm4
|
|
; SSE2-NEXT: pcmpgtd %xmm3, %xmm4
|
|
; SSE2-NEXT: pand %xmm4, %xmm0
|
|
; SSE2-NEXT: pandn %xmm1, %xmm4
|
|
; SSE2-NEXT: por %xmm0, %xmm4
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm4[2,3,0,1]
|
|
; SSE2-NEXT: movdqa %xmm4, %xmm1
|
|
; SSE2-NEXT: pxor %xmm2, %xmm1
|
|
; SSE2-NEXT: movdqa %xmm0, %xmm3
|
|
; SSE2-NEXT: pxor %xmm2, %xmm3
|
|
; SSE2-NEXT: pcmpgtd %xmm3, %xmm1
|
|
; SSE2-NEXT: pand %xmm1, %xmm4
|
|
; SSE2-NEXT: pandn %xmm0, %xmm1
|
|
; SSE2-NEXT: por %xmm4, %xmm1
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,2,3]
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm3
|
|
; SSE2-NEXT: pxor %xmm2, %xmm3
|
|
; SSE2-NEXT: pxor %xmm0, %xmm2
|
|
; SSE2-NEXT: pcmpgtd %xmm2, %xmm3
|
|
; SSE2-NEXT: pand %xmm3, %xmm1
|
|
; SSE2-NEXT: pandn %xmm0, %xmm3
|
|
; SSE2-NEXT: por %xmm1, %xmm3
|
|
; SSE2-NEXT: movd %xmm3, %eax
|
|
; SSE2-NEXT: retq
|
|
;
|
|
; SSE41-LABEL: test_v8i32:
|
|
; SSE41: # %bb.0:
|
|
; SSE41-NEXT: pmaxud %xmm1, %xmm0
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; SSE41-NEXT: pmaxud %xmm0, %xmm1
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,2,3]
|
|
; SSE41-NEXT: pmaxud %xmm1, %xmm0
|
|
; SSE41-NEXT: movd %xmm0, %eax
|
|
; SSE41-NEXT: retq
|
|
;
|
|
; AVX1-LABEL: test_v8i32:
|
|
; AVX1: # %bb.0:
|
|
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
|
|
; AVX1-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; AVX1-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
|
|
; AVX1-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX1-NEXT: vmovd %xmm0, %eax
|
|
; AVX1-NEXT: vzeroupper
|
|
; AVX1-NEXT: retq
|
|
;
|
|
; AVX2-LABEL: test_v8i32:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX2-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; AVX2-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
|
|
; AVX2-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX2-NEXT: vmovd %xmm0, %eax
|
|
; AVX2-NEXT: vzeroupper
|
|
; AVX2-NEXT: retq
|
|
;
|
|
; AVX512-LABEL: test_v8i32:
|
|
; AVX512: # %bb.0:
|
|
; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX512-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; AVX512-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
|
|
; AVX512-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX512-NEXT: vmovd %xmm0, %eax
|
|
; AVX512-NEXT: vzeroupper
|
|
; AVX512-NEXT: retq
|
|
%1 = call i32 @llvm.experimental.vector.reduce.umax.v8i32(<8 x i32> %a0)
|
|
ret i32 %1
|
|
}
|
|
|
|
define i32 @test_v16i32(<16 x i32> %a0) {
|
|
; SSE2-LABEL: test_v16i32:
|
|
; SSE2: # %bb.0:
|
|
; SSE2-NEXT: movdqa {{.*#+}} xmm4 = [2147483648,2147483648,2147483648,2147483648]
|
|
; SSE2-NEXT: movdqa %xmm2, %xmm6
|
|
; SSE2-NEXT: pxor %xmm4, %xmm6
|
|
; SSE2-NEXT: movdqa %xmm0, %xmm5
|
|
; SSE2-NEXT: pxor %xmm4, %xmm5
|
|
; SSE2-NEXT: pcmpgtd %xmm6, %xmm5
|
|
; SSE2-NEXT: pand %xmm5, %xmm0
|
|
; SSE2-NEXT: pandn %xmm2, %xmm5
|
|
; SSE2-NEXT: por %xmm0, %xmm5
|
|
; SSE2-NEXT: movdqa %xmm3, %xmm0
|
|
; SSE2-NEXT: pxor %xmm4, %xmm0
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm2
|
|
; SSE2-NEXT: pxor %xmm4, %xmm2
|
|
; SSE2-NEXT: pcmpgtd %xmm0, %xmm2
|
|
; SSE2-NEXT: pand %xmm2, %xmm1
|
|
; SSE2-NEXT: pandn %xmm3, %xmm2
|
|
; SSE2-NEXT: por %xmm1, %xmm2
|
|
; SSE2-NEXT: movdqa %xmm2, %xmm0
|
|
; SSE2-NEXT: pxor %xmm4, %xmm0
|
|
; SSE2-NEXT: movdqa %xmm5, %xmm1
|
|
; SSE2-NEXT: pxor %xmm4, %xmm1
|
|
; SSE2-NEXT: pcmpgtd %xmm0, %xmm1
|
|
; SSE2-NEXT: pand %xmm1, %xmm5
|
|
; SSE2-NEXT: pandn %xmm2, %xmm1
|
|
; SSE2-NEXT: por %xmm5, %xmm1
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm1[2,3,0,1]
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm2
|
|
; SSE2-NEXT: pxor %xmm4, %xmm2
|
|
; SSE2-NEXT: movdqa %xmm0, %xmm3
|
|
; SSE2-NEXT: pxor %xmm4, %xmm3
|
|
; SSE2-NEXT: pcmpgtd %xmm3, %xmm2
|
|
; SSE2-NEXT: pand %xmm2, %xmm1
|
|
; SSE2-NEXT: pandn %xmm0, %xmm2
|
|
; SSE2-NEXT: por %xmm1, %xmm2
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm2[1,1,2,3]
|
|
; SSE2-NEXT: movdqa %xmm2, %xmm1
|
|
; SSE2-NEXT: pxor %xmm4, %xmm1
|
|
; SSE2-NEXT: pxor %xmm0, %xmm4
|
|
; SSE2-NEXT: pcmpgtd %xmm4, %xmm1
|
|
; SSE2-NEXT: pand %xmm1, %xmm2
|
|
; SSE2-NEXT: pandn %xmm0, %xmm1
|
|
; SSE2-NEXT: por %xmm2, %xmm1
|
|
; SSE2-NEXT: movd %xmm1, %eax
|
|
; SSE2-NEXT: retq
|
|
;
|
|
; SSE41-LABEL: test_v16i32:
|
|
; SSE41: # %bb.0:
|
|
; SSE41-NEXT: pmaxud %xmm3, %xmm1
|
|
; SSE41-NEXT: pmaxud %xmm2, %xmm1
|
|
; SSE41-NEXT: pmaxud %xmm0, %xmm1
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[2,3,0,1]
|
|
; SSE41-NEXT: pmaxud %xmm1, %xmm0
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
|
|
; SSE41-NEXT: pmaxud %xmm0, %xmm1
|
|
; SSE41-NEXT: movd %xmm1, %eax
|
|
; SSE41-NEXT: retq
|
|
;
|
|
; AVX1-LABEL: test_v16i32:
|
|
; AVX1: # %bb.0:
|
|
; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm2
|
|
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm3
|
|
; AVX1-NEXT: vpmaxud %xmm2, %xmm3, %xmm2
|
|
; AVX1-NEXT: vpmaxud %xmm2, %xmm1, %xmm1
|
|
; AVX1-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; AVX1-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
|
|
; AVX1-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX1-NEXT: vmovd %xmm0, %eax
|
|
; AVX1-NEXT: vzeroupper
|
|
; AVX1-NEXT: retq
|
|
;
|
|
; AVX2-LABEL: test_v16i32:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vpmaxud %ymm1, %ymm0, %ymm0
|
|
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX2-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; AVX2-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
|
|
; AVX2-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX2-NEXT: vmovd %xmm0, %eax
|
|
; AVX2-NEXT: vzeroupper
|
|
; AVX2-NEXT: retq
|
|
;
|
|
; AVX512-LABEL: test_v16i32:
|
|
; AVX512: # %bb.0:
|
|
; AVX512-NEXT: vextracti64x4 $1, %zmm0, %ymm1
|
|
; AVX512-NEXT: vpmaxud %zmm1, %zmm0, %zmm0
|
|
; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX512-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; AVX512-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
|
|
; AVX512-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX512-NEXT: vmovd %xmm0, %eax
|
|
; AVX512-NEXT: vzeroupper
|
|
; AVX512-NEXT: retq
|
|
%1 = call i32 @llvm.experimental.vector.reduce.umax.v16i32(<16 x i32> %a0)
|
|
ret i32 %1
|
|
}
|
|
|
|
define i32 @test_v32i32(<32 x i32> %a0) {
|
|
; SSE2-LABEL: test_v32i32:
|
|
; SSE2: # %bb.0:
|
|
; SSE2-NEXT: movdqa {{.*#+}} xmm8 = [2147483648,2147483648,2147483648,2147483648]
|
|
; SSE2-NEXT: movdqa %xmm5, %xmm10
|
|
; SSE2-NEXT: pxor %xmm8, %xmm10
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm9
|
|
; SSE2-NEXT: pxor %xmm8, %xmm9
|
|
; SSE2-NEXT: pcmpgtd %xmm10, %xmm9
|
|
; SSE2-NEXT: pand %xmm9, %xmm1
|
|
; SSE2-NEXT: pandn %xmm5, %xmm9
|
|
; SSE2-NEXT: por %xmm1, %xmm9
|
|
; SSE2-NEXT: movdqa %xmm7, %xmm5
|
|
; SSE2-NEXT: pxor %xmm8, %xmm5
|
|
; SSE2-NEXT: movdqa %xmm3, %xmm1
|
|
; SSE2-NEXT: pxor %xmm8, %xmm1
|
|
; SSE2-NEXT: pcmpgtd %xmm5, %xmm1
|
|
; SSE2-NEXT: pand %xmm1, %xmm3
|
|
; SSE2-NEXT: pandn %xmm7, %xmm1
|
|
; SSE2-NEXT: por %xmm3, %xmm1
|
|
; SSE2-NEXT: movdqa %xmm4, %xmm5
|
|
; SSE2-NEXT: pxor %xmm8, %xmm5
|
|
; SSE2-NEXT: movdqa %xmm0, %xmm3
|
|
; SSE2-NEXT: pxor %xmm8, %xmm3
|
|
; SSE2-NEXT: pcmpgtd %xmm5, %xmm3
|
|
; SSE2-NEXT: pand %xmm3, %xmm0
|
|
; SSE2-NEXT: pandn %xmm4, %xmm3
|
|
; SSE2-NEXT: por %xmm0, %xmm3
|
|
; SSE2-NEXT: movdqa %xmm6, %xmm0
|
|
; SSE2-NEXT: pxor %xmm8, %xmm0
|
|
; SSE2-NEXT: movdqa %xmm2, %xmm4
|
|
; SSE2-NEXT: pxor %xmm8, %xmm4
|
|
; SSE2-NEXT: pcmpgtd %xmm0, %xmm4
|
|
; SSE2-NEXT: pand %xmm4, %xmm2
|
|
; SSE2-NEXT: pandn %xmm6, %xmm4
|
|
; SSE2-NEXT: por %xmm2, %xmm4
|
|
; SSE2-NEXT: movdqa %xmm4, %xmm2
|
|
; SSE2-NEXT: pxor %xmm8, %xmm2
|
|
; SSE2-NEXT: movdqa %xmm3, %xmm0
|
|
; SSE2-NEXT: pxor %xmm8, %xmm0
|
|
; SSE2-NEXT: pcmpgtd %xmm2, %xmm0
|
|
; SSE2-NEXT: pand %xmm0, %xmm3
|
|
; SSE2-NEXT: pandn %xmm4, %xmm0
|
|
; SSE2-NEXT: por %xmm3, %xmm0
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm2
|
|
; SSE2-NEXT: pxor %xmm8, %xmm2
|
|
; SSE2-NEXT: movdqa %xmm9, %xmm3
|
|
; SSE2-NEXT: pxor %xmm8, %xmm3
|
|
; SSE2-NEXT: pcmpgtd %xmm2, %xmm3
|
|
; SSE2-NEXT: pand %xmm3, %xmm9
|
|
; SSE2-NEXT: pandn %xmm1, %xmm3
|
|
; SSE2-NEXT: por %xmm9, %xmm3
|
|
; SSE2-NEXT: movdqa %xmm3, %xmm1
|
|
; SSE2-NEXT: pxor %xmm8, %xmm1
|
|
; SSE2-NEXT: movdqa %xmm0, %xmm2
|
|
; SSE2-NEXT: pxor %xmm8, %xmm2
|
|
; SSE2-NEXT: pcmpgtd %xmm1, %xmm2
|
|
; SSE2-NEXT: pand %xmm2, %xmm0
|
|
; SSE2-NEXT: pandn %xmm3, %xmm2
|
|
; SSE2-NEXT: por %xmm0, %xmm2
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm2[2,3,0,1]
|
|
; SSE2-NEXT: movdqa %xmm2, %xmm1
|
|
; SSE2-NEXT: pxor %xmm8, %xmm1
|
|
; SSE2-NEXT: movdqa %xmm0, %xmm3
|
|
; SSE2-NEXT: pxor %xmm8, %xmm3
|
|
; SSE2-NEXT: pcmpgtd %xmm3, %xmm1
|
|
; SSE2-NEXT: pand %xmm1, %xmm2
|
|
; SSE2-NEXT: pandn %xmm0, %xmm1
|
|
; SSE2-NEXT: por %xmm2, %xmm1
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,2,3]
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm2
|
|
; SSE2-NEXT: pxor %xmm8, %xmm2
|
|
; SSE2-NEXT: pxor %xmm0, %xmm8
|
|
; SSE2-NEXT: pcmpgtd %xmm8, %xmm2
|
|
; SSE2-NEXT: pand %xmm2, %xmm1
|
|
; SSE2-NEXT: pandn %xmm0, %xmm2
|
|
; SSE2-NEXT: por %xmm1, %xmm2
|
|
; SSE2-NEXT: movd %xmm2, %eax
|
|
; SSE2-NEXT: retq
|
|
;
|
|
; SSE41-LABEL: test_v32i32:
|
|
; SSE41: # %bb.0:
|
|
; SSE41-NEXT: pmaxud %xmm6, %xmm2
|
|
; SSE41-NEXT: pmaxud %xmm7, %xmm3
|
|
; SSE41-NEXT: pmaxud %xmm5, %xmm3
|
|
; SSE41-NEXT: pmaxud %xmm1, %xmm3
|
|
; SSE41-NEXT: pmaxud %xmm4, %xmm2
|
|
; SSE41-NEXT: pmaxud %xmm3, %xmm2
|
|
; SSE41-NEXT: pmaxud %xmm0, %xmm2
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm2[2,3,0,1]
|
|
; SSE41-NEXT: pmaxud %xmm2, %xmm0
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
|
|
; SSE41-NEXT: pmaxud %xmm0, %xmm1
|
|
; SSE41-NEXT: movd %xmm1, %eax
|
|
; SSE41-NEXT: retq
|
|
;
|
|
; AVX1-LABEL: test_v32i32:
|
|
; AVX1: # %bb.0:
|
|
; AVX1-NEXT: vpmaxud %xmm3, %xmm1, %xmm4
|
|
; AVX1-NEXT: vextractf128 $1, %ymm3, %xmm3
|
|
; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm1
|
|
; AVX1-NEXT: vpmaxud %xmm3, %xmm1, %xmm1
|
|
; AVX1-NEXT: vextractf128 $1, %ymm2, %xmm3
|
|
; AVX1-NEXT: vpmaxud %xmm1, %xmm3, %xmm1
|
|
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm3
|
|
; AVX1-NEXT: vpmaxud %xmm1, %xmm3, %xmm1
|
|
; AVX1-NEXT: vpmaxud %xmm4, %xmm2, %xmm2
|
|
; AVX1-NEXT: vpmaxud %xmm1, %xmm2, %xmm1
|
|
; AVX1-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; AVX1-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
|
|
; AVX1-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX1-NEXT: vmovd %xmm0, %eax
|
|
; AVX1-NEXT: vzeroupper
|
|
; AVX1-NEXT: retq
|
|
;
|
|
; AVX2-LABEL: test_v32i32:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vpmaxud %ymm3, %ymm1, %ymm1
|
|
; AVX2-NEXT: vpmaxud %ymm1, %ymm2, %ymm1
|
|
; AVX2-NEXT: vpmaxud %ymm1, %ymm0, %ymm0
|
|
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX2-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; AVX2-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
|
|
; AVX2-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX2-NEXT: vmovd %xmm0, %eax
|
|
; AVX2-NEXT: vzeroupper
|
|
; AVX2-NEXT: retq
|
|
;
|
|
; AVX512-LABEL: test_v32i32:
|
|
; AVX512: # %bb.0:
|
|
; AVX512-NEXT: vpmaxud %zmm1, %zmm0, %zmm0
|
|
; AVX512-NEXT: vextracti64x4 $1, %zmm0, %ymm1
|
|
; AVX512-NEXT: vpmaxud %zmm1, %zmm0, %zmm0
|
|
; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX512-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; AVX512-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
|
|
; AVX512-NEXT: vpmaxud %xmm1, %xmm0, %xmm0
|
|
; AVX512-NEXT: vmovd %xmm0, %eax
|
|
; AVX512-NEXT: vzeroupper
|
|
; AVX512-NEXT: retq
|
|
%1 = call i32 @llvm.experimental.vector.reduce.umax.v32i32(<32 x i32> %a0)
|
|
ret i32 %1
|
|
}
|
|
|
|
;
|
|
; vXi16
|
|
;
|
|
|
|
define i16 @test_v2i16(<2 x i16> %a0) {
|
|
; SSE2-LABEL: test_v2i16:
|
|
; SSE2: # %bb.0:
|
|
; SSE2-NEXT: movdqa {{.*#+}} xmm1 = [65535,0,0,0,65535,0,0,0]
|
|
; SSE2-NEXT: pand %xmm0, %xmm1
|
|
; SSE2-NEXT: pxor %xmm2, %xmm2
|
|
; SSE2-NEXT: punpckhwd {{.*#+}} xmm0 = xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
|
|
; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
|
|
; SSE2-NEXT: movdqa {{.*#+}} xmm2 = [2147483648,2147483648]
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm3
|
|
; SSE2-NEXT: por %xmm2, %xmm3
|
|
; SSE2-NEXT: por %xmm0, %xmm2
|
|
; SSE2-NEXT: movdqa %xmm3, %xmm4
|
|
; SSE2-NEXT: pcmpgtd %xmm2, %xmm4
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
|
|
; SSE2-NEXT: pcmpeqd %xmm3, %xmm2
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
|
|
; SSE2-NEXT: pand %xmm5, %xmm2
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
|
|
; SSE2-NEXT: por %xmm2, %xmm3
|
|
; SSE2-NEXT: pand %xmm3, %xmm1
|
|
; SSE2-NEXT: pandn %xmm0, %xmm3
|
|
; SSE2-NEXT: por %xmm1, %xmm3
|
|
; SSE2-NEXT: movd %xmm3, %eax
|
|
; SSE2-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; SSE2-NEXT: retq
|
|
;
|
|
; SSE41-LABEL: test_v2i16:
|
|
; SSE41: # %bb.0:
|
|
; SSE41-NEXT: pxor %xmm1, %xmm1
|
|
; SSE41-NEXT: pblendw {{.*#+}} xmm1 = xmm0[0],xmm1[1,2,3],xmm0[4],xmm1[5,6,7]
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
|
|
; SSE41-NEXT: pmovzxwq {{.*#+}} xmm2 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero
|
|
; SSE41-NEXT: movdqa {{.*#+}} xmm0 = [2147483648,2147483648]
|
|
; SSE41-NEXT: movdqa %xmm1, %xmm3
|
|
; SSE41-NEXT: por %xmm0, %xmm3
|
|
; SSE41-NEXT: por %xmm2, %xmm0
|
|
; SSE41-NEXT: movdqa %xmm3, %xmm4
|
|
; SSE41-NEXT: pcmpeqd %xmm0, %xmm4
|
|
; SSE41-NEXT: pcmpgtd %xmm0, %xmm3
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm3[0,0,2,2]
|
|
; SSE41-NEXT: pand %xmm4, %xmm0
|
|
; SSE41-NEXT: por %xmm3, %xmm0
|
|
; SSE41-NEXT: blendvpd %xmm0, %xmm1, %xmm2
|
|
; SSE41-NEXT: movd %xmm2, %eax
|
|
; SSE41-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; SSE41-NEXT: retq
|
|
;
|
|
; AVX-LABEL: test_v2i16:
|
|
; AVX: # %bb.0:
|
|
; AVX-NEXT: vpxor %xmm1, %xmm1, %xmm1
|
|
; AVX-NEXT: vpblendw {{.*#+}} xmm1 = xmm0[0],xmm1[1,2,3],xmm0[4],xmm1[5,6,7]
|
|
; AVX-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
|
|
; AVX-NEXT: vpmovzxwq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero
|
|
; AVX-NEXT: vpcmpgtq %xmm0, %xmm1, %xmm2
|
|
; AVX-NEXT: vblendvpd %xmm2, %xmm1, %xmm0, %xmm0
|
|
; AVX-NEXT: vmovd %xmm0, %eax
|
|
; AVX-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; AVX-NEXT: retq
|
|
;
|
|
; AVX512BW-LABEL: test_v2i16:
|
|
; AVX512BW: # %bb.0:
|
|
; AVX512BW-NEXT: vpxor %xmm1, %xmm1, %xmm1
|
|
; AVX512BW-NEXT: vpblendw {{.*#+}} xmm1 = xmm0[0],xmm1[1,2,3],xmm0[4],xmm1[5,6,7]
|
|
; AVX512BW-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
|
|
; AVX512BW-NEXT: vpmovzxwq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero
|
|
; AVX512BW-NEXT: vpmaxuq %zmm0, %zmm1, %zmm0
|
|
; AVX512BW-NEXT: vmovd %xmm0, %eax
|
|
; AVX512BW-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; AVX512BW-NEXT: vzeroupper
|
|
; AVX512BW-NEXT: retq
|
|
;
|
|
; AVX512VL-LABEL: test_v2i16:
|
|
; AVX512VL: # %bb.0:
|
|
; AVX512VL-NEXT: vpxor %xmm1, %xmm1, %xmm1
|
|
; AVX512VL-NEXT: vpblendw {{.*#+}} xmm1 = xmm0[0],xmm1[1,2,3],xmm0[4],xmm1[5,6,7]
|
|
; AVX512VL-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
|
|
; AVX512VL-NEXT: vpmovzxwq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero
|
|
; AVX512VL-NEXT: vpmaxuq %xmm0, %xmm1, %xmm0
|
|
; AVX512VL-NEXT: vmovd %xmm0, %eax
|
|
; AVX512VL-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; AVX512VL-NEXT: retq
|
|
%1 = call i16 @llvm.experimental.vector.reduce.umax.v2i16(<2 x i16> %a0)
|
|
ret i16 %1
|
|
}
|
|
|
|
define i16 @test_v4i16(<4 x i16> %a0) {
|
|
; SSE2-LABEL: test_v4i16:
|
|
; SSE2: # %bb.0:
|
|
; SSE2-NEXT: movdqa {{.*#+}} xmm1 = [65535,0,65535,0,65535,0,65535,0]
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm0[2,3,0,1]
|
|
; SSE2-NEXT: pand %xmm1, %xmm0
|
|
; SSE2-NEXT: pand {{.*}}(%rip), %xmm2
|
|
; SSE2-NEXT: movdqa %xmm0, %xmm3
|
|
; SSE2-NEXT: pcmpgtd %xmm2, %xmm3
|
|
; SSE2-NEXT: pand %xmm3, %xmm0
|
|
; SSE2-NEXT: pandn %xmm2, %xmm3
|
|
; SSE2-NEXT: por %xmm0, %xmm3
|
|
; SSE2-NEXT: pand %xmm3, %xmm1
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm3[1,1,2,3]
|
|
; SSE2-NEXT: pxor %xmm2, %xmm2
|
|
; SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
|
|
; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm2
|
|
; SSE2-NEXT: pcmpgtd %xmm0, %xmm2
|
|
; SSE2-NEXT: pand %xmm2, %xmm1
|
|
; SSE2-NEXT: pandn %xmm0, %xmm2
|
|
; SSE2-NEXT: por %xmm1, %xmm2
|
|
; SSE2-NEXT: movd %xmm2, %eax
|
|
; SSE2-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; SSE2-NEXT: retq
|
|
;
|
|
; SSE41-LABEL: test_v4i16:
|
|
; SSE41: # %bb.0:
|
|
; SSE41-NEXT: pxor %xmm1, %xmm1
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm2 = xmm0[2,3,0,1]
|
|
; SSE41-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3],xmm0[4],xmm1[5],xmm0[6],xmm1[7]
|
|
; SSE41-NEXT: pblendw {{.*#+}} xmm2 = xmm2[0],xmm1[1],xmm2[2],xmm1[3,4,5,6,7]
|
|
; SSE41-NEXT: pmaxud %xmm0, %xmm2
|
|
; SSE41-NEXT: pblendw {{.*#+}} xmm1 = xmm2[0],xmm1[1],xmm2[2],xmm1[3],xmm2[4],xmm1[5],xmm2[6],xmm1[7]
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm2[1,1,2,3]
|
|
; SSE41-NEXT: pmovzxwq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero
|
|
; SSE41-NEXT: pmaxud %xmm1, %xmm0
|
|
; SSE41-NEXT: movd %xmm0, %eax
|
|
; SSE41-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; SSE41-NEXT: retq
|
|
;
|
|
; AVX-LABEL: test_v4i16:
|
|
; AVX: # %bb.0:
|
|
; AVX-NEXT: vpxor %xmm1, %xmm1, %xmm1
|
|
; AVX-NEXT: vpblendw {{.*#+}} xmm2 = xmm0[0],xmm1[1],xmm0[2],xmm1[3],xmm0[4],xmm1[5],xmm0[6],xmm1[7]
|
|
; AVX-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
|
|
; AVX-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3,4,5,6,7]
|
|
; AVX-NEXT: vpmaxud %xmm0, %xmm2, %xmm0
|
|
; AVX-NEXT: vpblendw {{.*#+}} xmm1 = xmm0[0],xmm1[1],xmm0[2],xmm1[3],xmm0[4],xmm1[5],xmm0[6],xmm1[7]
|
|
; AVX-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
|
|
; AVX-NEXT: vpmovzxwq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero
|
|
; AVX-NEXT: vpmaxud %xmm0, %xmm1, %xmm0
|
|
; AVX-NEXT: vmovd %xmm0, %eax
|
|
; AVX-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; AVX-NEXT: retq
|
|
;
|
|
; AVX512-LABEL: test_v4i16:
|
|
; AVX512: # %bb.0:
|
|
; AVX512-NEXT: vpxor %xmm1, %xmm1, %xmm1
|
|
; AVX512-NEXT: vpblendw {{.*#+}} xmm2 = xmm0[0],xmm1[1],xmm0[2],xmm1[3],xmm0[4],xmm1[5],xmm0[6],xmm1[7]
|
|
; AVX512-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
|
|
; AVX512-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3,4,5,6,7]
|
|
; AVX512-NEXT: vpmaxud %xmm0, %xmm2, %xmm0
|
|
; AVX512-NEXT: vpblendw {{.*#+}} xmm1 = xmm0[0],xmm1[1],xmm0[2],xmm1[3],xmm0[4],xmm1[5],xmm0[6],xmm1[7]
|
|
; AVX512-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
|
|
; AVX512-NEXT: vpmovzxwq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero
|
|
; AVX512-NEXT: vpmaxud %xmm0, %xmm1, %xmm0
|
|
; AVX512-NEXT: vmovd %xmm0, %eax
|
|
; AVX512-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; AVX512-NEXT: retq
|
|
%1 = call i16 @llvm.experimental.vector.reduce.umax.v4i16(<4 x i16> %a0)
|
|
ret i16 %1
|
|
}
|
|
|
|
define i16 @test_v8i16(<8 x i16> %a0) {
|
|
; SSE2-LABEL: test_v8i16:
|
|
; SSE2: # %bb.0:
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; SSE2-NEXT: movdqa {{.*#+}} xmm2 = [32768,32768,32768,32768,32768,32768,32768,32768]
|
|
; SSE2-NEXT: pxor %xmm2, %xmm0
|
|
; SSE2-NEXT: pxor %xmm2, %xmm1
|
|
; SSE2-NEXT: pmaxsw %xmm0, %xmm1
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm0
|
|
; SSE2-NEXT: pxor %xmm2, %xmm0
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
|
|
; SSE2-NEXT: pxor %xmm2, %xmm0
|
|
; SSE2-NEXT: pmaxsw %xmm1, %xmm0
|
|
; SSE2-NEXT: movdqa %xmm0, %xmm1
|
|
; SSE2-NEXT: pxor %xmm2, %xmm1
|
|
; SSE2-NEXT: psrld $16, %xmm1
|
|
; SSE2-NEXT: pxor %xmm2, %xmm1
|
|
; SSE2-NEXT: pmaxsw %xmm0, %xmm1
|
|
; SSE2-NEXT: movd %xmm1, %eax
|
|
; SSE2-NEXT: xorl $32768, %eax # imm = 0x8000
|
|
; SSE2-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; SSE2-NEXT: retq
|
|
;
|
|
; SSE41-LABEL: test_v8i16:
|
|
; SSE41: # %bb.0:
|
|
; SSE41-NEXT: pcmpeqd %xmm1, %xmm1
|
|
; SSE41-NEXT: pxor %xmm0, %xmm1
|
|
; SSE41-NEXT: phminposuw %xmm1, %xmm0
|
|
; SSE41-NEXT: movd %xmm0, %eax
|
|
; SSE41-NEXT: notl %eax
|
|
; SSE41-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; SSE41-NEXT: retq
|
|
;
|
|
; AVX-LABEL: test_v8i16:
|
|
; AVX: # %bb.0:
|
|
; AVX-NEXT: vpcmpeqd %xmm1, %xmm1, %xmm1
|
|
; AVX-NEXT: vpxor %xmm1, %xmm0, %xmm0
|
|
; AVX-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX-NEXT: vmovd %xmm0, %eax
|
|
; AVX-NEXT: notl %eax
|
|
; AVX-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; AVX-NEXT: retq
|
|
;
|
|
; AVX512BW-LABEL: test_v8i16:
|
|
; AVX512BW: # %bb.0:
|
|
; AVX512BW-NEXT: # kill: def $xmm0 killed $xmm0 def $zmm0
|
|
; AVX512BW-NEXT: vpternlogq $15, %zmm0, %zmm0, %zmm0
|
|
; AVX512BW-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX512BW-NEXT: vmovd %xmm0, %eax
|
|
; AVX512BW-NEXT: notl %eax
|
|
; AVX512BW-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; AVX512BW-NEXT: vzeroupper
|
|
; AVX512BW-NEXT: retq
|
|
;
|
|
; AVX512VL-LABEL: test_v8i16:
|
|
; AVX512VL: # %bb.0:
|
|
; AVX512VL-NEXT: vpternlogq $15, %xmm0, %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vmovd %xmm0, %eax
|
|
; AVX512VL-NEXT: notl %eax
|
|
; AVX512VL-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; AVX512VL-NEXT: retq
|
|
%1 = call i16 @llvm.experimental.vector.reduce.umax.v8i16(<8 x i16> %a0)
|
|
ret i16 %1
|
|
}
|
|
|
|
define i16 @test_v16i16(<16 x i16> %a0) {
|
|
; SSE2-LABEL: test_v16i16:
|
|
; SSE2: # %bb.0:
|
|
; SSE2-NEXT: movdqa {{.*#+}} xmm2 = [32768,32768,32768,32768,32768,32768,32768,32768]
|
|
; SSE2-NEXT: pxor %xmm2, %xmm1
|
|
; SSE2-NEXT: pxor %xmm2, %xmm0
|
|
; SSE2-NEXT: pmaxsw %xmm1, %xmm0
|
|
; SSE2-NEXT: movdqa %xmm0, %xmm1
|
|
; SSE2-NEXT: pxor %xmm2, %xmm1
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
|
|
; SSE2-NEXT: pxor %xmm2, %xmm1
|
|
; SSE2-NEXT: pmaxsw %xmm0, %xmm1
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm0
|
|
; SSE2-NEXT: pxor %xmm2, %xmm0
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
|
|
; SSE2-NEXT: pxor %xmm2, %xmm0
|
|
; SSE2-NEXT: pmaxsw %xmm1, %xmm0
|
|
; SSE2-NEXT: movdqa %xmm0, %xmm1
|
|
; SSE2-NEXT: pxor %xmm2, %xmm1
|
|
; SSE2-NEXT: psrld $16, %xmm1
|
|
; SSE2-NEXT: pxor %xmm2, %xmm1
|
|
; SSE2-NEXT: pmaxsw %xmm0, %xmm1
|
|
; SSE2-NEXT: movd %xmm1, %eax
|
|
; SSE2-NEXT: xorl $32768, %eax # imm = 0x8000
|
|
; SSE2-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; SSE2-NEXT: retq
|
|
;
|
|
; SSE41-LABEL: test_v16i16:
|
|
; SSE41: # %bb.0:
|
|
; SSE41-NEXT: pmaxuw %xmm1, %xmm0
|
|
; SSE41-NEXT: pcmpeqd %xmm1, %xmm1
|
|
; SSE41-NEXT: pxor %xmm0, %xmm1
|
|
; SSE41-NEXT: phminposuw %xmm1, %xmm0
|
|
; SSE41-NEXT: movd %xmm0, %eax
|
|
; SSE41-NEXT: notl %eax
|
|
; SSE41-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; SSE41-NEXT: retq
|
|
;
|
|
; AVX1-LABEL: test_v16i16:
|
|
; AVX1: # %bb.0:
|
|
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
|
|
; AVX1-NEXT: vpmaxuw %xmm1, %xmm0, %xmm0
|
|
; AVX1-NEXT: vpcmpeqd %xmm1, %xmm1, %xmm1
|
|
; AVX1-NEXT: vpxor %xmm1, %xmm0, %xmm0
|
|
; AVX1-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX1-NEXT: vmovd %xmm0, %eax
|
|
; AVX1-NEXT: notl %eax
|
|
; AVX1-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; AVX1-NEXT: vzeroupper
|
|
; AVX1-NEXT: retq
|
|
;
|
|
; AVX2-LABEL: test_v16i16:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX2-NEXT: vpmaxuw %xmm1, %xmm0, %xmm0
|
|
; AVX2-NEXT: vpcmpeqd %xmm1, %xmm1, %xmm1
|
|
; AVX2-NEXT: vpxor %xmm1, %xmm0, %xmm0
|
|
; AVX2-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX2-NEXT: vmovd %xmm0, %eax
|
|
; AVX2-NEXT: notl %eax
|
|
; AVX2-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; AVX2-NEXT: vzeroupper
|
|
; AVX2-NEXT: retq
|
|
;
|
|
; AVX512BW-LABEL: test_v16i16:
|
|
; AVX512BW: # %bb.0:
|
|
; AVX512BW-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX512BW-NEXT: vpmaxuw %xmm1, %xmm0, %xmm0
|
|
; AVX512BW-NEXT: vpternlogq $15, %zmm0, %zmm0, %zmm0
|
|
; AVX512BW-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX512BW-NEXT: vmovd %xmm0, %eax
|
|
; AVX512BW-NEXT: notl %eax
|
|
; AVX512BW-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; AVX512BW-NEXT: vzeroupper
|
|
; AVX512BW-NEXT: retq
|
|
;
|
|
; AVX512VL-LABEL: test_v16i16:
|
|
; AVX512VL: # %bb.0:
|
|
; AVX512VL-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX512VL-NEXT: vpmaxuw %xmm1, %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vpternlogq $15, %xmm0, %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vmovd %xmm0, %eax
|
|
; AVX512VL-NEXT: notl %eax
|
|
; AVX512VL-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; AVX512VL-NEXT: vzeroupper
|
|
; AVX512VL-NEXT: retq
|
|
%1 = call i16 @llvm.experimental.vector.reduce.umax.v16i16(<16 x i16> %a0)
|
|
ret i16 %1
|
|
}
|
|
|
|
define i16 @test_v32i16(<32 x i16> %a0) {
|
|
; SSE2-LABEL: test_v32i16:
|
|
; SSE2: # %bb.0:
|
|
; SSE2-NEXT: movdqa {{.*#+}} xmm4 = [32768,32768,32768,32768,32768,32768,32768,32768]
|
|
; SSE2-NEXT: pxor %xmm4, %xmm3
|
|
; SSE2-NEXT: pxor %xmm4, %xmm1
|
|
; SSE2-NEXT: pmaxsw %xmm3, %xmm1
|
|
; SSE2-NEXT: pxor %xmm4, %xmm2
|
|
; SSE2-NEXT: pmaxsw %xmm1, %xmm2
|
|
; SSE2-NEXT: pxor %xmm4, %xmm0
|
|
; SSE2-NEXT: pmaxsw %xmm2, %xmm0
|
|
; SSE2-NEXT: movdqa %xmm0, %xmm1
|
|
; SSE2-NEXT: pxor %xmm4, %xmm1
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
|
|
; SSE2-NEXT: pxor %xmm4, %xmm1
|
|
; SSE2-NEXT: pmaxsw %xmm0, %xmm1
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm0
|
|
; SSE2-NEXT: pxor %xmm4, %xmm0
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
|
|
; SSE2-NEXT: pxor %xmm4, %xmm0
|
|
; SSE2-NEXT: pmaxsw %xmm1, %xmm0
|
|
; SSE2-NEXT: movdqa %xmm0, %xmm1
|
|
; SSE2-NEXT: pxor %xmm4, %xmm1
|
|
; SSE2-NEXT: psrld $16, %xmm1
|
|
; SSE2-NEXT: pxor %xmm4, %xmm1
|
|
; SSE2-NEXT: pmaxsw %xmm0, %xmm1
|
|
; SSE2-NEXT: movd %xmm1, %eax
|
|
; SSE2-NEXT: xorl $32768, %eax # imm = 0x8000
|
|
; SSE2-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; SSE2-NEXT: retq
|
|
;
|
|
; SSE41-LABEL: test_v32i16:
|
|
; SSE41: # %bb.0:
|
|
; SSE41-NEXT: pmaxuw %xmm3, %xmm1
|
|
; SSE41-NEXT: pmaxuw %xmm2, %xmm1
|
|
; SSE41-NEXT: pmaxuw %xmm0, %xmm1
|
|
; SSE41-NEXT: pcmpeqd %xmm0, %xmm0
|
|
; SSE41-NEXT: pxor %xmm1, %xmm0
|
|
; SSE41-NEXT: phminposuw %xmm0, %xmm0
|
|
; SSE41-NEXT: movd %xmm0, %eax
|
|
; SSE41-NEXT: notl %eax
|
|
; SSE41-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; SSE41-NEXT: retq
|
|
;
|
|
; AVX1-LABEL: test_v32i16:
|
|
; AVX1: # %bb.0:
|
|
; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm2
|
|
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm3
|
|
; AVX1-NEXT: vpmaxuw %xmm2, %xmm3, %xmm2
|
|
; AVX1-NEXT: vpmaxuw %xmm2, %xmm1, %xmm1
|
|
; AVX1-NEXT: vpmaxuw %xmm1, %xmm0, %xmm0
|
|
; AVX1-NEXT: vpcmpeqd %xmm1, %xmm1, %xmm1
|
|
; AVX1-NEXT: vpxor %xmm1, %xmm0, %xmm0
|
|
; AVX1-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX1-NEXT: vmovd %xmm0, %eax
|
|
; AVX1-NEXT: notl %eax
|
|
; AVX1-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; AVX1-NEXT: vzeroupper
|
|
; AVX1-NEXT: retq
|
|
;
|
|
; AVX2-LABEL: test_v32i16:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vpmaxuw %ymm1, %ymm0, %ymm0
|
|
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX2-NEXT: vpmaxuw %xmm1, %xmm0, %xmm0
|
|
; AVX2-NEXT: vpcmpeqd %xmm1, %xmm1, %xmm1
|
|
; AVX2-NEXT: vpxor %xmm1, %xmm0, %xmm0
|
|
; AVX2-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX2-NEXT: vmovd %xmm0, %eax
|
|
; AVX2-NEXT: notl %eax
|
|
; AVX2-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; AVX2-NEXT: vzeroupper
|
|
; AVX2-NEXT: retq
|
|
;
|
|
; AVX512BW-LABEL: test_v32i16:
|
|
; AVX512BW: # %bb.0:
|
|
; AVX512BW-NEXT: vextracti64x4 $1, %zmm0, %ymm1
|
|
; AVX512BW-NEXT: vpmaxuw %ymm1, %ymm0, %ymm0
|
|
; AVX512BW-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX512BW-NEXT: vpmaxuw %xmm1, %xmm0, %xmm0
|
|
; AVX512BW-NEXT: vpternlogq $15, %zmm0, %zmm0, %zmm0
|
|
; AVX512BW-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX512BW-NEXT: vmovd %xmm0, %eax
|
|
; AVX512BW-NEXT: notl %eax
|
|
; AVX512BW-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; AVX512BW-NEXT: vzeroupper
|
|
; AVX512BW-NEXT: retq
|
|
;
|
|
; AVX512VL-LABEL: test_v32i16:
|
|
; AVX512VL: # %bb.0:
|
|
; AVX512VL-NEXT: vextracti64x4 $1, %zmm0, %ymm1
|
|
; AVX512VL-NEXT: vpmaxuw %ymm1, %ymm0, %ymm0
|
|
; AVX512VL-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX512VL-NEXT: vpmaxuw %xmm1, %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vpternlogq $15, %xmm0, %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vmovd %xmm0, %eax
|
|
; AVX512VL-NEXT: notl %eax
|
|
; AVX512VL-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; AVX512VL-NEXT: vzeroupper
|
|
; AVX512VL-NEXT: retq
|
|
%1 = call i16 @llvm.experimental.vector.reduce.umax.v32i16(<32 x i16> %a0)
|
|
ret i16 %1
|
|
}
|
|
|
|
define i16 @test_v64i16(<64 x i16> %a0) {
|
|
; SSE2-LABEL: test_v64i16:
|
|
; SSE2: # %bb.0:
|
|
; SSE2-NEXT: movdqa {{.*#+}} xmm8 = [32768,32768,32768,32768,32768,32768,32768,32768]
|
|
; SSE2-NEXT: pxor %xmm8, %xmm6
|
|
; SSE2-NEXT: pxor %xmm8, %xmm2
|
|
; SSE2-NEXT: pmaxsw %xmm6, %xmm2
|
|
; SSE2-NEXT: pxor %xmm8, %xmm4
|
|
; SSE2-NEXT: pmaxsw %xmm2, %xmm4
|
|
; SSE2-NEXT: pxor %xmm8, %xmm0
|
|
; SSE2-NEXT: pxor %xmm8, %xmm7
|
|
; SSE2-NEXT: pxor %xmm8, %xmm3
|
|
; SSE2-NEXT: pmaxsw %xmm7, %xmm3
|
|
; SSE2-NEXT: pxor %xmm8, %xmm5
|
|
; SSE2-NEXT: pmaxsw %xmm3, %xmm5
|
|
; SSE2-NEXT: pxor %xmm8, %xmm1
|
|
; SSE2-NEXT: pmaxsw %xmm5, %xmm1
|
|
; SSE2-NEXT: pmaxsw %xmm4, %xmm1
|
|
; SSE2-NEXT: pmaxsw %xmm0, %xmm1
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm0
|
|
; SSE2-NEXT: pxor %xmm8, %xmm0
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
|
|
; SSE2-NEXT: pxor %xmm8, %xmm0
|
|
; SSE2-NEXT: pmaxsw %xmm1, %xmm0
|
|
; SSE2-NEXT: movdqa %xmm0, %xmm1
|
|
; SSE2-NEXT: pxor %xmm8, %xmm1
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,1,2,3]
|
|
; SSE2-NEXT: pxor %xmm8, %xmm1
|
|
; SSE2-NEXT: pmaxsw %xmm0, %xmm1
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm0
|
|
; SSE2-NEXT: pxor %xmm8, %xmm0
|
|
; SSE2-NEXT: psrld $16, %xmm0
|
|
; SSE2-NEXT: pxor %xmm8, %xmm0
|
|
; SSE2-NEXT: pmaxsw %xmm1, %xmm0
|
|
; SSE2-NEXT: movd %xmm0, %eax
|
|
; SSE2-NEXT: xorl $32768, %eax # imm = 0x8000
|
|
; SSE2-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; SSE2-NEXT: retq
|
|
;
|
|
; SSE41-LABEL: test_v64i16:
|
|
; SSE41: # %bb.0:
|
|
; SSE41-NEXT: pmaxuw %xmm7, %xmm3
|
|
; SSE41-NEXT: pmaxuw %xmm5, %xmm3
|
|
; SSE41-NEXT: pmaxuw %xmm1, %xmm3
|
|
; SSE41-NEXT: pmaxuw %xmm6, %xmm2
|
|
; SSE41-NEXT: pmaxuw %xmm4, %xmm2
|
|
; SSE41-NEXT: pmaxuw %xmm3, %xmm2
|
|
; SSE41-NEXT: pmaxuw %xmm0, %xmm2
|
|
; SSE41-NEXT: pcmpeqd %xmm0, %xmm0
|
|
; SSE41-NEXT: pxor %xmm2, %xmm0
|
|
; SSE41-NEXT: phminposuw %xmm0, %xmm0
|
|
; SSE41-NEXT: movd %xmm0, %eax
|
|
; SSE41-NEXT: notl %eax
|
|
; SSE41-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; SSE41-NEXT: retq
|
|
;
|
|
; AVX1-LABEL: test_v64i16:
|
|
; AVX1: # %bb.0:
|
|
; AVX1-NEXT: vextractf128 $1, %ymm3, %xmm4
|
|
; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm5
|
|
; AVX1-NEXT: vpmaxuw %xmm4, %xmm5, %xmm4
|
|
; AVX1-NEXT: vextractf128 $1, %ymm2, %xmm5
|
|
; AVX1-NEXT: vpmaxuw %xmm4, %xmm5, %xmm4
|
|
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm5
|
|
; AVX1-NEXT: vpmaxuw %xmm4, %xmm5, %xmm4
|
|
; AVX1-NEXT: vpmaxuw %xmm3, %xmm1, %xmm1
|
|
; AVX1-NEXT: vpmaxuw %xmm1, %xmm2, %xmm1
|
|
; AVX1-NEXT: vpmaxuw %xmm4, %xmm1, %xmm1
|
|
; AVX1-NEXT: vpmaxuw %xmm1, %xmm0, %xmm0
|
|
; AVX1-NEXT: vpcmpeqd %xmm1, %xmm1, %xmm1
|
|
; AVX1-NEXT: vpxor %xmm1, %xmm0, %xmm0
|
|
; AVX1-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX1-NEXT: vmovd %xmm0, %eax
|
|
; AVX1-NEXT: notl %eax
|
|
; AVX1-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; AVX1-NEXT: vzeroupper
|
|
; AVX1-NEXT: retq
|
|
;
|
|
; AVX2-LABEL: test_v64i16:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vpmaxuw %ymm3, %ymm1, %ymm1
|
|
; AVX2-NEXT: vpmaxuw %ymm1, %ymm2, %ymm1
|
|
; AVX2-NEXT: vpmaxuw %ymm1, %ymm0, %ymm0
|
|
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX2-NEXT: vpmaxuw %xmm1, %xmm0, %xmm0
|
|
; AVX2-NEXT: vpcmpeqd %xmm1, %xmm1, %xmm1
|
|
; AVX2-NEXT: vpxor %xmm1, %xmm0, %xmm0
|
|
; AVX2-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX2-NEXT: vmovd %xmm0, %eax
|
|
; AVX2-NEXT: notl %eax
|
|
; AVX2-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; AVX2-NEXT: vzeroupper
|
|
; AVX2-NEXT: retq
|
|
;
|
|
; AVX512BW-LABEL: test_v64i16:
|
|
; AVX512BW: # %bb.0:
|
|
; AVX512BW-NEXT: vpmaxuw %zmm1, %zmm0, %zmm0
|
|
; AVX512BW-NEXT: vextracti64x4 $1, %zmm0, %ymm1
|
|
; AVX512BW-NEXT: vpmaxuw %ymm1, %ymm0, %ymm0
|
|
; AVX512BW-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX512BW-NEXT: vpmaxuw %xmm1, %xmm0, %xmm0
|
|
; AVX512BW-NEXT: vpternlogq $15, %zmm0, %zmm0, %zmm0
|
|
; AVX512BW-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX512BW-NEXT: vmovd %xmm0, %eax
|
|
; AVX512BW-NEXT: notl %eax
|
|
; AVX512BW-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; AVX512BW-NEXT: vzeroupper
|
|
; AVX512BW-NEXT: retq
|
|
;
|
|
; AVX512VL-LABEL: test_v64i16:
|
|
; AVX512VL: # %bb.0:
|
|
; AVX512VL-NEXT: vpmaxuw %zmm1, %zmm0, %zmm0
|
|
; AVX512VL-NEXT: vextracti64x4 $1, %zmm0, %ymm1
|
|
; AVX512VL-NEXT: vpmaxuw %ymm1, %ymm0, %ymm0
|
|
; AVX512VL-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX512VL-NEXT: vpmaxuw %xmm1, %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vpternlogq $15, %xmm0, %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vmovd %xmm0, %eax
|
|
; AVX512VL-NEXT: notl %eax
|
|
; AVX512VL-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; AVX512VL-NEXT: vzeroupper
|
|
; AVX512VL-NEXT: retq
|
|
%1 = call i16 @llvm.experimental.vector.reduce.umax.v64i16(<64 x i16> %a0)
|
|
ret i16 %1
|
|
}
|
|
|
|
;
|
|
; vXi8
|
|
;
|
|
|
|
define i8 @test_v2i8(<2 x i8> %a0) {
|
|
; SSE2-LABEL: test_v2i8:
|
|
; SSE2: # %bb.0:
|
|
; SSE2-NEXT: movdqa {{.*#+}} xmm1 = [255,0,0,0,0,0,0,0,255,0,0,0,0,0,0,0]
|
|
; SSE2-NEXT: pand %xmm0, %xmm1
|
|
; SSE2-NEXT: pxor %xmm2, %xmm2
|
|
; SSE2-NEXT: punpckhbw {{.*#+}} xmm0 = xmm0[8],xmm2[8],xmm0[9],xmm2[9],xmm0[10],xmm2[10],xmm0[11],xmm2[11],xmm0[12],xmm2[12],xmm0[13],xmm2[13],xmm0[14],xmm2[14],xmm0[15],xmm2[15]
|
|
; SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
|
|
; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
|
|
; SSE2-NEXT: movdqa {{.*#+}} xmm2 = [2147483648,2147483648]
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm3
|
|
; SSE2-NEXT: por %xmm2, %xmm3
|
|
; SSE2-NEXT: por %xmm0, %xmm2
|
|
; SSE2-NEXT: movdqa %xmm3, %xmm4
|
|
; SSE2-NEXT: pcmpgtd %xmm2, %xmm4
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
|
|
; SSE2-NEXT: pcmpeqd %xmm3, %xmm2
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
|
|
; SSE2-NEXT: pand %xmm5, %xmm2
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
|
|
; SSE2-NEXT: por %xmm2, %xmm3
|
|
; SSE2-NEXT: pand %xmm3, %xmm1
|
|
; SSE2-NEXT: pandn %xmm0, %xmm3
|
|
; SSE2-NEXT: por %xmm1, %xmm3
|
|
; SSE2-NEXT: movd %xmm3, %eax
|
|
; SSE2-NEXT: # kill: def $al killed $al killed $eax
|
|
; SSE2-NEXT: retq
|
|
;
|
|
; SSE41-LABEL: test_v2i8:
|
|
; SSE41: # %bb.0:
|
|
; SSE41-NEXT: movdqa %xmm0, %xmm1
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
|
|
; SSE41-NEXT: pand {{.*}}(%rip), %xmm1
|
|
; SSE41-NEXT: pmovzxbq {{.*#+}} xmm2 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
|
|
; SSE41-NEXT: movdqa {{.*#+}} xmm3 = [2147483648,2147483648]
|
|
; SSE41-NEXT: movdqa %xmm1, %xmm4
|
|
; SSE41-NEXT: por %xmm3, %xmm4
|
|
; SSE41-NEXT: por %xmm2, %xmm3
|
|
; SSE41-NEXT: movdqa %xmm4, %xmm0
|
|
; SSE41-NEXT: pcmpeqd %xmm3, %xmm0
|
|
; SSE41-NEXT: pcmpgtd %xmm3, %xmm4
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm3 = xmm4[0,0,2,2]
|
|
; SSE41-NEXT: pand %xmm3, %xmm0
|
|
; SSE41-NEXT: por %xmm4, %xmm0
|
|
; SSE41-NEXT: blendvpd %xmm0, %xmm1, %xmm2
|
|
; SSE41-NEXT: pextrb $0, %xmm2, %eax
|
|
; SSE41-NEXT: # kill: def $al killed $al killed $eax
|
|
; SSE41-NEXT: retq
|
|
;
|
|
; AVX-LABEL: test_v2i8:
|
|
; AVX: # %bb.0:
|
|
; AVX-NEXT: vpand {{.*}}(%rip), %xmm0, %xmm1
|
|
; AVX-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
|
|
; AVX-NEXT: vpmovzxbq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
|
|
; AVX-NEXT: vpcmpgtq %xmm0, %xmm1, %xmm2
|
|
; AVX-NEXT: vblendvpd %xmm2, %xmm1, %xmm0, %xmm0
|
|
; AVX-NEXT: vpextrb $0, %xmm0, %eax
|
|
; AVX-NEXT: # kill: def $al killed $al killed $eax
|
|
; AVX-NEXT: retq
|
|
;
|
|
; AVX512BW-LABEL: test_v2i8:
|
|
; AVX512BW: # %bb.0:
|
|
; AVX512BW-NEXT: vpand {{.*}}(%rip), %xmm0, %xmm1
|
|
; AVX512BW-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
|
|
; AVX512BW-NEXT: vpmovzxbq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
|
|
; AVX512BW-NEXT: vpmaxuq %zmm0, %zmm1, %zmm0
|
|
; AVX512BW-NEXT: vpextrb $0, %xmm0, %eax
|
|
; AVX512BW-NEXT: # kill: def $al killed $al killed $eax
|
|
; AVX512BW-NEXT: vzeroupper
|
|
; AVX512BW-NEXT: retq
|
|
;
|
|
; AVX512VL-LABEL: test_v2i8:
|
|
; AVX512VL: # %bb.0:
|
|
; AVX512VL-NEXT: vpand {{.*}}(%rip), %xmm0, %xmm1
|
|
; AVX512VL-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
|
|
; AVX512VL-NEXT: vpmovzxbq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
|
|
; AVX512VL-NEXT: vpmaxuq %xmm0, %xmm1, %xmm0
|
|
; AVX512VL-NEXT: vpextrb $0, %xmm0, %eax
|
|
; AVX512VL-NEXT: # kill: def $al killed $al killed $eax
|
|
; AVX512VL-NEXT: retq
|
|
%1 = call i8 @llvm.experimental.vector.reduce.umax.v2i8(<2 x i8> %a0)
|
|
ret i8 %1
|
|
}
|
|
|
|
define i8 @test_v4i8(<4 x i8> %a0) {
|
|
; SSE2-LABEL: test_v4i8:
|
|
; SSE2: # %bb.0:
|
|
; SSE2-NEXT: movdqa {{.*#+}} xmm1 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm0[2,3,0,1]
|
|
; SSE2-NEXT: pand %xmm1, %xmm0
|
|
; SSE2-NEXT: pand {{.*}}(%rip), %xmm2
|
|
; SSE2-NEXT: movdqa %xmm0, %xmm3
|
|
; SSE2-NEXT: pcmpgtd %xmm2, %xmm3
|
|
; SSE2-NEXT: pand %xmm3, %xmm0
|
|
; SSE2-NEXT: pandn %xmm2, %xmm3
|
|
; SSE2-NEXT: por %xmm0, %xmm3
|
|
; SSE2-NEXT: pand %xmm3, %xmm1
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm3[1,1,2,3]
|
|
; SSE2-NEXT: pxor %xmm2, %xmm2
|
|
; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
|
|
; SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
|
|
; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm2
|
|
; SSE2-NEXT: pcmpgtd %xmm0, %xmm2
|
|
; SSE2-NEXT: pand %xmm2, %xmm1
|
|
; SSE2-NEXT: pandn %xmm0, %xmm2
|
|
; SSE2-NEXT: por %xmm1, %xmm2
|
|
; SSE2-NEXT: movd %xmm2, %eax
|
|
; SSE2-NEXT: # kill: def $al killed $al killed $eax
|
|
; SSE2-NEXT: retq
|
|
;
|
|
; SSE41-LABEL: test_v4i8:
|
|
; SSE41: # %bb.0:
|
|
; SSE41-NEXT: movdqa {{.*#+}} xmm1 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
|
|
; SSE41-NEXT: movdqa %xmm0, %xmm2
|
|
; SSE41-NEXT: pand %xmm1, %xmm2
|
|
; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[8],zero,zero,zero,xmm0[12],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
|
|
; SSE41-NEXT: pmaxud %xmm2, %xmm0
|
|
; SSE41-NEXT: pand %xmm0, %xmm1
|
|
; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
|
|
; SSE41-NEXT: pmovzxbq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
|
|
; SSE41-NEXT: pmaxud %xmm1, %xmm0
|
|
; SSE41-NEXT: pextrb $0, %xmm0, %eax
|
|
; SSE41-NEXT: # kill: def $al killed $al killed $eax
|
|
; SSE41-NEXT: retq
|
|
;
|
|
; AVX1-LABEL: test_v4i8:
|
|
; AVX1: # %bb.0:
|
|
; AVX1-NEXT: vbroadcastss {{.*#+}} xmm1 = [3.57331108E-43,3.57331108E-43,3.57331108E-43,3.57331108E-43]
|
|
; AVX1-NEXT: vandps %xmm1, %xmm0, %xmm2
|
|
; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8],zero,zero,zero,xmm0[12],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
|
|
; AVX1-NEXT: vpmaxud %xmm0, %xmm2, %xmm0
|
|
; AVX1-NEXT: vpand %xmm1, %xmm0, %xmm1
|
|
; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
|
|
; AVX1-NEXT: vpmovzxbq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
|
|
; AVX1-NEXT: vpmaxud %xmm0, %xmm1, %xmm0
|
|
; AVX1-NEXT: vpextrb $0, %xmm0, %eax
|
|
; AVX1-NEXT: # kill: def $al killed $al killed $eax
|
|
; AVX1-NEXT: retq
|
|
;
|
|
; AVX2-LABEL: test_v4i8:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vpbroadcastd {{.*#+}} xmm1 = [255,255,255,255]
|
|
; AVX2-NEXT: vpand %xmm1, %xmm0, %xmm2
|
|
; AVX2-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8],zero,zero,zero,xmm0[12],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
|
|
; AVX2-NEXT: vpmaxud %xmm0, %xmm2, %xmm0
|
|
; AVX2-NEXT: vpand %xmm1, %xmm0, %xmm1
|
|
; AVX2-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
|
|
; AVX2-NEXT: vpmovzxbq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
|
|
; AVX2-NEXT: vpmaxud %xmm0, %xmm1, %xmm0
|
|
; AVX2-NEXT: vpextrb $0, %xmm0, %eax
|
|
; AVX2-NEXT: # kill: def $al killed $al killed $eax
|
|
; AVX2-NEXT: retq
|
|
;
|
|
; AVX512-LABEL: test_v4i8:
|
|
; AVX512: # %bb.0:
|
|
; AVX512-NEXT: vpbroadcastd {{.*#+}} xmm1 = [255,255,255,255]
|
|
; AVX512-NEXT: vpand %xmm1, %xmm0, %xmm2
|
|
; AVX512-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8],zero,zero,zero,xmm0[12],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
|
|
; AVX512-NEXT: vpmaxud %xmm0, %xmm2, %xmm0
|
|
; AVX512-NEXT: vpand %xmm1, %xmm0, %xmm1
|
|
; AVX512-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
|
|
; AVX512-NEXT: vpmovzxbq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
|
|
; AVX512-NEXT: vpmaxud %xmm0, %xmm1, %xmm0
|
|
; AVX512-NEXT: vpextrb $0, %xmm0, %eax
|
|
; AVX512-NEXT: # kill: def $al killed $al killed $eax
|
|
; AVX512-NEXT: retq
|
|
%1 = call i8 @llvm.experimental.vector.reduce.umax.v4i8(<4 x i8> %a0)
|
|
ret i8 %1
|
|
}
|
|
|
|
define i8 @test_v8i8(<8 x i8> %a0) {
|
|
; SSE2-LABEL: test_v8i8:
|
|
; SSE2: # %bb.0:
|
|
; SSE2-NEXT: movdqa {{.*#+}} xmm1 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm0[2,3,0,1]
|
|
; SSE2-NEXT: pand %xmm1, %xmm0
|
|
; SSE2-NEXT: pand {{.*}}(%rip), %xmm2
|
|
; SSE2-NEXT: pmaxsw %xmm0, %xmm2
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm2[1,1,2,3]
|
|
; SSE2-NEXT: pand %xmm1, %xmm2
|
|
; SSE2-NEXT: pand {{.*}}(%rip), %xmm0
|
|
; SSE2-NEXT: pmaxsw %xmm2, %xmm0
|
|
; SSE2-NEXT: pand %xmm0, %xmm1
|
|
; SSE2-NEXT: psrld $16, %xmm0
|
|
; SSE2-NEXT: pxor %xmm2, %xmm2
|
|
; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
|
|
; SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
|
|
; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
|
|
; SSE2-NEXT: pmaxsw %xmm1, %xmm0
|
|
; SSE2-NEXT: movd %xmm0, %eax
|
|
; SSE2-NEXT: # kill: def $al killed $al killed $eax
|
|
; SSE2-NEXT: retq
|
|
;
|
|
; SSE41-LABEL: test_v8i8:
|
|
; SSE41: # %bb.0:
|
|
; SSE41-NEXT: movdqa {{.*#+}} xmm1 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
|
|
; SSE41-NEXT: movdqa %xmm0, %xmm2
|
|
; SSE41-NEXT: pand %xmm1, %xmm2
|
|
; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[8],zero,xmm0[10],zero,xmm0[12],zero,xmm0[14],zero,zero,zero,zero,zero,zero,zero,zero,zero
|
|
; SSE41-NEXT: pmaxuw %xmm2, %xmm0
|
|
; SSE41-NEXT: movdqa %xmm0, %xmm2
|
|
; SSE41-NEXT: pand %xmm1, %xmm2
|
|
; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[4],zero,xmm0[6],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
|
|
; SSE41-NEXT: pmaxuw %xmm2, %xmm0
|
|
; SSE41-NEXT: pand %xmm0, %xmm1
|
|
; SSE41-NEXT: psrld $16, %xmm0
|
|
; SSE41-NEXT: pmovzxbq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
|
|
; SSE41-NEXT: pmaxuw %xmm1, %xmm0
|
|
; SSE41-NEXT: pextrb $0, %xmm0, %eax
|
|
; SSE41-NEXT: # kill: def $al killed $al killed $eax
|
|
; SSE41-NEXT: retq
|
|
;
|
|
; AVX1-LABEL: test_v8i8:
|
|
; AVX1: # %bb.0:
|
|
; AVX1-NEXT: vmovdqa {{.*#+}} xmm1 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
|
|
; AVX1-NEXT: vpand %xmm1, %xmm0, %xmm2
|
|
; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8],zero,xmm0[10],zero,xmm0[12],zero,xmm0[14],zero,zero,zero,zero,zero,zero,zero,zero,zero
|
|
; AVX1-NEXT: vpmaxuw %xmm0, %xmm2, %xmm0
|
|
; AVX1-NEXT: vpand %xmm1, %xmm0, %xmm2
|
|
; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[4],zero,xmm0[6],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
|
|
; AVX1-NEXT: vpmaxuw %xmm0, %xmm2, %xmm0
|
|
; AVX1-NEXT: vpand %xmm1, %xmm0, %xmm1
|
|
; AVX1-NEXT: vpsrld $16, %xmm0, %xmm0
|
|
; AVX1-NEXT: vpmovzxbq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
|
|
; AVX1-NEXT: vpmaxuw %xmm0, %xmm1, %xmm0
|
|
; AVX1-NEXT: vpextrb $0, %xmm0, %eax
|
|
; AVX1-NEXT: # kill: def $al killed $al killed $eax
|
|
; AVX1-NEXT: retq
|
|
;
|
|
; AVX2-LABEL: test_v8i8:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vpbroadcastw {{.*#+}} xmm1 = [255,255,255,255,255,255,255,255]
|
|
; AVX2-NEXT: vpand %xmm1, %xmm0, %xmm2
|
|
; AVX2-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8],zero,xmm0[10],zero,xmm0[12],zero,xmm0[14],zero,zero,zero,zero,zero,zero,zero,zero,zero
|
|
; AVX2-NEXT: vpmaxuw %xmm0, %xmm2, %xmm0
|
|
; AVX2-NEXT: vpand %xmm1, %xmm0, %xmm2
|
|
; AVX2-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[4],zero,xmm0[6],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
|
|
; AVX2-NEXT: vpmaxuw %xmm0, %xmm2, %xmm0
|
|
; AVX2-NEXT: vpand %xmm1, %xmm0, %xmm1
|
|
; AVX2-NEXT: vpsrld $16, %xmm0, %xmm0
|
|
; AVX2-NEXT: vpmovzxbq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
|
|
; AVX2-NEXT: vpmaxuw %xmm0, %xmm1, %xmm0
|
|
; AVX2-NEXT: vpextrb $0, %xmm0, %eax
|
|
; AVX2-NEXT: # kill: def $al killed $al killed $eax
|
|
; AVX2-NEXT: retq
|
|
;
|
|
; AVX512-LABEL: test_v8i8:
|
|
; AVX512: # %bb.0:
|
|
; AVX512-NEXT: vpbroadcastw {{.*#+}} xmm1 = [255,255,255,255,255,255,255,255]
|
|
; AVX512-NEXT: vpand %xmm1, %xmm0, %xmm2
|
|
; AVX512-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8],zero,xmm0[10],zero,xmm0[12],zero,xmm0[14],zero,zero,zero,zero,zero,zero,zero,zero,zero
|
|
; AVX512-NEXT: vpmaxuw %xmm0, %xmm2, %xmm0
|
|
; AVX512-NEXT: vpand %xmm1, %xmm0, %xmm2
|
|
; AVX512-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[4],zero,xmm0[6],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
|
|
; AVX512-NEXT: vpmaxuw %xmm0, %xmm2, %xmm0
|
|
; AVX512-NEXT: vpand %xmm1, %xmm0, %xmm1
|
|
; AVX512-NEXT: vpsrld $16, %xmm0, %xmm0
|
|
; AVX512-NEXT: vpmovzxbq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
|
|
; AVX512-NEXT: vpmaxuw %xmm0, %xmm1, %xmm0
|
|
; AVX512-NEXT: vpextrb $0, %xmm0, %eax
|
|
; AVX512-NEXT: # kill: def $al killed $al killed $eax
|
|
; AVX512-NEXT: retq
|
|
%1 = call i8 @llvm.experimental.vector.reduce.umax.v8i8(<8 x i8> %a0)
|
|
ret i8 %1
|
|
}
|
|
|
|
define i8 @test_v16i8(<16 x i8> %a0) {
|
|
; SSE2-LABEL: test_v16i8:
|
|
; SSE2: # %bb.0:
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; SSE2-NEXT: pmaxub %xmm0, %xmm1
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,2,3]
|
|
; SSE2-NEXT: pmaxub %xmm1, %xmm0
|
|
; SSE2-NEXT: movdqa %xmm0, %xmm1
|
|
; SSE2-NEXT: psrld $16, %xmm1
|
|
; SSE2-NEXT: pmaxub %xmm0, %xmm1
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm0
|
|
; SSE2-NEXT: psrlw $8, %xmm0
|
|
; SSE2-NEXT: pmaxub %xmm1, %xmm0
|
|
; SSE2-NEXT: movd %xmm0, %eax
|
|
; SSE2-NEXT: # kill: def $al killed $al killed $eax
|
|
; SSE2-NEXT: retq
|
|
;
|
|
; SSE41-LABEL: test_v16i8:
|
|
; SSE41: # %bb.0:
|
|
; SSE41-NEXT: pcmpeqd %xmm1, %xmm1
|
|
; SSE41-NEXT: pxor %xmm0, %xmm1
|
|
; SSE41-NEXT: movdqa %xmm1, %xmm0
|
|
; SSE41-NEXT: psrlw $8, %xmm0
|
|
; SSE41-NEXT: pminub %xmm1, %xmm0
|
|
; SSE41-NEXT: phminposuw %xmm0, %xmm0
|
|
; SSE41-NEXT: pextrb $0, %xmm0, %eax
|
|
; SSE41-NEXT: notb %al
|
|
; SSE41-NEXT: # kill: def $al killed $al killed $eax
|
|
; SSE41-NEXT: retq
|
|
;
|
|
; AVX-LABEL: test_v16i8:
|
|
; AVX: # %bb.0:
|
|
; AVX-NEXT: vpcmpeqd %xmm1, %xmm1, %xmm1
|
|
; AVX-NEXT: vpxor %xmm1, %xmm0, %xmm0
|
|
; AVX-NEXT: vpsrlw $8, %xmm0, %xmm1
|
|
; AVX-NEXT: vpminub %xmm1, %xmm0, %xmm0
|
|
; AVX-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX-NEXT: vpextrb $0, %xmm0, %eax
|
|
; AVX-NEXT: notb %al
|
|
; AVX-NEXT: # kill: def $al killed $al killed $eax
|
|
; AVX-NEXT: retq
|
|
;
|
|
; AVX512BW-LABEL: test_v16i8:
|
|
; AVX512BW: # %bb.0:
|
|
; AVX512BW-NEXT: # kill: def $xmm0 killed $xmm0 def $zmm0
|
|
; AVX512BW-NEXT: vpternlogq $15, %zmm0, %zmm0, %zmm0
|
|
; AVX512BW-NEXT: vpsrlw $8, %xmm0, %xmm1
|
|
; AVX512BW-NEXT: vpminub %xmm1, %xmm0, %xmm0
|
|
; AVX512BW-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX512BW-NEXT: vpextrb $0, %xmm0, %eax
|
|
; AVX512BW-NEXT: notb %al
|
|
; AVX512BW-NEXT: # kill: def $al killed $al killed $eax
|
|
; AVX512BW-NEXT: vzeroupper
|
|
; AVX512BW-NEXT: retq
|
|
;
|
|
; AVX512VL-LABEL: test_v16i8:
|
|
; AVX512VL: # %bb.0:
|
|
; AVX512VL-NEXT: vpternlogq $15, %xmm0, %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vpsrlw $8, %xmm0, %xmm1
|
|
; AVX512VL-NEXT: vpminub %xmm1, %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vpextrb $0, %xmm0, %eax
|
|
; AVX512VL-NEXT: notb %al
|
|
; AVX512VL-NEXT: # kill: def $al killed $al killed $eax
|
|
; AVX512VL-NEXT: retq
|
|
%1 = call i8 @llvm.experimental.vector.reduce.umax.v16i8(<16 x i8> %a0)
|
|
ret i8 %1
|
|
}
|
|
|
|
define i8 @test_v32i8(<32 x i8> %a0) {
|
|
; SSE2-LABEL: test_v32i8:
|
|
; SSE2: # %bb.0:
|
|
; SSE2-NEXT: pmaxub %xmm1, %xmm0
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
|
|
; SSE2-NEXT: pmaxub %xmm0, %xmm1
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,2,3]
|
|
; SSE2-NEXT: pmaxub %xmm1, %xmm0
|
|
; SSE2-NEXT: movdqa %xmm0, %xmm1
|
|
; SSE2-NEXT: psrld $16, %xmm1
|
|
; SSE2-NEXT: pmaxub %xmm0, %xmm1
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm0
|
|
; SSE2-NEXT: psrlw $8, %xmm0
|
|
; SSE2-NEXT: pmaxub %xmm1, %xmm0
|
|
; SSE2-NEXT: movd %xmm0, %eax
|
|
; SSE2-NEXT: # kill: def $al killed $al killed $eax
|
|
; SSE2-NEXT: retq
|
|
;
|
|
; SSE41-LABEL: test_v32i8:
|
|
; SSE41: # %bb.0:
|
|
; SSE41-NEXT: pmaxub %xmm1, %xmm0
|
|
; SSE41-NEXT: pcmpeqd %xmm1, %xmm1
|
|
; SSE41-NEXT: pxor %xmm0, %xmm1
|
|
; SSE41-NEXT: movdqa %xmm1, %xmm0
|
|
; SSE41-NEXT: psrlw $8, %xmm0
|
|
; SSE41-NEXT: pminub %xmm1, %xmm0
|
|
; SSE41-NEXT: phminposuw %xmm0, %xmm0
|
|
; SSE41-NEXT: pextrb $0, %xmm0, %eax
|
|
; SSE41-NEXT: notb %al
|
|
; SSE41-NEXT: # kill: def $al killed $al killed $eax
|
|
; SSE41-NEXT: retq
|
|
;
|
|
; AVX1-LABEL: test_v32i8:
|
|
; AVX1: # %bb.0:
|
|
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
|
|
; AVX1-NEXT: vpmaxub %xmm1, %xmm0, %xmm0
|
|
; AVX1-NEXT: vpcmpeqd %xmm1, %xmm1, %xmm1
|
|
; AVX1-NEXT: vpxor %xmm1, %xmm0, %xmm0
|
|
; AVX1-NEXT: vpsrlw $8, %xmm0, %xmm1
|
|
; AVX1-NEXT: vpminub %xmm1, %xmm0, %xmm0
|
|
; AVX1-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX1-NEXT: vpextrb $0, %xmm0, %eax
|
|
; AVX1-NEXT: notb %al
|
|
; AVX1-NEXT: # kill: def $al killed $al killed $eax
|
|
; AVX1-NEXT: vzeroupper
|
|
; AVX1-NEXT: retq
|
|
;
|
|
; AVX2-LABEL: test_v32i8:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX2-NEXT: vpmaxub %xmm1, %xmm0, %xmm0
|
|
; AVX2-NEXT: vpcmpeqd %xmm1, %xmm1, %xmm1
|
|
; AVX2-NEXT: vpxor %xmm1, %xmm0, %xmm0
|
|
; AVX2-NEXT: vpsrlw $8, %xmm0, %xmm1
|
|
; AVX2-NEXT: vpminub %xmm1, %xmm0, %xmm0
|
|
; AVX2-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX2-NEXT: vpextrb $0, %xmm0, %eax
|
|
; AVX2-NEXT: notb %al
|
|
; AVX2-NEXT: # kill: def $al killed $al killed $eax
|
|
; AVX2-NEXT: vzeroupper
|
|
; AVX2-NEXT: retq
|
|
;
|
|
; AVX512BW-LABEL: test_v32i8:
|
|
; AVX512BW: # %bb.0:
|
|
; AVX512BW-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX512BW-NEXT: vpmaxub %xmm1, %xmm0, %xmm0
|
|
; AVX512BW-NEXT: vpternlogq $15, %zmm0, %zmm0, %zmm0
|
|
; AVX512BW-NEXT: vpsrlw $8, %xmm0, %xmm1
|
|
; AVX512BW-NEXT: vpminub %xmm1, %xmm0, %xmm0
|
|
; AVX512BW-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX512BW-NEXT: vpextrb $0, %xmm0, %eax
|
|
; AVX512BW-NEXT: notb %al
|
|
; AVX512BW-NEXT: # kill: def $al killed $al killed $eax
|
|
; AVX512BW-NEXT: vzeroupper
|
|
; AVX512BW-NEXT: retq
|
|
;
|
|
; AVX512VL-LABEL: test_v32i8:
|
|
; AVX512VL: # %bb.0:
|
|
; AVX512VL-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX512VL-NEXT: vpmaxub %xmm1, %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vpternlogq $15, %xmm0, %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vpsrlw $8, %xmm0, %xmm1
|
|
; AVX512VL-NEXT: vpminub %xmm1, %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vpextrb $0, %xmm0, %eax
|
|
; AVX512VL-NEXT: notb %al
|
|
; AVX512VL-NEXT: # kill: def $al killed $al killed $eax
|
|
; AVX512VL-NEXT: vzeroupper
|
|
; AVX512VL-NEXT: retq
|
|
%1 = call i8 @llvm.experimental.vector.reduce.umax.v32i8(<32 x i8> %a0)
|
|
ret i8 %1
|
|
}
|
|
|
|
define i8 @test_v64i8(<64 x i8> %a0) {
|
|
; SSE2-LABEL: test_v64i8:
|
|
; SSE2: # %bb.0:
|
|
; SSE2-NEXT: pmaxub %xmm3, %xmm1
|
|
; SSE2-NEXT: pmaxub %xmm2, %xmm1
|
|
; SSE2-NEXT: pmaxub %xmm0, %xmm1
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm1[2,3,0,1]
|
|
; SSE2-NEXT: pmaxub %xmm1, %xmm0
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
|
|
; SSE2-NEXT: pmaxub %xmm0, %xmm1
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm0
|
|
; SSE2-NEXT: psrld $16, %xmm0
|
|
; SSE2-NEXT: pmaxub %xmm1, %xmm0
|
|
; SSE2-NEXT: movdqa %xmm0, %xmm1
|
|
; SSE2-NEXT: psrlw $8, %xmm1
|
|
; SSE2-NEXT: pmaxub %xmm0, %xmm1
|
|
; SSE2-NEXT: movd %xmm1, %eax
|
|
; SSE2-NEXT: # kill: def $al killed $al killed $eax
|
|
; SSE2-NEXT: retq
|
|
;
|
|
; SSE41-LABEL: test_v64i8:
|
|
; SSE41: # %bb.0:
|
|
; SSE41-NEXT: pmaxub %xmm3, %xmm1
|
|
; SSE41-NEXT: pmaxub %xmm2, %xmm1
|
|
; SSE41-NEXT: pmaxub %xmm0, %xmm1
|
|
; SSE41-NEXT: pcmpeqd %xmm0, %xmm0
|
|
; SSE41-NEXT: pxor %xmm1, %xmm0
|
|
; SSE41-NEXT: movdqa %xmm0, %xmm1
|
|
; SSE41-NEXT: psrlw $8, %xmm1
|
|
; SSE41-NEXT: pminub %xmm0, %xmm1
|
|
; SSE41-NEXT: phminposuw %xmm1, %xmm0
|
|
; SSE41-NEXT: pextrb $0, %xmm0, %eax
|
|
; SSE41-NEXT: notb %al
|
|
; SSE41-NEXT: # kill: def $al killed $al killed $eax
|
|
; SSE41-NEXT: retq
|
|
;
|
|
; AVX1-LABEL: test_v64i8:
|
|
; AVX1: # %bb.0:
|
|
; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm2
|
|
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm3
|
|
; AVX1-NEXT: vpmaxub %xmm2, %xmm3, %xmm2
|
|
; AVX1-NEXT: vpmaxub %xmm2, %xmm1, %xmm1
|
|
; AVX1-NEXT: vpmaxub %xmm1, %xmm0, %xmm0
|
|
; AVX1-NEXT: vpcmpeqd %xmm1, %xmm1, %xmm1
|
|
; AVX1-NEXT: vpxor %xmm1, %xmm0, %xmm0
|
|
; AVX1-NEXT: vpsrlw $8, %xmm0, %xmm1
|
|
; AVX1-NEXT: vpminub %xmm1, %xmm0, %xmm0
|
|
; AVX1-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX1-NEXT: vpextrb $0, %xmm0, %eax
|
|
; AVX1-NEXT: notb %al
|
|
; AVX1-NEXT: # kill: def $al killed $al killed $eax
|
|
; AVX1-NEXT: vzeroupper
|
|
; AVX1-NEXT: retq
|
|
;
|
|
; AVX2-LABEL: test_v64i8:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vpmaxub %ymm1, %ymm0, %ymm0
|
|
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX2-NEXT: vpmaxub %xmm1, %xmm0, %xmm0
|
|
; AVX2-NEXT: vpcmpeqd %xmm1, %xmm1, %xmm1
|
|
; AVX2-NEXT: vpxor %xmm1, %xmm0, %xmm0
|
|
; AVX2-NEXT: vpsrlw $8, %xmm0, %xmm1
|
|
; AVX2-NEXT: vpminub %xmm1, %xmm0, %xmm0
|
|
; AVX2-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX2-NEXT: vpextrb $0, %xmm0, %eax
|
|
; AVX2-NEXT: notb %al
|
|
; AVX2-NEXT: # kill: def $al killed $al killed $eax
|
|
; AVX2-NEXT: vzeroupper
|
|
; AVX2-NEXT: retq
|
|
;
|
|
; AVX512BW-LABEL: test_v64i8:
|
|
; AVX512BW: # %bb.0:
|
|
; AVX512BW-NEXT: vextracti64x4 $1, %zmm0, %ymm1
|
|
; AVX512BW-NEXT: vpmaxub %ymm1, %ymm0, %ymm0
|
|
; AVX512BW-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX512BW-NEXT: vpmaxub %xmm1, %xmm0, %xmm0
|
|
; AVX512BW-NEXT: vpternlogq $15, %zmm0, %zmm0, %zmm0
|
|
; AVX512BW-NEXT: vpsrlw $8, %xmm0, %xmm1
|
|
; AVX512BW-NEXT: vpminub %xmm1, %xmm0, %xmm0
|
|
; AVX512BW-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX512BW-NEXT: vpextrb $0, %xmm0, %eax
|
|
; AVX512BW-NEXT: notb %al
|
|
; AVX512BW-NEXT: # kill: def $al killed $al killed $eax
|
|
; AVX512BW-NEXT: vzeroupper
|
|
; AVX512BW-NEXT: retq
|
|
;
|
|
; AVX512VL-LABEL: test_v64i8:
|
|
; AVX512VL: # %bb.0:
|
|
; AVX512VL-NEXT: vextracti64x4 $1, %zmm0, %ymm1
|
|
; AVX512VL-NEXT: vpmaxub %ymm1, %ymm0, %ymm0
|
|
; AVX512VL-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX512VL-NEXT: vpmaxub %xmm1, %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vpternlogq $15, %xmm0, %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vpsrlw $8, %xmm0, %xmm1
|
|
; AVX512VL-NEXT: vpminub %xmm1, %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vpextrb $0, %xmm0, %eax
|
|
; AVX512VL-NEXT: notb %al
|
|
; AVX512VL-NEXT: # kill: def $al killed $al killed $eax
|
|
; AVX512VL-NEXT: vzeroupper
|
|
; AVX512VL-NEXT: retq
|
|
%1 = call i8 @llvm.experimental.vector.reduce.umax.v64i8(<64 x i8> %a0)
|
|
ret i8 %1
|
|
}
|
|
|
|
define i8 @test_v128i8(<128 x i8> %a0) {
|
|
; SSE2-LABEL: test_v128i8:
|
|
; SSE2: # %bb.0:
|
|
; SSE2-NEXT: pmaxub %xmm6, %xmm2
|
|
; SSE2-NEXT: pmaxub %xmm7, %xmm3
|
|
; SSE2-NEXT: pmaxub %xmm5, %xmm3
|
|
; SSE2-NEXT: pmaxub %xmm1, %xmm3
|
|
; SSE2-NEXT: pmaxub %xmm4, %xmm2
|
|
; SSE2-NEXT: pmaxub %xmm3, %xmm2
|
|
; SSE2-NEXT: pmaxub %xmm0, %xmm2
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm2[2,3,0,1]
|
|
; SSE2-NEXT: pmaxub %xmm2, %xmm0
|
|
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
|
|
; SSE2-NEXT: pmaxub %xmm0, %xmm1
|
|
; SSE2-NEXT: movdqa %xmm1, %xmm0
|
|
; SSE2-NEXT: psrld $16, %xmm0
|
|
; SSE2-NEXT: pmaxub %xmm1, %xmm0
|
|
; SSE2-NEXT: movdqa %xmm0, %xmm1
|
|
; SSE2-NEXT: psrlw $8, %xmm1
|
|
; SSE2-NEXT: pmaxub %xmm0, %xmm1
|
|
; SSE2-NEXT: movd %xmm1, %eax
|
|
; SSE2-NEXT: # kill: def $al killed $al killed $eax
|
|
; SSE2-NEXT: retq
|
|
;
|
|
; SSE41-LABEL: test_v128i8:
|
|
; SSE41: # %bb.0:
|
|
; SSE41-NEXT: pmaxub %xmm7, %xmm3
|
|
; SSE41-NEXT: pmaxub %xmm5, %xmm3
|
|
; SSE41-NEXT: pmaxub %xmm1, %xmm3
|
|
; SSE41-NEXT: pmaxub %xmm6, %xmm2
|
|
; SSE41-NEXT: pmaxub %xmm4, %xmm2
|
|
; SSE41-NEXT: pmaxub %xmm3, %xmm2
|
|
; SSE41-NEXT: pmaxub %xmm0, %xmm2
|
|
; SSE41-NEXT: pcmpeqd %xmm0, %xmm0
|
|
; SSE41-NEXT: pxor %xmm2, %xmm0
|
|
; SSE41-NEXT: movdqa %xmm0, %xmm1
|
|
; SSE41-NEXT: psrlw $8, %xmm1
|
|
; SSE41-NEXT: pminub %xmm0, %xmm1
|
|
; SSE41-NEXT: phminposuw %xmm1, %xmm0
|
|
; SSE41-NEXT: pextrb $0, %xmm0, %eax
|
|
; SSE41-NEXT: notb %al
|
|
; SSE41-NEXT: # kill: def $al killed $al killed $eax
|
|
; SSE41-NEXT: retq
|
|
;
|
|
; AVX1-LABEL: test_v128i8:
|
|
; AVX1: # %bb.0:
|
|
; AVX1-NEXT: vextractf128 $1, %ymm3, %xmm4
|
|
; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm5
|
|
; AVX1-NEXT: vpmaxub %xmm4, %xmm5, %xmm4
|
|
; AVX1-NEXT: vextractf128 $1, %ymm2, %xmm5
|
|
; AVX1-NEXT: vpmaxub %xmm4, %xmm5, %xmm4
|
|
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm5
|
|
; AVX1-NEXT: vpmaxub %xmm4, %xmm5, %xmm4
|
|
; AVX1-NEXT: vpmaxub %xmm3, %xmm1, %xmm1
|
|
; AVX1-NEXT: vpmaxub %xmm1, %xmm2, %xmm1
|
|
; AVX1-NEXT: vpmaxub %xmm4, %xmm1, %xmm1
|
|
; AVX1-NEXT: vpmaxub %xmm1, %xmm0, %xmm0
|
|
; AVX1-NEXT: vpcmpeqd %xmm1, %xmm1, %xmm1
|
|
; AVX1-NEXT: vpxor %xmm1, %xmm0, %xmm0
|
|
; AVX1-NEXT: vpsrlw $8, %xmm0, %xmm1
|
|
; AVX1-NEXT: vpminub %xmm1, %xmm0, %xmm0
|
|
; AVX1-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX1-NEXT: vpextrb $0, %xmm0, %eax
|
|
; AVX1-NEXT: notb %al
|
|
; AVX1-NEXT: # kill: def $al killed $al killed $eax
|
|
; AVX1-NEXT: vzeroupper
|
|
; AVX1-NEXT: retq
|
|
;
|
|
; AVX2-LABEL: test_v128i8:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vpmaxub %ymm3, %ymm1, %ymm1
|
|
; AVX2-NEXT: vpmaxub %ymm1, %ymm2, %ymm1
|
|
; AVX2-NEXT: vpmaxub %ymm1, %ymm0, %ymm0
|
|
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX2-NEXT: vpmaxub %xmm1, %xmm0, %xmm0
|
|
; AVX2-NEXT: vpcmpeqd %xmm1, %xmm1, %xmm1
|
|
; AVX2-NEXT: vpxor %xmm1, %xmm0, %xmm0
|
|
; AVX2-NEXT: vpsrlw $8, %xmm0, %xmm1
|
|
; AVX2-NEXT: vpminub %xmm1, %xmm0, %xmm0
|
|
; AVX2-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX2-NEXT: vpextrb $0, %xmm0, %eax
|
|
; AVX2-NEXT: notb %al
|
|
; AVX2-NEXT: # kill: def $al killed $al killed $eax
|
|
; AVX2-NEXT: vzeroupper
|
|
; AVX2-NEXT: retq
|
|
;
|
|
; AVX512BW-LABEL: test_v128i8:
|
|
; AVX512BW: # %bb.0:
|
|
; AVX512BW-NEXT: vpmaxub %zmm1, %zmm0, %zmm0
|
|
; AVX512BW-NEXT: vextracti64x4 $1, %zmm0, %ymm1
|
|
; AVX512BW-NEXT: vpmaxub %ymm1, %ymm0, %ymm0
|
|
; AVX512BW-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX512BW-NEXT: vpmaxub %xmm1, %xmm0, %xmm0
|
|
; AVX512BW-NEXT: vpternlogq $15, %zmm0, %zmm0, %zmm0
|
|
; AVX512BW-NEXT: vpsrlw $8, %xmm0, %xmm1
|
|
; AVX512BW-NEXT: vpminub %xmm1, %xmm0, %xmm0
|
|
; AVX512BW-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX512BW-NEXT: vpextrb $0, %xmm0, %eax
|
|
; AVX512BW-NEXT: notb %al
|
|
; AVX512BW-NEXT: # kill: def $al killed $al killed $eax
|
|
; AVX512BW-NEXT: vzeroupper
|
|
; AVX512BW-NEXT: retq
|
|
;
|
|
; AVX512VL-LABEL: test_v128i8:
|
|
; AVX512VL: # %bb.0:
|
|
; AVX512VL-NEXT: vpmaxub %zmm1, %zmm0, %zmm0
|
|
; AVX512VL-NEXT: vextracti64x4 $1, %zmm0, %ymm1
|
|
; AVX512VL-NEXT: vpmaxub %ymm1, %ymm0, %ymm0
|
|
; AVX512VL-NEXT: vextracti128 $1, %ymm0, %xmm1
|
|
; AVX512VL-NEXT: vpmaxub %xmm1, %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vpternlogq $15, %xmm0, %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vpsrlw $8, %xmm0, %xmm1
|
|
; AVX512VL-NEXT: vpminub %xmm1, %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vphminposuw %xmm0, %xmm0
|
|
; AVX512VL-NEXT: vpextrb $0, %xmm0, %eax
|
|
; AVX512VL-NEXT: notb %al
|
|
; AVX512VL-NEXT: # kill: def $al killed $al killed $eax
|
|
; AVX512VL-NEXT: vzeroupper
|
|
; AVX512VL-NEXT: retq
|
|
%1 = call i8 @llvm.experimental.vector.reduce.umax.v128i8(<128 x i8> %a0)
|
|
ret i8 %1
|
|
}
|
|
|
|
declare i64 @llvm.experimental.vector.reduce.umax.v2i64(<2 x i64>)
|
|
declare i64 @llvm.experimental.vector.reduce.umax.v4i64(<4 x i64>)
|
|
declare i64 @llvm.experimental.vector.reduce.umax.v8i64(<8 x i64>)
|
|
declare i64 @llvm.experimental.vector.reduce.umax.v16i64(<16 x i64>)
|
|
|
|
declare i32 @llvm.experimental.vector.reduce.umax.v2i32(<2 x i32>)
|
|
declare i32 @llvm.experimental.vector.reduce.umax.v4i32(<4 x i32>)
|
|
declare i32 @llvm.experimental.vector.reduce.umax.v8i32(<8 x i32>)
|
|
declare i32 @llvm.experimental.vector.reduce.umax.v16i32(<16 x i32>)
|
|
declare i32 @llvm.experimental.vector.reduce.umax.v32i32(<32 x i32>)
|
|
|
|
declare i16 @llvm.experimental.vector.reduce.umax.v2i16(<2 x i16>)
|
|
declare i16 @llvm.experimental.vector.reduce.umax.v4i16(<4 x i16>)
|
|
declare i16 @llvm.experimental.vector.reduce.umax.v8i16(<8 x i16>)
|
|
declare i16 @llvm.experimental.vector.reduce.umax.v16i16(<16 x i16>)
|
|
declare i16 @llvm.experimental.vector.reduce.umax.v32i16(<32 x i16>)
|
|
declare i16 @llvm.experimental.vector.reduce.umax.v64i16(<64 x i16>)
|
|
|
|
declare i8 @llvm.experimental.vector.reduce.umax.v2i8(<2 x i8>)
|
|
declare i8 @llvm.experimental.vector.reduce.umax.v4i8(<4 x i8>)
|
|
declare i8 @llvm.experimental.vector.reduce.umax.v8i8(<8 x i8>)
|
|
declare i8 @llvm.experimental.vector.reduce.umax.v16i8(<16 x i8>)
|
|
declare i8 @llvm.experimental.vector.reduce.umax.v32i8(<32 x i8>)
|
|
declare i8 @llvm.experimental.vector.reduce.umax.v64i8(<64 x i8>)
|
|
declare i8 @llvm.experimental.vector.reduce.umax.v128i8(<128 x i8>)
|