Custom lower vector type bitreverse to scalar bitrev and vshuf4i instructions. Keep `v2i64` and `v4i64` bitreverse `Expand`, it's good enough.
72 lines
2.4 KiB
LLVM
72 lines
2.4 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
|
|
; RUN: llc --mtriple=loongarch64 -mattr=+lsx --verify-machineinstrs < %s \
|
|
; RUN: | FileCheck %s
|
|
|
|
declare <16 x i8> @llvm.bitreverse.v16i8(<16 x i8>)
|
|
|
|
define <16 x i8> @test_bitreverse_v16i8(<16 x i8> %a) nounwind {
|
|
; CHECK-LABEL: test_bitreverse_v16i8:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 0
|
|
; CHECK-NEXT: bitrev.8b $a0, $a0
|
|
; CHECK-NEXT: vinsgr2vr.d $vr1, $a0, 0
|
|
; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 1
|
|
; CHECK-NEXT: bitrev.8b $a0, $a0
|
|
; CHECK-NEXT: vinsgr2vr.d $vr1, $a0, 1
|
|
; CHECK-NEXT: vori.b $vr0, $vr1, 0
|
|
; CHECK-NEXT: ret
|
|
%b = call <16 x i8> @llvm.bitreverse.v16i8(<16 x i8> %a)
|
|
ret <16 x i8> %b
|
|
}
|
|
|
|
declare <8 x i16> @llvm.bitreverse.v8i16(<8 x i16>)
|
|
|
|
define <8 x i16> @test_bitreverse_v8i16(<8 x i16> %a) nounwind {
|
|
; CHECK-LABEL: test_bitreverse_v8i16:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 0
|
|
; CHECK-NEXT: bitrev.d $a0, $a0
|
|
; CHECK-NEXT: vinsgr2vr.d $vr1, $a0, 0
|
|
; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 1
|
|
; CHECK-NEXT: bitrev.d $a0, $a0
|
|
; CHECK-NEXT: vinsgr2vr.d $vr1, $a0, 1
|
|
; CHECK-NEXT: vshuf4i.h $vr0, $vr1, 27
|
|
; CHECK-NEXT: ret
|
|
%b = call <8 x i16> @llvm.bitreverse.v8i16(<8 x i16> %a)
|
|
ret <8 x i16> %b
|
|
}
|
|
|
|
declare <4 x i32> @llvm.bitreverse.v4i32(<4 x i32>)
|
|
|
|
define <4 x i32> @test_bitreverse_v4i32(<4 x i32> %a) nounwind {
|
|
; CHECK-LABEL: test_bitreverse_v4i32:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 0
|
|
; CHECK-NEXT: bitrev.d $a0, $a0
|
|
; CHECK-NEXT: vinsgr2vr.d $vr1, $a0, 0
|
|
; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 1
|
|
; CHECK-NEXT: bitrev.d $a0, $a0
|
|
; CHECK-NEXT: vinsgr2vr.d $vr1, $a0, 1
|
|
; CHECK-NEXT: vshuf4i.w $vr0, $vr1, 177
|
|
; CHECK-NEXT: ret
|
|
%b = call <4 x i32> @llvm.bitreverse.v4i32(<4 x i32> %a)
|
|
ret <4 x i32> %b
|
|
}
|
|
|
|
declare <2 x i64> @llvm.bitreverse.v2i64(<2 x i64>)
|
|
|
|
define <2 x i64> @test_bitreverse_v2i64(<2 x i64> %a) nounwind {
|
|
; CHECK-LABEL: test_bitreverse_v2i64:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 0
|
|
; CHECK-NEXT: bitrev.d $a0, $a0
|
|
; CHECK-NEXT: vinsgr2vr.d $vr1, $a0, 0
|
|
; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 1
|
|
; CHECK-NEXT: bitrev.d $a0, $a0
|
|
; CHECK-NEXT: vinsgr2vr.d $vr1, $a0, 1
|
|
; CHECK-NEXT: vori.b $vr0, $vr1, 0
|
|
; CHECK-NEXT: ret
|
|
%b = call <2 x i64> @llvm.bitreverse.v2i64(<2 x i64> %a)
|
|
ret <2 x i64> %b
|
|
}
|