-; RUN: llc < %s -mcpu=x86-64 | FileCheck %s --check-prefix=CHECK-NOSSSE3
-; RUN: llc < %s -mcpu=core2 | FileCheck %s --check-prefix=CHECK-SSSE3
-; RUN: llc < %s -mcpu=core-avx2 | FileCheck %s --check-prefix=CHECK-AVX2
+; RUN: llc < %s -mcpu=x86-64 | FileCheck %s --check-prefix=CHECK-ALL --check-prefix=CHECK-SSE --check-prefix=CHECK-NOSSSE3
+; RUN: llc < %s -mcpu=core2 | FileCheck %s --check-prefix=CHECK-ALL --check-prefix=CHECK-SSE --check-prefix=CHECK-SSSE3
+; RUN: llc < %s -mcpu=core-avx2 | FileCheck %s --check-prefix=CHECK-AVX --check-prefix=CHECK-AVX2
; RUN: llc < %s -mcpu=core-avx2 -x86-experimental-vector-widening-legalization | FileCheck %s --check-prefix=CHECK-WIDE-AVX2
target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
%r = call <4 x i16> @llvm.bswap.v4i16(<4 x i16> %v)
ret <4 x i16> %r
}
+
+;
+; Double BSWAP -> Identity
+;
+
+define <8 x i16> @identity_v8i16(<8 x i16> %v) {
+; CHECK-ALL-LABEL: identity_v8i16:
+; CHECK-ALL: # BB#0: # %entry
+; CHECK-ALL: retq
+entry:
+ %bs1 = call <8 x i16> @llvm.bswap.v8i16(<8 x i16> %v)
+ %bs2 = call <8 x i16> @llvm.bswap.v8i16(<8 x i16> %bs1)
+ ret <8 x i16> %bs2
+}
+
+define <4 x i32> @identity_v4i32(<4 x i32> %v) {
+; CHECK-ALL-LABEL: identity_v4i32:
+; CHECK-ALL: # BB#0: # %entry
+; CHECK-ALL-NEXT: retq
+entry:
+ %bs1 = call <4 x i32> @llvm.bswap.v4i32(<4 x i32> %v)
+ %bs2 = call <4 x i32> @llvm.bswap.v4i32(<4 x i32> %bs1)
+ ret <4 x i32> %bs2
+}
+
+define <2 x i64> @identity_v2i64(<2 x i64> %v) {
+; CHECK-ALL-LABEL: identity_v2i64:
+; CHECK-ALL: # BB#0: # %entry
+; CHECK-ALL-NEXT: retq
+entry:
+ %bs1 = call <2 x i64> @llvm.bswap.v2i64(<2 x i64> %v)
+ %bs2 = call <2 x i64> @llvm.bswap.v2i64(<2 x i64> %bs1)
+ ret <2 x i64> %bs2
+}
+
+define <16 x i16> @identity_v16i16(<16 x i16> %v) {
+; CHECK-ALL-LABEL: identity_v16i16:
+; CHECK-ALL: # BB#0: # %entry
+; CHECK-ALL-NEXT: retq
+entry:
+ %bs1 = call <16 x i16> @llvm.bswap.v16i16(<16 x i16> %v)
+ %bs2 = call <16 x i16> @llvm.bswap.v16i16(<16 x i16> %bs1)
+ ret <16 x i16> %bs2
+}
+
+define <8 x i32> @identity_v8i32(<8 x i32> %v) {
+; CHECK-ALL-LABEL: identity_v8i32:
+; CHECK-ALL: # BB#0: # %entry
+; CHECK-ALL-NEXT: retq
+entry:
+ %bs1 = call <8 x i32> @llvm.bswap.v8i32(<8 x i32> %v)
+ %bs2 = call <8 x i32> @llvm.bswap.v8i32(<8 x i32> %bs1)
+ ret <8 x i32> %bs2
+}
+
+define <4 x i64> @identity_v4i64(<4 x i64> %v) {
+; CHECK-ALL-LABEL: identity_v4i64:
+; CHECK-ALL: # BB#0: # %entry
+; CHECK-ALL-NEXT: retq
+entry:
+ %bs1 = call <4 x i64> @llvm.bswap.v4i64(<4 x i64> %v)
+ %bs2 = call <4 x i64> @llvm.bswap.v4i64(<4 x i64> %bs1)
+ ret <4 x i64> %bs2
+}
+
+define <4 x i16> @identity_v4i16(<4 x i16> %v) {
+; CHECK-ALL-LABEL: identity_v4i16:
+; CHECK-ALL: # BB#0: # %entry
+; CHECK-ALL-NEXT: retq
+entry:
+ %bs1 = call <4 x i16> @llvm.bswap.v4i16(<4 x i16> %v)
+ %bs2 = call <4 x i16> @llvm.bswap.v4i16(<4 x i16> %bs1)
+ ret <4 x i16> %bs2
+}
+
+;
+; Constant Folding
+;
+
+define <8 x i16> @fold_v8i16() {
+; CHECK-SSE-LABEL: fold_v8i16:
+; CHECK-SSE: # BB#0: # %entry
+; CHECK-SSE-NEXT: movaps {{.*#+}} xmm0 = [0,256,65535,512,65023,1024,64511,1536]
+; CHECK-SSE-NEXT: retq
+;
+; CHECK-AVX-LABEL: fold_v8i16:
+; CHECK-AVX: # BB#0: # %entry
+; CHECK-AVX-NEXT: vmovaps {{.*#+}} xmm0 = [0,256,65535,512,65023,1024,64511,1536]
+; CHECK-AVX-NEXT: retq
+entry:
+ %r = call <8 x i16> @llvm.bswap.v8i16(<8 x i16> <i16 0, i16 1, i16 -1, i16 2, i16 -3, i16 4, i16 -5, i16 6>)
+ ret <8 x i16> %r
+}
+
+define <4 x i32> @fold_v4i32() {
+; CHECK-SSE-LABEL: fold_v4i32:
+; CHECK-SSE: # BB#0: # %entry
+; CHECK-SSE-NEXT: movaps {{.*#+}} xmm0 = [0,4294967295,33554432,4261412863]
+; CHECK-SSE-NEXT: retq
+;
+; CHECK-AVX-LABEL: fold_v4i32:
+; CHECK-AVX: # BB#0: # %entry
+; CHECK-AVX-NEXT: vmovaps {{.*#+}} xmm0 = [0,4294967295,33554432,4261412863]
+; CHECK-AVX-NEXT: retq
+entry:
+ %r = call <4 x i32> @llvm.bswap.v4i32(<4 x i32> <i32 0, i32 -1, i32 2, i32 -3>)
+ ret <4 x i32> %r
+}
+
+define <2 x i64> @fold_v2i64() {
+; CHECK-SSE-LABEL: fold_v2i64:
+; CHECK-SSE: # BB#0: # %entry
+; CHECK-SSE-NEXT: movaps {{.*#+}} xmm0 = [18374686479671623680,18446744073709551615]
+; CHECK-SSE-NEXT: retq
+;
+; CHECK-AVX-LABEL: fold_v2i64:
+; CHECK-AVX: # BB#0: # %entry
+; CHECK-AVX-NEXT: vmovaps {{.*#+}} xmm0 = [18374686479671623680,18446744073709551615]
+; CHECK-AVX-NEXT: retq
+entry:
+ %r = call <2 x i64> @llvm.bswap.v2i64(<2 x i64> <i64 255, i64 -1>)
+ ret <2 x i64> %r
+}
+
+define <16 x i16> @fold_v16i16() {
+; CHECK-SSE-LABEL: fold_v16i16:
+; CHECK-SSE: # BB#0: # %entry
+; CHECK-SSE-NEXT: movaps {{.*#+}} xmm0 = [0,256,65535,512,65023,1024,64511,1536]
+; CHECK-SSE-NEXT: movaps {{.*#+}} xmm1 = [63999,2048,63487,2560,62975,3072,62463,3584]
+; CHECK-SSE-NEXT: retq
+;
+; CHECK-AVX-LABEL: fold_v16i16:
+; CHECK-AVX: # BB#0: # %entry
+; CHECK-AVX-NEXT: vmovaps {{.*#+}} ymm0 = [0,256,65535,512,65023,1024,64511,1536,63999,2048,63487,2560,62975,3072,62463,3584]
+; CHECK-AVX-NEXT: retq
+entry:
+ %r = call <16 x i16> @llvm.bswap.v16i16(<16 x i16> <i16 0, i16 1, i16 -1, i16 2, i16 -3, i16 4, i16 -5, i16 6, i16 -7, i16 8, i16 -9, i16 10, i16 -11, i16 12, i16 -13, i16 14>)
+ ret <16 x i16> %r
+}
+
+define <8 x i32> @fold_v8i32() {
+; CHECK-SSE-LABEL: fold_v8i32:
+; CHECK-SSE: # BB#0: # %entry
+; CHECK-SSE-NEXT: movaps {{.*#+}} xmm0 = [0,16777216,4294967295,33554432]
+; CHECK-SSE-NEXT: movaps {{.*#+}} xmm1 = [4261412863,67108864,4227858431,100663296]
+; CHECK-SSE-NEXT: retq
+;
+; CHECK-AVX-LABEL: fold_v8i32:
+; CHECK-AVX: # BB#0: # %entry
+; CHECK-AVX-NEXT: vmovaps {{.*#+}} ymm0 = [0,16777216,4294967295,33554432,4261412863,67108864,4227858431,100663296]
+; CHECK-AVX-NEXT: retq
+entry:
+ %r = call <8 x i32> @llvm.bswap.v8i32(<8 x i32> <i32 0, i32 1, i32 -1, i32 2, i32 -3, i32 4, i32 -5, i32 6>)
+ ret <8 x i32> %r
+}
+
+define <4 x i64> @fold_v4i64() {
+; CHECK-SSE-LABEL: fold_v4i64:
+; CHECK-SSE: # BB#0: # %entry
+; CHECK-SSE-NEXT: movaps {{.*#+}} xmm0 = [18374686479671623680,18446744073709551615]
+; CHECK-SSE-NEXT: movaps {{.*#+}} xmm1 = [18446462598732840960,72056494526300160]
+; CHECK-SSE-NEXT: retq
+;
+; CHECK-AVX-LABEL: fold_v4i64:
+; CHECK-AVX: # BB#0: # %entry
+; CHECK-AVX-NEXT: vmovaps {{.*#+}} ymm0 = [18374686479671623680,18446744073709551615,18446462598732840960,72056494526300160]
+; CHECK-AVX-NEXT: retq
+entry:
+ %r = call <4 x i64> @llvm.bswap.v4i64(<4 x i64> <i64 255, i64 -1, i64 65535, i64 16776960>)
+ ret <4 x i64> %r
+}
projects / oota-llvm.git / blobdiff