05b16aff0f
This patch adds two DAG combines:
1. vector_interleave(splat, splat, ...) -> {splat,splat,...}
2. concat_vectors(splat, splat, ...) -> wide_splat
where all the input splats are identical. Both of these
together enable us to fold
concat_vectors(vector_interleave(splat, splat, ...))
into a wide splat. Post-legalisation we must only do the
concat_vector combine if the wider type and splat operation
is legal.
For fixed-width vectors the DAG combine only occurs for
interleave factors of 3 or more, however it's not currently
safe to test this for AArch64 since there isn't any lowering
support for fixed-width interleaves. I've only added
fixed-width tests for RISCV.
214 lines
7.9 KiB
LLVM
214 lines
7.9 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc -mtriple=aarch64-none-linux-gnu %s -o - | FileCheck %s --check-prefixes=CHECK,CHECK-SD
|
|
; RUN: llc -mtriple=aarch64-none-linux-gnu -global-isel %s -o - 2>&1 | FileCheck %s --check-prefixes=CHECK,CHECK-GI
|
|
|
|
define <4 x half> @interleave2_v4f16(<2 x half> %vec0, <2 x half> %vec1) {
|
|
; CHECK-LABEL: interleave2_v4f16:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: zip1 v0.4h, v0.4h, v1.4h
|
|
; CHECK-NEXT: ret
|
|
%retval = call <4 x half> @llvm.vector.interleave2.v4f16(<2 x half> %vec0, <2 x half> %vec1)
|
|
ret <4 x half> %retval
|
|
}
|
|
|
|
define <8 x half> @interleave2_v8f16(<4 x half> %vec0, <4 x half> %vec1) {
|
|
; CHECK-SD-LABEL: interleave2_v8f16:
|
|
; CHECK-SD: // %bb.0:
|
|
; CHECK-SD-NEXT: // kill: def $d0 killed $d0 def $q0
|
|
; CHECK-SD-NEXT: // kill: def $d1 killed $d1 def $q1
|
|
; CHECK-SD-NEXT: adrp x8, .LCPI1_0
|
|
; CHECK-SD-NEXT: mov v0.d[1], v1.d[0]
|
|
; CHECK-SD-NEXT: ldr q1, [x8, :lo12:.LCPI1_0]
|
|
; CHECK-SD-NEXT: tbl v0.16b, { v0.16b }, v1.16b
|
|
; CHECK-SD-NEXT: ret
|
|
;
|
|
; CHECK-GI-LABEL: interleave2_v8f16:
|
|
; CHECK-GI: // %bb.0:
|
|
; CHECK-GI-NEXT: // kill: def $d0 killed $d0 def $q0
|
|
; CHECK-GI-NEXT: // kill: def $d1 killed $d1 def $q1
|
|
; CHECK-GI-NEXT: zip1 v0.8h, v0.8h, v1.8h
|
|
; CHECK-GI-NEXT: ret
|
|
%retval = call <8 x half> @llvm.vector.interleave2.v8f16(<4 x half> %vec0, <4 x half> %vec1)
|
|
ret <8 x half> %retval
|
|
}
|
|
|
|
define <16 x half> @interleave2_v16f16(<8 x half> %vec0, <8 x half> %vec1) {
|
|
; CHECK-LABEL: interleave2_v16f16:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: zip1 v2.8h, v0.8h, v1.8h
|
|
; CHECK-NEXT: zip2 v1.8h, v0.8h, v1.8h
|
|
; CHECK-NEXT: mov v0.16b, v2.16b
|
|
; CHECK-NEXT: ret
|
|
%retval = call <16 x half> @llvm.vector.interleave2.v16f16(<8 x half> %vec0, <8 x half> %vec1)
|
|
ret <16 x half> %retval
|
|
}
|
|
|
|
define <4 x float> @interleave2_v4f32(<2 x float> %vec0, <2 x float> %vec1) {
|
|
; CHECK-SD-LABEL: interleave2_v4f32:
|
|
; CHECK-SD: // %bb.0:
|
|
; CHECK-SD-NEXT: // kill: def $d0 killed $d0 def $q0
|
|
; CHECK-SD-NEXT: // kill: def $d1 killed $d1 def $q1
|
|
; CHECK-SD-NEXT: mov v0.d[1], v1.d[0]
|
|
; CHECK-SD-NEXT: rev64 v1.4s, v0.4s
|
|
; CHECK-SD-NEXT: uzp1 v0.4s, v0.4s, v1.4s
|
|
; CHECK-SD-NEXT: ret
|
|
;
|
|
; CHECK-GI-LABEL: interleave2_v4f32:
|
|
; CHECK-GI: // %bb.0:
|
|
; CHECK-GI-NEXT: // kill: def $d0 killed $d0 def $q0
|
|
; CHECK-GI-NEXT: // kill: def $d1 killed $d1 def $q1
|
|
; CHECK-GI-NEXT: zip1 v0.4s, v0.4s, v1.4s
|
|
; CHECK-GI-NEXT: ret
|
|
%retval = call <4 x float> @llvm.vector.interleave2.v4f32(<2 x float> %vec0, <2 x float> %vec1)
|
|
ret <4 x float> %retval
|
|
}
|
|
|
|
define <8 x float> @interleave2_v8f32(<4 x float> %vec0, <4 x float> %vec1) {
|
|
; CHECK-LABEL: interleave2_v8f32:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: zip1 v2.4s, v0.4s, v1.4s
|
|
; CHECK-NEXT: zip2 v1.4s, v0.4s, v1.4s
|
|
; CHECK-NEXT: mov v0.16b, v2.16b
|
|
; CHECK-NEXT: ret
|
|
%retval = call <8 x float> @llvm.vector.interleave2.v8f32(<4 x float> %vec0, <4 x float> %vec1)
|
|
ret <8 x float> %retval
|
|
}
|
|
|
|
define <4 x double> @interleave2_v4f64(<2 x double> %vec0, <2 x double> %vec1) {
|
|
; CHECK-LABEL: interleave2_v4f64:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: zip1 v2.2d, v0.2d, v1.2d
|
|
; CHECK-NEXT: zip2 v1.2d, v0.2d, v1.2d
|
|
; CHECK-NEXT: mov v0.16b, v2.16b
|
|
; CHECK-NEXT: ret
|
|
%retval = call <4 x double>@llvm.vector.interleave2.v4f64(<2 x double> %vec0, <2 x double> %vec1)
|
|
ret <4 x double> %retval
|
|
}
|
|
|
|
; Integers
|
|
|
|
define <32 x i8> @interleave2_v32i8(<16 x i8> %vec0, <16 x i8> %vec1) {
|
|
; CHECK-LABEL: interleave2_v32i8:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: zip1 v2.16b, v0.16b, v1.16b
|
|
; CHECK-NEXT: zip2 v1.16b, v0.16b, v1.16b
|
|
; CHECK-NEXT: mov v0.16b, v2.16b
|
|
; CHECK-NEXT: ret
|
|
%retval = call <32 x i8> @llvm.vector.interleave2.v32i8(<16 x i8> %vec0, <16 x i8> %vec1)
|
|
ret <32 x i8> %retval
|
|
}
|
|
|
|
define <16 x i16> @interleave2_v16i16(<8 x i16> %vec0, <8 x i16> %vec1) {
|
|
; CHECK-LABEL: interleave2_v16i16:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: zip1 v2.8h, v0.8h, v1.8h
|
|
; CHECK-NEXT: zip2 v1.8h, v0.8h, v1.8h
|
|
; CHECK-NEXT: mov v0.16b, v2.16b
|
|
; CHECK-NEXT: ret
|
|
%retval = call <16 x i16> @llvm.vector.interleave2.v16i16(<8 x i16> %vec0, <8 x i16> %vec1)
|
|
ret <16 x i16> %retval
|
|
}
|
|
|
|
define <8 x i32> @interleave2_v8i32(<4 x i32> %vec0, <4 x i32> %vec1) {
|
|
; CHECK-LABEL: interleave2_v8i32:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: zip1 v2.4s, v0.4s, v1.4s
|
|
; CHECK-NEXT: zip2 v1.4s, v0.4s, v1.4s
|
|
; CHECK-NEXT: mov v0.16b, v2.16b
|
|
; CHECK-NEXT: ret
|
|
%retval = call <8 x i32> @llvm.vector.interleave2.v8i32(<4 x i32> %vec0, <4 x i32> %vec1)
|
|
ret <8 x i32> %retval
|
|
}
|
|
|
|
define <4 x i64> @interleave2_v4i64(<2 x i64> %vec0, <2 x i64> %vec1) {
|
|
; CHECK-LABEL: interleave2_v4i64:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: zip1 v2.2d, v0.2d, v1.2d
|
|
; CHECK-NEXT: zip2 v1.2d, v0.2d, v1.2d
|
|
; CHECK-NEXT: mov v0.16b, v2.16b
|
|
; CHECK-NEXT: ret
|
|
%retval = call <4 x i64> @llvm.vector.interleave2.v4i64(<2 x i64> %vec0, <2 x i64> %vec1)
|
|
ret <4 x i64> %retval
|
|
}
|
|
|
|
define <4 x i16> @interleave2_same_const_splat_v4i16() {
|
|
; CHECK-SD-LABEL: interleave2_same_const_splat_v4i16:
|
|
; CHECK-SD: // %bb.0:
|
|
; CHECK-SD-NEXT: movi v0.4h, #3
|
|
; CHECK-SD-NEXT: ret
|
|
;
|
|
; CHECK-GI-LABEL: interleave2_same_const_splat_v4i16:
|
|
; CHECK-GI: // %bb.0:
|
|
; CHECK-GI-NEXT: mov w8, #3 // =0x3
|
|
; CHECK-GI-NEXT: fmov s0, w8
|
|
; CHECK-GI-NEXT: mov v0.h[1], w8
|
|
; CHECK-GI-NEXT: zip1 v0.4h, v0.4h, v0.4h
|
|
; CHECK-GI-NEXT: ret
|
|
%retval = call <4 x i16> @llvm.vector.interleave2.v4i16(<2 x i16> splat(i16 3), <2 x i16> splat(i16 3))
|
|
ret <4 x i16> %retval
|
|
}
|
|
|
|
define <4 x i16> @interleave2_diff_const_splat_v4i16() {
|
|
; CHECK-SD-LABEL: interleave2_diff_const_splat_v4i16:
|
|
; CHECK-SD: // %bb.0:
|
|
; CHECK-SD-NEXT: adrp x8, .LCPI11_0
|
|
; CHECK-SD-NEXT: ldr d0, [x8, :lo12:.LCPI11_0]
|
|
; CHECK-SD-NEXT: ret
|
|
;
|
|
; CHECK-GI-LABEL: interleave2_diff_const_splat_v4i16:
|
|
; CHECK-GI: // %bb.0:
|
|
; CHECK-GI-NEXT: mov w8, #3 // =0x3
|
|
; CHECK-GI-NEXT: mov w9, #4 // =0x4
|
|
; CHECK-GI-NEXT: fmov s0, w8
|
|
; CHECK-GI-NEXT: fmov s1, w9
|
|
; CHECK-GI-NEXT: mov v0.h[1], w8
|
|
; CHECK-GI-NEXT: mov v1.h[1], w9
|
|
; CHECK-GI-NEXT: zip1 v0.4h, v0.4h, v1.4h
|
|
; CHECK-GI-NEXT: ret
|
|
%retval = call <4 x i16> @llvm.vector.interleave2.v4i16(<2 x i16> splat(i16 3), <2 x i16> splat(i16 4))
|
|
ret <4 x i16> %retval
|
|
}
|
|
|
|
define <4 x i16> @interleave2_same_nonconst_splat_v4i16(i16 %a) {
|
|
; CHECK-SD-LABEL: interleave2_same_nonconst_splat_v4i16:
|
|
; CHECK-SD: // %bb.0:
|
|
; CHECK-SD-NEXT: dup v0.4h, w0
|
|
; CHECK-SD-NEXT: ret
|
|
;
|
|
; CHECK-GI-LABEL: interleave2_same_nonconst_splat_v4i16:
|
|
; CHECK-GI: // %bb.0:
|
|
; CHECK-GI-NEXT: dup v0.4h, w0
|
|
; CHECK-GI-NEXT: zip1 v0.4h, v0.4h, v0.4h
|
|
; CHECK-GI-NEXT: ret
|
|
%ins = insertelement <2 x i16> poison, i16 %a, i32 0
|
|
%splat = shufflevector <2 x i16> %ins, <2 x i16> poison, <2 x i32> <i32 0, i32 0>
|
|
%retval = call <4 x i16> @llvm.vector.interleave2.v4i16(<2 x i16> %splat, <2 x i16> %splat)
|
|
ret <4 x i16> %retval
|
|
}
|
|
|
|
define <4 x i16> @interleave2_diff_nonconst_splat_v4i16(i16 %a, i16 %b) {
|
|
; CHECK-SD-LABEL: interleave2_diff_nonconst_splat_v4i16:
|
|
; CHECK-SD: // %bb.0:
|
|
; CHECK-SD-NEXT: fmov s0, w0
|
|
; CHECK-SD-NEXT: mov v0.h[1], w0
|
|
; CHECK-SD-NEXT: mov v0.h[2], w1
|
|
; CHECK-SD-NEXT: mov v0.h[3], w1
|
|
; CHECK-SD-NEXT: rev32 v1.4h, v0.4h
|
|
; CHECK-SD-NEXT: uzp1 v0.4h, v0.4h, v1.4h
|
|
; CHECK-SD-NEXT: ret
|
|
;
|
|
; CHECK-GI-LABEL: interleave2_diff_nonconst_splat_v4i16:
|
|
; CHECK-GI: // %bb.0:
|
|
; CHECK-GI-NEXT: dup v0.4h, w0
|
|
; CHECK-GI-NEXT: dup v1.4h, w1
|
|
; CHECK-GI-NEXT: zip1 v0.4h, v0.4h, v1.4h
|
|
; CHECK-GI-NEXT: ret
|
|
%ins1 = insertelement <2 x i16> poison, i16 %a, i32 0
|
|
%splat1 = shufflevector <2 x i16> %ins1, <2 x i16> poison, <2 x i32> <i32 0, i32 0>
|
|
%ins2 = insertelement <2 x i16> poison, i16 %b, i32 0
|
|
%splat2 = shufflevector <2 x i16> %ins2, <2 x i16> poison, <2 x i32> <i32 0, i32 0>
|
|
%retval = call <4 x i16> @llvm.vector.interleave2.v4i16(<2 x i16> %splat1, <2 x i16> %splat2)
|
|
ret <4 x i16> %retval
|
|
}
|
|
|