derek.holloway
7502bf64a4
CodeQL / Analyze (actions) (push) Canceled after 0s
CodeQL / Analyze (python) (push) Canceled after 0s
pre-commit / pre-commit (push) Canceled after 0s
Multi-Arch CI / setup (push) Canceled after 0s
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Setup (Multi-Arch) / setup (push) Canceled after 0s
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Multi-Arch CI / CI Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
442 lines
24 KiB
LLVM
442 lines
24 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
|
|
; RUN: llc -mtriple=riscv32 -mattr=+v,+zvfh -verify-machineinstrs < %s | FileCheck %s -check-prefixes=CHECK
|
|
|
|
; Can be optimized as a merge followed by a shuffle
|
|
define <16 x i16> @shuffle_shuffle_disjoint(<16 x i16> %op0, <16 x i16> %op1) {
|
|
; CHECK-LABEL: LCPI0_0
|
|
; CHECK-NEXT: .byte 0 # 0x0
|
|
; CHECK-NEXT: .byte 4 # 0x4
|
|
; CHECK-NEXT: .byte 8 # 0x8
|
|
; CHECK-NEXT: .byte 12 # 0xc
|
|
; CHECK-NEXT: .byte 1 # 0x1
|
|
; CHECK-NEXT: .byte 5 # 0x5
|
|
; CHECK-NEXT: .byte 9 # 0x9
|
|
; CHECK-NEXT: .byte 13 # 0xd
|
|
; CHECK-NEXT: .byte 3 # 0x3
|
|
; CHECK-NEXT: .byte 7 # 0x7
|
|
; CHECK-NEXT: .byte 11 # 0xb
|
|
; CHECK-NEXT: .byte 15 # 0xf
|
|
; CHECK-NEXT: .byte 2 # 0x2
|
|
; CHECK-NEXT: .byte 6 # 0x6
|
|
; CHECK-NEXT: .byte 10 # 0xa
|
|
; CHECK-NEXT: .byte 14 # 0xe
|
|
; CHECK-LABEL: shuffle_shuffle_disjoint:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: lui a0, %hi(.LCPI0_0)
|
|
; CHECK-NEXT: addi a0, a0, %lo(.LCPI0_0)
|
|
; CHECK-NEXT: vsetivli zero, 16, e16, m2, ta, ma
|
|
; CHECK-NEXT: vle8.v v14, (a0)
|
|
; CHECK-NEXT: lui a0, 3
|
|
; CHECK-NEXT: addi a0, a0, 819
|
|
; CHECK-NEXT: vmv.s.x v0, a0
|
|
; CHECK-NEXT: vmerge.vvm v10, v10, v8, v0
|
|
; CHECK-NEXT: vsext.vf2 v12, v14
|
|
; CHECK-NEXT: vrgather.vv v8, v10, v12
|
|
; CHECK-NEXT: ret
|
|
%shuff0 = shufflevector <16 x i16> %op0, <16 x i16> poison, <16 x i32> <i32 0, i32 4, i32 8, i32 12, i32 1, i32 5, i32 9, i32 13, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
|
|
%shuff1 = shufflevector <16 x i16> %op1, <16 x i16> poison, <16 x i32> <i32 3, i32 7, i32 11, i32 15, i32 2, i32 6, i32 10, i32 14, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
|
|
%merge = shufflevector <16 x i16> %shuff0, <16 x i16> %shuff1, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23>
|
|
ret <16 x i16> %merge
|
|
}
|
|
|
|
; Can be optimized as a merge followed by a shuffle
|
|
define <16 x i16> @shuffle_shuffle_disjoint_unordered(<16 x i16> %op0, <16 x i16> %op1) {
|
|
; CHECK-LABEL: LCPI1_0
|
|
; CHECK-NEXT: .byte 0 # 0x0
|
|
; CHECK-NEXT: .byte 15 # 0xf
|
|
; CHECK-NEXT: .byte 14 # 0xe
|
|
; CHECK-NEXT: .byte 12 # 0xc
|
|
; CHECK-NEXT: .byte 1 # 0x1
|
|
; CHECK-NEXT: .byte 2 # 0x2
|
|
; CHECK-NEXT: .byte 9 # 0x9
|
|
; CHECK-NEXT: .byte 13 # 0xd
|
|
; CHECK-NEXT: .byte 3 # 0x3
|
|
; CHECK-NEXT: .byte 7 # 0x7
|
|
; CHECK-NEXT: .byte 11 # 0xb
|
|
; CHECK-NEXT: .byte 4 # 0x4
|
|
; CHECK-NEXT: .byte 5 # 0x5
|
|
; CHECK-NEXT: .byte 6 # 0x6
|
|
; CHECK-NEXT: .byte 10 # 0xa
|
|
; CHECK-NEXT: .byte 8 # 0x8
|
|
; CHECK-LABEL: shuffle_shuffle_disjoint_unordered:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: lui a0, %hi(.LCPI1_0)
|
|
; CHECK-NEXT: addi a0, a0, %lo(.LCPI1_0)
|
|
; CHECK-NEXT: vsetivli zero, 16, e16, m2, ta, ma
|
|
; CHECK-NEXT: vle8.v v14, (a0)
|
|
; CHECK-NEXT: lui a0, 3
|
|
; CHECK-NEXT: addi a0, a0, 819
|
|
; CHECK-NEXT: vmv.s.x v0, a0
|
|
; CHECK-NEXT: vmerge.vvm v10, v10, v8, v0
|
|
; CHECK-NEXT: vsext.vf2 v12, v14
|
|
; CHECK-NEXT: vrgather.vv v8, v10, v12
|
|
; CHECK-NEXT: ret
|
|
%shuff0 = shufflevector <16 x i16> %op0, <16 x i16> poison, <16 x i32> <i32 0, i32 4, i32 8, i32 12, i32 1, i32 5, i32 9, i32 13, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
|
|
%shuff1 = shufflevector <16 x i16> %op1, <16 x i16> poison, <16 x i32> <i32 3, i32 7, i32 11, i32 15, i32 2, i32 6, i32 10, i32 14, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
|
|
%merge = shufflevector <16 x i16> %shuff0, <16 x i16> %shuff1, <16 x i32> <i32 0, i32 19, i32 23, i32 3, i32 4, i32 20, i32 6, i32 7, i32 16, i32 17, i32 18, i32 1, i32 5, i32 21, i32 22, i32 2>
|
|
ret <16 x i16> %merge
|
|
}
|
|
|
|
; Can be optimized since the lanes are disjoint, but a single lane is used multiple times by one of the vectors
|
|
define <16 x i16> @shuffle_shuffle_duplicated_within_operand(<16 x i16> %op0, <16 x i16> %op1) {
|
|
; CHECK-LABEL: LCPI2_0
|
|
; CHECK-NEXT: .byte 0 # 0x0
|
|
; CHECK-NEXT: .byte 0 # 0x0
|
|
; CHECK-NEXT: .byte 8 # 0x8
|
|
; CHECK-NEXT: .byte 12 # 0xc
|
|
; CHECK-NEXT: .byte 1 # 0x1
|
|
; CHECK-NEXT: .byte 5 # 0x5
|
|
; CHECK-NEXT: .byte 9 # 0x9
|
|
; CHECK-NEXT: .byte 13 # 0xd
|
|
; CHECK-NEXT: .byte 3 # 0x3
|
|
; CHECK-NEXT: .byte 7 # 0x7
|
|
; CHECK-NEXT: .byte 11 # 0xb
|
|
; CHECK-NEXT: .byte 15 # 0xf
|
|
; CHECK-NEXT: .byte 2 # 0x2
|
|
; CHECK-NEXT: .byte 6 # 0x6
|
|
; CHECK-NEXT: .byte 10 # 0xa
|
|
; CHECK-NEXT: .byte 14 # 0xe
|
|
; CHECK-LABEL: shuffle_shuffle_duplicated_within_operand:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: lui a0, %hi(.LCPI2_0)
|
|
; CHECK-NEXT: addi a0, a0, %lo(.LCPI2_0)
|
|
; CHECK-NEXT: vsetivli zero, 16, e16, m2, ta, ma
|
|
; CHECK-NEXT: vle8.v v14, (a0)
|
|
; CHECK-NEXT: lui a0, 3
|
|
; CHECK-NEXT: addi a0, a0, 803
|
|
; CHECK-NEXT: vmv.s.x v0, a0
|
|
; CHECK-NEXT: vmerge.vvm v10, v10, v8, v0
|
|
; CHECK-NEXT: vsext.vf2 v12, v14
|
|
; CHECK-NEXT: vrgather.vv v8, v10, v12
|
|
; CHECK-NEXT: ret
|
|
%shuff0 = shufflevector <16 x i16> %op0, <16 x i16> poison, <16 x i32> <i32 0, i32 0, i32 8, i32 12, i32 1, i32 5, i32 9, i32 13, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
|
|
%shuff1 = shufflevector <16 x i16> %op1, <16 x i16> poison, <16 x i32> <i32 3, i32 7, i32 11, i32 15, i32 2, i32 6, i32 10, i32 14, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
|
|
%merge = shufflevector <16 x i16> %shuff0, <16 x i16> %shuff1, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23>
|
|
ret <16 x i16> %merge
|
|
}
|
|
|
|
; Can't be optimized as merge-shuffle since the same lane (8) is used from both operands
|
|
define <16 x i16> @shuffle_shuffle_duplicated_lane(<16 x i16> %op0, <16 x i16> %op1) {
|
|
; CHECK-LABEL: LCPI3_0
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .half 3 # 0x3
|
|
; CHECK-NEXT: .half 7 # 0x7
|
|
; CHECK-NEXT: .half 8 # 0x8
|
|
; CHECK-NEXT: .half 15 # 0xf
|
|
; CHECK-NEXT: .half 2 # 0x2
|
|
; CHECK-NEXT: .half 6 # 0x6
|
|
; CHECK-NEXT: .half 10 # 0xa
|
|
; CHECK-NEXT: .half 14 # 0xe
|
|
; CHECK-LABEL: LCPI3_1
|
|
; CHECK-NEXT: .half 0 # 0x0
|
|
; CHECK-NEXT: .half 0 # 0x0
|
|
; CHECK-NEXT: .half 8 # 0x8
|
|
; CHECK-NEXT: .half 12 # 0xc
|
|
; CHECK-NEXT: .half 1 # 0x1
|
|
; CHECK-NEXT: .half 5 # 0x5
|
|
; CHECK-NEXT: .half 9 # 0x9
|
|
; CHECK-NEXT: .half 13 # 0xd
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-LABEL: shuffle_shuffle_duplicated_lane:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: lui a0, %hi(.LCPI3_1)
|
|
; CHECK-NEXT: addi a0, a0, %lo(.LCPI3_1)
|
|
; CHECK-NEXT: vsetivli zero, 16, e16, m2, ta, mu
|
|
; CHECK-NEXT: vle16.v v14, (a0)
|
|
; CHECK-NEXT: lui a0, %hi(.LCPI3_0)
|
|
; CHECK-NEXT: addi a0, a0, %lo(.LCPI3_0)
|
|
; CHECK-NEXT: vle16.v v16, (a0)
|
|
; CHECK-NEXT: li a0, -256
|
|
; CHECK-NEXT: vmv.s.x v0, a0
|
|
; CHECK-NEXT: vrgather.vv v12, v8, v14
|
|
; CHECK-NEXT: vrgather.vv v12, v10, v16, v0.t
|
|
; CHECK-NEXT: vmv.v.v v8, v12
|
|
; CHECK-NEXT: ret
|
|
%shuff0 = shufflevector <16 x i16> %op0, <16 x i16> poison, <16 x i32> <i32 0, i32 0, i32 8, i32 12, i32 1, i32 5, i32 9, i32 13, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
|
|
%shuff1 = shufflevector <16 x i16> %op1, <16 x i16> poison, <16 x i32> <i32 3, i32 7, i32 8, i32 15, i32 2, i32 6, i32 10, i32 14, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
|
|
%merge = shufflevector <16 x i16> %shuff0, <16 x i16> %shuff1, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23>
|
|
ret <16 x i16> %merge
|
|
}
|
|
|
|
; Can't be optimized since shuff0 is used twice
|
|
define <16 x i16> @shuffle_shuffle_multiple_uses(<16 x i16> %op0, <16 x i16> %op1) {
|
|
; CHECK-LABEL: LCPI4_0
|
|
; CHECK-NEXT: .half 0 # 0x0
|
|
; CHECK-NEXT: .half 4 # 0x4
|
|
; CHECK-NEXT: .half 8 # 0x8
|
|
; CHECK-NEXT: .half 12 # 0xc
|
|
; CHECK-NEXT: .half 1 # 0x1
|
|
; CHECK-NEXT: .half 5 # 0x5
|
|
; CHECK-NEXT: .half 9 # 0x9
|
|
; CHECK-NEXT: .half 13 # 0xd
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-LABEL: LCPI4_1
|
|
; CHECK-NEXT: .half 3 # 0x3
|
|
; CHECK-NEXT: .half 7 # 0x7
|
|
; CHECK-NEXT: .half 11 # 0xb
|
|
; CHECK-NEXT: .half 15 # 0xf
|
|
; CHECK-NEXT: .half 2 # 0x2
|
|
; CHECK-NEXT: .half 6 # 0x6
|
|
; CHECK-NEXT: .half 10 # 0xa
|
|
; CHECK-NEXT: .half 14 # 0xe
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-LABEL: shuffle_shuffle_multiple_uses:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: lui a0, %hi(.LCPI4_0)
|
|
; CHECK-NEXT: addi a0, a0, %lo(.LCPI4_0)
|
|
; CHECK-NEXT: vsetivli zero, 16, e16, m2, ta, ma
|
|
; CHECK-NEXT: vle16.v v12, (a0)
|
|
; CHECK-NEXT: lui a0, %hi(.LCPI4_1)
|
|
; CHECK-NEXT: addi a0, a0, %lo(.LCPI4_1)
|
|
; CHECK-NEXT: vle16.v v14, (a0)
|
|
; CHECK-NEXT: vrgather.vv v16, v8, v12
|
|
; CHECK-NEXT: vrgather.vv v8, v10, v14
|
|
; CHECK-NEXT: vmv.v.v v10, v16
|
|
; CHECK-NEXT: vslideup.vi v10, v8, 8
|
|
; CHECK-NEXT: vadd.vv v8, v16, v10
|
|
; CHECK-NEXT: ret
|
|
%shuff0 = shufflevector <16 x i16> %op0, <16 x i16> poison, <16 x i32> <i32 0, i32 4, i32 8, i32 12, i32 1, i32 5, i32 9, i32 13, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
|
|
%shuff1 = shufflevector <16 x i16> %op1, <16 x i16> poison, <16 x i32> <i32 3, i32 7, i32 11, i32 15, i32 2, i32 6, i32 10, i32 14, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
|
|
%merge = shufflevector <16 x i16> %shuff0, <16 x i16> %shuff1, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23>
|
|
%add = add <16 x i16> %shuff0, %merge
|
|
ret <16 x i16> %add
|
|
}
|
|
|
|
; Can be optimized as a merge followed by a shuffle
|
|
define <16 x i16> @shuffle_shuffle_unbalanced(<16 x i16> %op0, <16 x i16> %op1) {
|
|
; CHECK-LABEL: LCPI5_0
|
|
; CHECK-NEXT: .byte 0 # 0x0
|
|
; CHECK-NEXT: .byte 4 # 0x4
|
|
; CHECK-NEXT: .byte 8 # 0x8
|
|
; CHECK-NEXT: .byte 12 # 0xc
|
|
; CHECK-NEXT: .byte 1 # 0x1
|
|
; CHECK-NEXT: .byte 5 # 0x5
|
|
; CHECK-NEXT: .byte 9 # 0x9
|
|
; CHECK-NEXT: .byte 13 # 0xd
|
|
; CHECK-NEXT: .byte 3 # 0x3
|
|
; CHECK-NEXT: .byte 7 # 0x7
|
|
; CHECK-NEXT: .byte 11 # 0xb
|
|
; CHECK-NEXT: .byte 15 # 0xf
|
|
; CHECK-NEXT: .byte 2 # 0x2
|
|
; CHECK-NEXT: .byte 6 # 0x6
|
|
; CHECK-NEXT: .byte 10 # 0xa
|
|
; CHECK-NEXT: .byte 14 # 0xe
|
|
; CHECK-LABEL: shuffle_shuffle_unbalanced:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: lui a0, %hi(.LCPI5_0)
|
|
; CHECK-NEXT: addi a0, a0, %lo(.LCPI5_0)
|
|
; CHECK-NEXT: vsetivli zero, 16, e16, m2, ta, ma
|
|
; CHECK-NEXT: vle8.v v14, (a0)
|
|
; CHECK-NEXT: lui a0, 7
|
|
; CHECK-NEXT: addi a0, a0, 1843
|
|
; CHECK-NEXT: vmv.s.x v0, a0
|
|
; CHECK-NEXT: vmerge.vvm v10, v10, v8, v0
|
|
; CHECK-NEXT: vsext.vf2 v12, v14
|
|
; CHECK-NEXT: vrgather.vv v8, v10, v12
|
|
; CHECK-NEXT: ret
|
|
%shuff0 = shufflevector <16 x i16> %op0, <16 x i16> poison, <16 x i32> <i32 0, i32 4, i32 8, i32 12, i32 1, i32 5, i32 9, i32 13, i32 10, i32 14, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
|
|
%shuff1 = shufflevector <16 x i16> %op1, <16 x i16> poison, <16 x i32> <i32 3, i32 7, i32 11, i32 15, i32 2, i32 6, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
|
|
%merge = shufflevector <16 x i16> %shuff0, <16 x i16> %shuff1, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 8, i32 9>
|
|
ret <16 x i16> %merge
|
|
}
|
|
|
|
; Can't be optimized since final one of the elements used is shuffled to a poison index
|
|
define <16 x i16> @shuffle_shuffle_poison(<16 x i16> %op0, <16 x i16> %op1) {
|
|
; CHECK-LABEL: LCPI6_0
|
|
; CHECK-NEXT: .half 0 # 0x0
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .half 8 # 0x8
|
|
; CHECK-NEXT: .half 12 # 0xc
|
|
; CHECK-NEXT: .half 1 # 0x1
|
|
; CHECK-NEXT: .half 5 # 0x5
|
|
; CHECK-NEXT: .half 9 # 0x9
|
|
; CHECK-NEXT: .half 13 # 0xd
|
|
; CHECK-NEXT: .half 3 # 0x3
|
|
; CHECK-NEXT: .half 7 # 0x7
|
|
; CHECK-NEXT: .half 11 # 0xb
|
|
; CHECK-NEXT: .half 15 # 0xf
|
|
; CHECK-NEXT: .half 2 # 0x2
|
|
; CHECK-NEXT: .half 6 # 0x6
|
|
; CHECK-NEXT: .half 10 # 0xa
|
|
; CHECK-NEXT: .half 14 # 0xe
|
|
; CHECK-LABEL: shuffle_shuffle_poison:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: lui a0, %hi(.LCPI6_0)
|
|
; CHECK-NEXT: addi a0, a0, %lo(.LCPI6_0)
|
|
; CHECK-NEXT: vsetivli zero, 16, e16, m2, ta, ma
|
|
; CHECK-NEXT: vle16.v v12, (a0)
|
|
; CHECK-NEXT: lui a0, 3
|
|
; CHECK-NEXT: addi a0, a0, 803
|
|
; CHECK-NEXT: vmv.s.x v0, a0
|
|
; CHECK-NEXT: vmerge.vvm v10, v10, v8, v0
|
|
; CHECK-NEXT: vrgather.vv v8, v10, v12
|
|
; CHECK-NEXT: ret
|
|
%shuff0 = shufflevector <16 x i16> %op0, <16 x i16> poison, <16 x i32> <i32 0, i32 poison, i32 8, i32 12, i32 1, i32 5, i32 9, i32 13, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
|
|
%shuff1 = shufflevector <16 x i16> %op1, <16 x i16> poison, <16 x i32> <i32 3, i32 7, i32 11, i32 15, i32 2, i32 6, i32 10, i32 14, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
|
|
%merge = shufflevector <16 x i16> %shuff0, <16 x i16> %shuff1, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23>
|
|
ret <16 x i16> %merge
|
|
}
|
|
|
|
; Can't be optimized since final one of the elements used is shuffled to a poison index
|
|
define <16 x i16> @shuffle_shuffle_poison2(<16 x i16> %op0, <16 x i16> %op1) {
|
|
; CHECK-LABEL: LCPI7_0
|
|
; CHECK-NEXT: .half 0 # 0x0
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .half 8 # 0x8
|
|
; CHECK-NEXT: .half 12 # 0xc
|
|
; CHECK-NEXT: .half 1 # 0x1
|
|
; CHECK-NEXT: .half 5 # 0x5
|
|
; CHECK-NEXT: .half 9 # 0x9
|
|
; CHECK-NEXT: .half 13 # 0xd
|
|
; CHECK-NEXT: .half 3 # 0x3
|
|
; CHECK-NEXT: .half 7 # 0x7
|
|
; CHECK-NEXT: .half 11 # 0xb
|
|
; CHECK-NEXT: .half 15 # 0xf
|
|
; CHECK-NEXT: .half 2 # 0x2
|
|
; CHECK-NEXT: .half 6 # 0x6
|
|
; CHECK-NEXT: .half 10 # 0xa
|
|
; CHECK-NEXT: .half 14 # 0xe
|
|
; CHECK-LABEL: shuffle_shuffle_poison2:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: lui a0, %hi(.LCPI7_0)
|
|
; CHECK-NEXT: addi a0, a0, %lo(.LCPI7_0)
|
|
; CHECK-NEXT: vsetivli zero, 16, e16, m2, ta, ma
|
|
; CHECK-NEXT: vle16.v v12, (a0)
|
|
; CHECK-NEXT: lui a0, 3
|
|
; CHECK-NEXT: addi a0, a0, 803
|
|
; CHECK-NEXT: vmv.s.x v0, a0
|
|
; CHECK-NEXT: vmerge.vvm v10, v10, v8, v0
|
|
; CHECK-NEXT: vrgather.vv v8, v10, v12
|
|
; CHECK-NEXT: ret
|
|
%shuff0 = shufflevector <16 x i16> %op0, <16 x i16> poison, <16 x i32> <i32 0, i32 4, i32 8, i32 12, i32 1, i32 5, i32 9, i32 13, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
|
|
%shuff1 = shufflevector <16 x i16> %op1, <16 x i16> poison, <16 x i32> <i32 3, i32 7, i32 11, i32 15, i32 2, i32 6, i32 10, i32 14, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
|
|
%merge = shufflevector <16 x i16> %shuff0, <16 x i16> %shuff1, <16 x i32> <i32 0, i32 poison, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23>
|
|
ret <16 x i16> %merge
|
|
}
|
|
|
|
; Can't be optimized since first shuffle uses from two different operands
|
|
define <16 x i16> @shuffle_shuffle_first_multi_shuffle(<16 x i16> %op0, <16 x i16> %op1) {
|
|
; CHECK-LABEL: LCPI8_0
|
|
; CHECK-NEXT: .half 0 # 0x0
|
|
; CHECK-NEXT: .half 4 # 0x4
|
|
; CHECK-NEXT: .half 8 # 0x8
|
|
; CHECK-NEXT: .half 12 # 0xc
|
|
; CHECK-NEXT: .half 1 # 0x1
|
|
; CHECK-NEXT: .half 5 # 0x5
|
|
; CHECK-NEXT: .half 9 # 0x9
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-LABEL: LCPI8_1
|
|
; CHECK-NEXT: .half 3 # 0x3
|
|
; CHECK-NEXT: .half 7 # 0x7
|
|
; CHECK-NEXT: .half 11 # 0xb
|
|
; CHECK-NEXT: .half 15 # 0xf
|
|
; CHECK-NEXT: .half 2 # 0x2
|
|
; CHECK-NEXT: .half 6 # 0x6
|
|
; CHECK-NEXT: .half 10 # 0xa
|
|
; CHECK-NEXT: .half 14 # 0xe
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-NEXT: .zero 2
|
|
; CHECK-LABEL: shuffle_shuffle_first_multi_shuffle:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: lui a0, %hi(.LCPI8_0)
|
|
; CHECK-NEXT: addi a0, a0, %lo(.LCPI8_0)
|
|
; CHECK-NEXT: vsetivli zero, 16, e16, m2, ta, mu
|
|
; CHECK-NEXT: vle16.v v14, (a0)
|
|
; CHECK-NEXT: lui a0, %hi(.LCPI8_1)
|
|
; CHECK-NEXT: addi a0, a0, %lo(.LCPI8_1)
|
|
; CHECK-NEXT: vle16.v v16, (a0)
|
|
; CHECK-NEXT: li a0, 128
|
|
; CHECK-NEXT: vmv.s.x v0, a0
|
|
; CHECK-NEXT: vrgather.vv v12, v8, v14
|
|
; CHECK-NEXT: vrgather.vi v12, v10, 0, v0.t
|
|
; CHECK-NEXT: vrgather.vv v8, v10, v16
|
|
; CHECK-NEXT: vslideup.vi v12, v8, 8
|
|
; CHECK-NEXT: vmv.v.v v8, v12
|
|
; CHECK-NEXT: ret
|
|
%shuff0 = shufflevector <16 x i16> %op0, <16 x i16> %op1, <16 x i32> <i32 0, i32 4, i32 8, i32 12, i32 1, i32 5, i32 9, i32 16, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
|
|
%shuff1 = shufflevector <16 x i16> %op1, <16 x i16> poison, <16 x i32> <i32 3, i32 7, i32 11, i32 15, i32 2, i32 6, i32 10, i32 14, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
|
|
%merge = shufflevector <16 x i16> %shuff0, <16 x i16> %shuff1, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23>
|
|
ret <16 x i16> %merge
|
|
}
|
|
|
|
; Can optimize, the first two shuffles use the same lanes, but only a disjoint
|
|
; set of lanes is used for the final shuffle
|
|
define <16 x i16> @shuffle_shuffle_duplicates_not_used(<16 x i16> %op0, <16 x i16> %op1) {
|
|
; CHECK-LABEL: LCPI9_0
|
|
; CHECK-NEXT: .byte 0 # 0x0
|
|
; CHECK-NEXT: .byte 4 # 0x4
|
|
; CHECK-NEXT: .byte 8 # 0x8
|
|
; CHECK-NEXT: .byte 12 # 0xc
|
|
; CHECK-NEXT: .byte 1 # 0x1
|
|
; CHECK-NEXT: .byte 5 # 0x5
|
|
; CHECK-NEXT: .byte 9 # 0x9
|
|
; CHECK-NEXT: .byte 13 # 0xd
|
|
; CHECK-NEXT: .byte 3 # 0x3
|
|
; CHECK-NEXT: .byte 7 # 0x7
|
|
; CHECK-NEXT: .byte 11 # 0xb
|
|
; CHECK-NEXT: .byte 15 # 0xf
|
|
; CHECK-NEXT: .byte 2 # 0x2
|
|
; CHECK-NEXT: .byte 6 # 0x6
|
|
; CHECK-NEXT: .byte 10 # 0xa
|
|
; CHECK-NEXT: .byte 14 # 0xe
|
|
; CHECK-LABEL: shuffle_shuffle_duplicates_not_used:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: lui a0, %hi(.LCPI9_0)
|
|
; CHECK-NEXT: addi a0, a0, %lo(.LCPI9_0)
|
|
; CHECK-NEXT: vsetivli zero, 16, e16, m2, ta, ma
|
|
; CHECK-NEXT: vle8.v v14, (a0)
|
|
; CHECK-NEXT: lui a0, 3
|
|
; CHECK-NEXT: addi a0, a0, 819
|
|
; CHECK-NEXT: vmv.s.x v0, a0
|
|
; CHECK-NEXT: vmerge.vvm v10, v10, v8, v0
|
|
; CHECK-NEXT: vsext.vf2 v12, v14
|
|
; CHECK-NEXT: vrgather.vv v8, v10, v12
|
|
; CHECK-NEXT: ret
|
|
%shuff0 = shufflevector <16 x i16> %op0, <16 x i16> poison, <16 x i32> <i32 0, i32 4, i32 8, i32 12, i32 1, i32 5, i32 9, i32 13, i32 3, i32 7, i32 11, i32 15, i32 2, i32 6, i32 10, i32 14>
|
|
%shuff1 = shufflevector <16 x i16> %op1, <16 x i16> poison, <16 x i32> <i32 3, i32 7, i32 11, i32 15, i32 2, i32 6, i32 10, i32 14, i32 0, i32 4, i32 8, i32 12, i32 1, i32 5, i32 9, i32 13>
|
|
%merge = shufflevector <16 x i16> %shuff0, <16 x i16> %shuff1, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23>
|
|
ret <16 x i16> %merge
|
|
}
|