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
1275 lines
51 KiB
LLVM
1275 lines
51 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc -mtriple=riscv64 -mattr=+v -verify-machineinstrs < %s | FileCheck %s --check-prefix=NOVLDEP
|
|
; RUN: llc -mtriple=riscv64 -mattr=+v,+vl-dependent-latency -verify-machineinstrs < %s | FileCheck %s --check-prefix=VLDEP
|
|
|
|
define <vscale x 1 x i1> @test_vp_splice_nxv1i1(<vscale x 1 x i1> %va, <vscale x 1 x i1> %vb, i32 zeroext %evla, i32 zeroext %evlb) #0 {
|
|
; NOVLDEP-LABEL: test_vp_splice_nxv1i1:
|
|
; NOVLDEP: # %bb.0:
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, mf8, ta, ma
|
|
; NOVLDEP-NEXT: vmv1r.v v9, v0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v8
|
|
; NOVLDEP-NEXT: vmv.v.i v8, 0
|
|
; NOVLDEP-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a0, e8, mf8, ta, ma
|
|
; NOVLDEP-NEXT: vmv.v.i v10, 0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v9
|
|
; NOVLDEP-NEXT: vmerge.vim v9, v10, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, mf8, ta, ma
|
|
; NOVLDEP-NEXT: vslidedown.vi v9, v9, 1
|
|
; NOVLDEP-NEXT: addi a0, a0, -1
|
|
; NOVLDEP-NEXT: vslideup.vx v9, v8, a0
|
|
; NOVLDEP-NEXT: vmsne.vi v0, v9, 0
|
|
; NOVLDEP-NEXT: ret
|
|
;
|
|
; VLDEP-LABEL: test_vp_splice_nxv1i1:
|
|
; VLDEP: # %bb.0:
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, mf8, ta, ma
|
|
; VLDEP-NEXT: vmv1r.v v9, v0
|
|
; VLDEP-NEXT: vmv1r.v v0, v8
|
|
; VLDEP-NEXT: vmv.v.i v8, 0
|
|
; VLDEP-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, mf8, ta, ma
|
|
; VLDEP-NEXT: vmv.v.i v10, 0
|
|
; VLDEP-NEXT: vmv1r.v v0, v9
|
|
; VLDEP-NEXT: vmerge.vim v9, v10, 1, v0
|
|
; VLDEP-NEXT: addi a0, a0, -1
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, mf8, ta, ma
|
|
; VLDEP-NEXT: vslidedown.vi v9, v9, 1
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, mf8, ta, ma
|
|
; VLDEP-NEXT: vslideup.vx v9, v8, a0
|
|
; VLDEP-NEXT: vmsne.vi v0, v9, 0
|
|
; VLDEP-NEXT: ret
|
|
|
|
%v = call <vscale x 1 x i1> @llvm.experimental.vp.splice.nxv1i1(<vscale x 1 x i1> %va, <vscale x 1 x i1> %vb, i32 1, <vscale x 1 x i1> splat (i1 1), i32 %evla, i32 %evlb)
|
|
ret <vscale x 1 x i1> %v
|
|
}
|
|
|
|
define <vscale x 1 x i1> @test_vp_splice_nxv1i1_negative_offset(<vscale x 1 x i1> %va, <vscale x 1 x i1> %vb, i32 zeroext %evla, i32 zeroext %evlb) #0 {
|
|
; NOVLDEP-LABEL: test_vp_splice_nxv1i1_negative_offset:
|
|
; NOVLDEP: # %bb.0:
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, mf8, ta, ma
|
|
; NOVLDEP-NEXT: vmv1r.v v9, v0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v8
|
|
; NOVLDEP-NEXT: vmv.v.i v8, 0
|
|
; NOVLDEP-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a0, e8, mf8, ta, ma
|
|
; NOVLDEP-NEXT: vmv.v.i v10, 0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v9
|
|
; NOVLDEP-NEXT: vmerge.vim v9, v10, 1, v0
|
|
; NOVLDEP-NEXT: addi a0, a0, -2
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, mf8, ta, ma
|
|
; NOVLDEP-NEXT: vslidedown.vx v9, v9, a0
|
|
; NOVLDEP-NEXT: vslideup.vi v9, v8, 2
|
|
; NOVLDEP-NEXT: vmsne.vi v0, v9, 0
|
|
; NOVLDEP-NEXT: ret
|
|
;
|
|
; VLDEP-LABEL: test_vp_splice_nxv1i1_negative_offset:
|
|
; VLDEP: # %bb.0:
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, mf8, ta, ma
|
|
; VLDEP-NEXT: vmv1r.v v9, v0
|
|
; VLDEP-NEXT: vmv1r.v v0, v8
|
|
; VLDEP-NEXT: vmv.v.i v8, 0
|
|
; VLDEP-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, mf8, ta, ma
|
|
; VLDEP-NEXT: vmv.v.i v10, 0
|
|
; VLDEP-NEXT: vmv1r.v v0, v9
|
|
; VLDEP-NEXT: vmerge.vim v9, v10, 1, v0
|
|
; VLDEP-NEXT: addi a0, a0, -2
|
|
; VLDEP-NEXT: vsetivli zero, 2, e8, mf8, ta, ma
|
|
; VLDEP-NEXT: vslidedown.vx v9, v9, a0
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, mf8, ta, ma
|
|
; VLDEP-NEXT: vslideup.vi v9, v8, 2
|
|
; VLDEP-NEXT: vmsne.vi v0, v9, 0
|
|
; VLDEP-NEXT: ret
|
|
|
|
%v = call <vscale x 1 x i1> @llvm.experimental.vp.splice.nxv1i1(<vscale x 1 x i1> %va, <vscale x 1 x i1> %vb, i32 -2, <vscale x 1 x i1> splat (i1 1), i32 %evla, i32 %evlb)
|
|
ret <vscale x 1 x i1> %v
|
|
}
|
|
|
|
define <vscale x 1 x i1> @test_vp_splice_nxv1i1_masked(<vscale x 1 x i1> %va, <vscale x 1 x i1> %vb, <vscale x 1 x i1> %mask, i32 zeroext %evla, i32 zeroext %evlb) #0 {
|
|
; NOVLDEP-LABEL: test_vp_splice_nxv1i1_masked:
|
|
; NOVLDEP: # %bb.0:
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, mf8, ta, ma
|
|
; NOVLDEP-NEXT: vmv1r.v v10, v0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v8
|
|
; NOVLDEP-NEXT: vmv.v.i v8, 0
|
|
; NOVLDEP-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a0, e8, mf8, ta, ma
|
|
; NOVLDEP-NEXT: vmv.v.i v11, 0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v10
|
|
; NOVLDEP-NEXT: vmerge.vim v10, v11, 1, v0
|
|
; NOVLDEP-NEXT: addi a0, a0, -1
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v9
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, mf8, ta, ma
|
|
; NOVLDEP-NEXT: vslidedown.vi v10, v10, 1, v0.t
|
|
; NOVLDEP-NEXT: vsetvli zero, zero, e8, mf8, ta, mu
|
|
; NOVLDEP-NEXT: vslideup.vx v10, v8, a0, v0.t
|
|
; NOVLDEP-NEXT: vsetvli zero, zero, e8, mf8, ta, ma
|
|
; NOVLDEP-NEXT: vmsne.vi v0, v10, 0, v0.t
|
|
; NOVLDEP-NEXT: ret
|
|
;
|
|
; VLDEP-LABEL: test_vp_splice_nxv1i1_masked:
|
|
; VLDEP: # %bb.0:
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, mf8, ta, ma
|
|
; VLDEP-NEXT: vmv1r.v v10, v0
|
|
; VLDEP-NEXT: vmv1r.v v0, v8
|
|
; VLDEP-NEXT: vmv.v.i v8, 0
|
|
; VLDEP-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, mf8, ta, ma
|
|
; VLDEP-NEXT: vmv.v.i v11, 0
|
|
; VLDEP-NEXT: vmv1r.v v0, v10
|
|
; VLDEP-NEXT: vmerge.vim v10, v11, 1, v0
|
|
; VLDEP-NEXT: addi a0, a0, -1
|
|
; VLDEP-NEXT: vmv1r.v v0, v9
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, mf8, ta, ma
|
|
; VLDEP-NEXT: vslidedown.vi v10, v10, 1, v0.t
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, mf8, ta, mu
|
|
; VLDEP-NEXT: vslideup.vx v10, v8, a0, v0.t
|
|
; VLDEP-NEXT: vsetvli zero, zero, e8, mf8, ta, ma
|
|
; VLDEP-NEXT: vmsne.vi v0, v10, 0, v0.t
|
|
; VLDEP-NEXT: ret
|
|
%v = call <vscale x 1 x i1> @llvm.experimental.vp.splice.nxv1i1(<vscale x 1 x i1> %va, <vscale x 1 x i1> %vb, i32 1, <vscale x 1 x i1> %mask, i32 %evla, i32 %evlb)
|
|
ret <vscale x 1 x i1> %v
|
|
}
|
|
|
|
define <vscale x 2 x i1> @test_vp_splice_nxv2i1(<vscale x 2 x i1> %va, <vscale x 2 x i1> %vb, i32 zeroext %evla, i32 zeroext %evlb) #0 {
|
|
; NOVLDEP-LABEL: test_vp_splice_nxv2i1:
|
|
; NOVLDEP: # %bb.0:
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, mf4, ta, ma
|
|
; NOVLDEP-NEXT: vmv1r.v v9, v0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v8
|
|
; NOVLDEP-NEXT: vmv.v.i v8, 0
|
|
; NOVLDEP-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a0, e8, mf4, ta, ma
|
|
; NOVLDEP-NEXT: vmv.v.i v10, 0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v9
|
|
; NOVLDEP-NEXT: vmerge.vim v9, v10, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, mf4, ta, ma
|
|
; NOVLDEP-NEXT: vslidedown.vi v9, v9, 3
|
|
; NOVLDEP-NEXT: addi a0, a0, -3
|
|
; NOVLDEP-NEXT: vslideup.vx v9, v8, a0
|
|
; NOVLDEP-NEXT: vmsne.vi v0, v9, 0
|
|
; NOVLDEP-NEXT: ret
|
|
;
|
|
; VLDEP-LABEL: test_vp_splice_nxv2i1:
|
|
; VLDEP: # %bb.0:
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, mf4, ta, ma
|
|
; VLDEP-NEXT: vmv1r.v v9, v0
|
|
; VLDEP-NEXT: vmv1r.v v0, v8
|
|
; VLDEP-NEXT: vmv.v.i v8, 0
|
|
; VLDEP-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, mf4, ta, ma
|
|
; VLDEP-NEXT: vmv.v.i v10, 0
|
|
; VLDEP-NEXT: vmv1r.v v0, v9
|
|
; VLDEP-NEXT: vmerge.vim v9, v10, 1, v0
|
|
; VLDEP-NEXT: addi a0, a0, -3
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, mf4, ta, ma
|
|
; VLDEP-NEXT: vslidedown.vi v9, v9, 3
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, mf4, ta, ma
|
|
; VLDEP-NEXT: vslideup.vx v9, v8, a0
|
|
; VLDEP-NEXT: vmsne.vi v0, v9, 0
|
|
; VLDEP-NEXT: ret
|
|
|
|
%v = call <vscale x 2 x i1> @llvm.experimental.vp.splice.nxv2i1(<vscale x 2 x i1> %va, <vscale x 2 x i1> %vb, i32 3, <vscale x 2 x i1> splat (i1 1), i32 %evla, i32 %evlb)
|
|
ret <vscale x 2 x i1> %v
|
|
}
|
|
|
|
define <vscale x 2 x i1> @test_vp_splice_nxv2i1_negative_offset(<vscale x 2 x i1> %va, <vscale x 2 x i1> %vb, i32 zeroext %evla, i32 zeroext %evlb) #0 {
|
|
; NOVLDEP-LABEL: test_vp_splice_nxv2i1_negative_offset:
|
|
; NOVLDEP: # %bb.0:
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, mf4, ta, ma
|
|
; NOVLDEP-NEXT: vmv1r.v v9, v0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v8
|
|
; NOVLDEP-NEXT: vmv.v.i v8, 0
|
|
; NOVLDEP-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a0, e8, mf4, ta, ma
|
|
; NOVLDEP-NEXT: vmv.v.i v10, 0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v9
|
|
; NOVLDEP-NEXT: vmerge.vim v9, v10, 1, v0
|
|
; NOVLDEP-NEXT: addi a0, a0, -4
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, mf4, ta, ma
|
|
; NOVLDEP-NEXT: vslidedown.vx v9, v9, a0
|
|
; NOVLDEP-NEXT: vslideup.vi v9, v8, 4
|
|
; NOVLDEP-NEXT: vmsne.vi v0, v9, 0
|
|
; NOVLDEP-NEXT: ret
|
|
;
|
|
; VLDEP-LABEL: test_vp_splice_nxv2i1_negative_offset:
|
|
; VLDEP: # %bb.0:
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, mf4, ta, ma
|
|
; VLDEP-NEXT: vmv1r.v v9, v0
|
|
; VLDEP-NEXT: vmv1r.v v0, v8
|
|
; VLDEP-NEXT: vmv.v.i v8, 0
|
|
; VLDEP-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, mf4, ta, ma
|
|
; VLDEP-NEXT: vmv.v.i v10, 0
|
|
; VLDEP-NEXT: vmv1r.v v0, v9
|
|
; VLDEP-NEXT: vmerge.vim v9, v10, 1, v0
|
|
; VLDEP-NEXT: addi a0, a0, -4
|
|
; VLDEP-NEXT: vsetivli zero, 4, e8, mf4, ta, ma
|
|
; VLDEP-NEXT: vslidedown.vx v9, v9, a0
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, mf4, ta, ma
|
|
; VLDEP-NEXT: vslideup.vi v9, v8, 4
|
|
; VLDEP-NEXT: vmsne.vi v0, v9, 0
|
|
; VLDEP-NEXT: ret
|
|
|
|
%v = call <vscale x 2 x i1> @llvm.experimental.vp.splice.nxv2i1(<vscale x 2 x i1> %va, <vscale x 2 x i1> %vb, i32 -4, <vscale x 2 x i1> splat (i1 1), i32 %evla, i32 %evlb)
|
|
ret <vscale x 2 x i1> %v
|
|
}
|
|
|
|
define <vscale x 2 x i1> @test_vp_splice_nxv2i1_masked(<vscale x 2 x i1> %va, <vscale x 2 x i1> %vb, <vscale x 2 x i1> %mask, i32 zeroext %evla, i32 zeroext %evlb) #0 {
|
|
; NOVLDEP-LABEL: test_vp_splice_nxv2i1_masked:
|
|
; NOVLDEP: # %bb.0:
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, mf4, ta, ma
|
|
; NOVLDEP-NEXT: vmv1r.v v10, v0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v8
|
|
; NOVLDEP-NEXT: vmv.v.i v8, 0
|
|
; NOVLDEP-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a0, e8, mf4, ta, ma
|
|
; NOVLDEP-NEXT: vmv.v.i v11, 0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v10
|
|
; NOVLDEP-NEXT: vmerge.vim v10, v11, 1, v0
|
|
; NOVLDEP-NEXT: addi a0, a0, -3
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v9
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, mf4, ta, ma
|
|
; NOVLDEP-NEXT: vslidedown.vi v10, v10, 3, v0.t
|
|
; NOVLDEP-NEXT: vsetvli zero, zero, e8, mf4, ta, mu
|
|
; NOVLDEP-NEXT: vslideup.vx v10, v8, a0, v0.t
|
|
; NOVLDEP-NEXT: vsetvli zero, zero, e8, mf4, ta, ma
|
|
; NOVLDEP-NEXT: vmsne.vi v0, v10, 0, v0.t
|
|
; NOVLDEP-NEXT: ret
|
|
;
|
|
; VLDEP-LABEL: test_vp_splice_nxv2i1_masked:
|
|
; VLDEP: # %bb.0:
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, mf4, ta, ma
|
|
; VLDEP-NEXT: vmv1r.v v10, v0
|
|
; VLDEP-NEXT: vmv1r.v v0, v8
|
|
; VLDEP-NEXT: vmv.v.i v8, 0
|
|
; VLDEP-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, mf4, ta, ma
|
|
; VLDEP-NEXT: vmv.v.i v11, 0
|
|
; VLDEP-NEXT: vmv1r.v v0, v10
|
|
; VLDEP-NEXT: vmerge.vim v10, v11, 1, v0
|
|
; VLDEP-NEXT: addi a0, a0, -3
|
|
; VLDEP-NEXT: vmv1r.v v0, v9
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, mf4, ta, ma
|
|
; VLDEP-NEXT: vslidedown.vi v10, v10, 3, v0.t
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, mf4, ta, mu
|
|
; VLDEP-NEXT: vslideup.vx v10, v8, a0, v0.t
|
|
; VLDEP-NEXT: vsetvli zero, zero, e8, mf4, ta, ma
|
|
; VLDEP-NEXT: vmsne.vi v0, v10, 0, v0.t
|
|
; VLDEP-NEXT: ret
|
|
%v = call <vscale x 2 x i1> @llvm.experimental.vp.splice.nxv2i1(<vscale x 2 x i1> %va, <vscale x 2 x i1> %vb, i32 3, <vscale x 2 x i1> %mask, i32 %evla, i32 %evlb)
|
|
ret <vscale x 2 x i1> %v
|
|
}
|
|
|
|
define <vscale x 4 x i1> @test_vp_splice_nxv4i1(<vscale x 4 x i1> %va, <vscale x 4 x i1> %vb, i32 zeroext %evla, i32 zeroext %evlb) #0 {
|
|
; NOVLDEP-LABEL: test_vp_splice_nxv4i1:
|
|
; NOVLDEP: # %bb.0:
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, mf2, ta, ma
|
|
; NOVLDEP-NEXT: vmv1r.v v9, v0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v8
|
|
; NOVLDEP-NEXT: vmv.v.i v8, 0
|
|
; NOVLDEP-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a0, e8, mf2, ta, ma
|
|
; NOVLDEP-NEXT: vmv.v.i v10, 0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v9
|
|
; NOVLDEP-NEXT: vmerge.vim v9, v10, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, mf2, ta, ma
|
|
; NOVLDEP-NEXT: vslidedown.vi v9, v9, 5
|
|
; NOVLDEP-NEXT: addi a0, a0, -5
|
|
; NOVLDEP-NEXT: vslideup.vx v9, v8, a0
|
|
; NOVLDEP-NEXT: vmsne.vi v0, v9, 0
|
|
; NOVLDEP-NEXT: ret
|
|
;
|
|
; VLDEP-LABEL: test_vp_splice_nxv4i1:
|
|
; VLDEP: # %bb.0:
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, mf2, ta, ma
|
|
; VLDEP-NEXT: vmv1r.v v9, v0
|
|
; VLDEP-NEXT: vmv1r.v v0, v8
|
|
; VLDEP-NEXT: vmv.v.i v8, 0
|
|
; VLDEP-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, mf2, ta, ma
|
|
; VLDEP-NEXT: vmv.v.i v10, 0
|
|
; VLDEP-NEXT: vmv1r.v v0, v9
|
|
; VLDEP-NEXT: vmerge.vim v9, v10, 1, v0
|
|
; VLDEP-NEXT: addi a0, a0, -5
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, mf2, ta, ma
|
|
; VLDEP-NEXT: vslidedown.vi v9, v9, 5
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, mf2, ta, ma
|
|
; VLDEP-NEXT: vslideup.vx v9, v8, a0
|
|
; VLDEP-NEXT: vmsne.vi v0, v9, 0
|
|
; VLDEP-NEXT: ret
|
|
|
|
%v = call <vscale x 4 x i1> @llvm.experimental.vp.splice.nxv4i1(<vscale x 4 x i1> %va, <vscale x 4 x i1> %vb, i32 5, <vscale x 4 x i1> splat (i1 1), i32 %evla, i32 %evlb)
|
|
ret <vscale x 4 x i1> %v
|
|
}
|
|
|
|
define <vscale x 4 x i1> @test_vp_splice_nxv4i1_negative_offset(<vscale x 4 x i1> %va, <vscale x 4 x i1> %vb, i32 zeroext %evla, i32 zeroext %evlb) #0 {
|
|
; NOVLDEP-LABEL: test_vp_splice_nxv4i1_negative_offset:
|
|
; NOVLDEP: # %bb.0:
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, mf2, ta, ma
|
|
; NOVLDEP-NEXT: vmv1r.v v9, v0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v8
|
|
; NOVLDEP-NEXT: vmv.v.i v8, 0
|
|
; NOVLDEP-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a0, e8, mf2, ta, ma
|
|
; NOVLDEP-NEXT: vmv.v.i v10, 0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v9
|
|
; NOVLDEP-NEXT: vmerge.vim v9, v10, 1, v0
|
|
; NOVLDEP-NEXT: addi a0, a0, -5
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, mf2, ta, ma
|
|
; NOVLDEP-NEXT: vslidedown.vx v9, v9, a0
|
|
; NOVLDEP-NEXT: vslideup.vi v9, v8, 5
|
|
; NOVLDEP-NEXT: vmsne.vi v0, v9, 0
|
|
; NOVLDEP-NEXT: ret
|
|
;
|
|
; VLDEP-LABEL: test_vp_splice_nxv4i1_negative_offset:
|
|
; VLDEP: # %bb.0:
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, mf2, ta, ma
|
|
; VLDEP-NEXT: vmv1r.v v9, v0
|
|
; VLDEP-NEXT: vmv1r.v v0, v8
|
|
; VLDEP-NEXT: vmv.v.i v8, 0
|
|
; VLDEP-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, mf2, ta, ma
|
|
; VLDEP-NEXT: vmv.v.i v10, 0
|
|
; VLDEP-NEXT: vmv1r.v v0, v9
|
|
; VLDEP-NEXT: vmerge.vim v9, v10, 1, v0
|
|
; VLDEP-NEXT: addi a0, a0, -5
|
|
; VLDEP-NEXT: vsetivli zero, 5, e8, mf2, ta, ma
|
|
; VLDEP-NEXT: vslidedown.vx v9, v9, a0
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, mf2, ta, ma
|
|
; VLDEP-NEXT: vslideup.vi v9, v8, 5
|
|
; VLDEP-NEXT: vmsne.vi v0, v9, 0
|
|
; VLDEP-NEXT: ret
|
|
|
|
%v = call <vscale x 4 x i1> @llvm.experimental.vp.splice.nxv4i1(<vscale x 4 x i1> %va, <vscale x 4 x i1> %vb, i32 -5, <vscale x 4 x i1> splat (i1 1), i32 %evla, i32 %evlb)
|
|
ret <vscale x 4 x i1> %v
|
|
}
|
|
|
|
define <vscale x 4 x i1> @test_vp_splice_nxv4i1_masked(<vscale x 4 x i1> %va, <vscale x 4 x i1> %vb, <vscale x 4 x i1> %mask, i32 zeroext %evla, i32 zeroext %evlb) #0 {
|
|
; NOVLDEP-LABEL: test_vp_splice_nxv4i1_masked:
|
|
; NOVLDEP: # %bb.0:
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, mf2, ta, ma
|
|
; NOVLDEP-NEXT: vmv1r.v v10, v0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v8
|
|
; NOVLDEP-NEXT: vmv.v.i v8, 0
|
|
; NOVLDEP-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a0, e8, mf2, ta, ma
|
|
; NOVLDEP-NEXT: vmv.v.i v11, 0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v10
|
|
; NOVLDEP-NEXT: vmerge.vim v10, v11, 1, v0
|
|
; NOVLDEP-NEXT: addi a0, a0, -5
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v9
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, mf2, ta, ma
|
|
; NOVLDEP-NEXT: vslidedown.vi v10, v10, 5, v0.t
|
|
; NOVLDEP-NEXT: vsetvli zero, zero, e8, mf2, ta, mu
|
|
; NOVLDEP-NEXT: vslideup.vx v10, v8, a0, v0.t
|
|
; NOVLDEP-NEXT: vsetvli zero, zero, e8, mf2, ta, ma
|
|
; NOVLDEP-NEXT: vmsne.vi v0, v10, 0, v0.t
|
|
; NOVLDEP-NEXT: ret
|
|
;
|
|
; VLDEP-LABEL: test_vp_splice_nxv4i1_masked:
|
|
; VLDEP: # %bb.0:
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, mf2, ta, ma
|
|
; VLDEP-NEXT: vmv1r.v v10, v0
|
|
; VLDEP-NEXT: vmv1r.v v0, v8
|
|
; VLDEP-NEXT: vmv.v.i v8, 0
|
|
; VLDEP-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, mf2, ta, ma
|
|
; VLDEP-NEXT: vmv.v.i v11, 0
|
|
; VLDEP-NEXT: vmv1r.v v0, v10
|
|
; VLDEP-NEXT: vmerge.vim v10, v11, 1, v0
|
|
; VLDEP-NEXT: addi a0, a0, -5
|
|
; VLDEP-NEXT: vmv1r.v v0, v9
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, mf2, ta, ma
|
|
; VLDEP-NEXT: vslidedown.vi v10, v10, 5, v0.t
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, mf2, ta, mu
|
|
; VLDEP-NEXT: vslideup.vx v10, v8, a0, v0.t
|
|
; VLDEP-NEXT: vsetvli zero, zero, e8, mf2, ta, ma
|
|
; VLDEP-NEXT: vmsne.vi v0, v10, 0, v0.t
|
|
; VLDEP-NEXT: ret
|
|
%v = call <vscale x 4 x i1> @llvm.experimental.vp.splice.nxv4i1(<vscale x 4 x i1> %va, <vscale x 4 x i1> %vb, i32 5, <vscale x 4 x i1> %mask, i32 %evla, i32 %evlb)
|
|
ret <vscale x 4 x i1> %v
|
|
}
|
|
|
|
define <vscale x 8 x i1> @test_vp_splice_nxv8i1(<vscale x 8 x i1> %va, <vscale x 8 x i1> %vb, i32 zeroext %evla, i32 zeroext %evlb) #0 {
|
|
; NOVLDEP-LABEL: test_vp_splice_nxv8i1:
|
|
; NOVLDEP: # %bb.0:
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m1, ta, ma
|
|
; NOVLDEP-NEXT: vmv1r.v v9, v0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v8
|
|
; NOVLDEP-NEXT: vmv.v.i v8, 0
|
|
; NOVLDEP-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a0, e8, m1, ta, ma
|
|
; NOVLDEP-NEXT: vmv.v.i v10, 0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v9
|
|
; NOVLDEP-NEXT: vmerge.vim v9, v10, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m1, ta, ma
|
|
; NOVLDEP-NEXT: vslidedown.vi v9, v9, 5
|
|
; NOVLDEP-NEXT: addi a0, a0, -5
|
|
; NOVLDEP-NEXT: vslideup.vx v9, v8, a0
|
|
; NOVLDEP-NEXT: vmsne.vi v0, v9, 0
|
|
; NOVLDEP-NEXT: ret
|
|
;
|
|
; VLDEP-LABEL: test_vp_splice_nxv8i1:
|
|
; VLDEP: # %bb.0:
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m1, ta, ma
|
|
; VLDEP-NEXT: vmv1r.v v9, v0
|
|
; VLDEP-NEXT: vmv1r.v v0, v8
|
|
; VLDEP-NEXT: vmv.v.i v8, 0
|
|
; VLDEP-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, m1, ta, ma
|
|
; VLDEP-NEXT: vmv.v.i v10, 0
|
|
; VLDEP-NEXT: vmv1r.v v0, v9
|
|
; VLDEP-NEXT: vmerge.vim v9, v10, 1, v0
|
|
; VLDEP-NEXT: addi a0, a0, -5
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, m1, ta, ma
|
|
; VLDEP-NEXT: vslidedown.vi v9, v9, 5
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m1, ta, ma
|
|
; VLDEP-NEXT: vslideup.vx v9, v8, a0
|
|
; VLDEP-NEXT: vmsne.vi v0, v9, 0
|
|
; VLDEP-NEXT: ret
|
|
|
|
%v = call <vscale x 8 x i1> @llvm.experimental.vp.splice.nxv8i1(<vscale x 8 x i1> %va, <vscale x 8 x i1> %vb, i32 5, <vscale x 8 x i1> splat (i1 1), i32 %evla, i32 %evlb)
|
|
ret <vscale x 8 x i1> %v
|
|
}
|
|
|
|
define <vscale x 8 x i1> @test_vp_splice_nxv8i1_negative_offset(<vscale x 8 x i1> %va, <vscale x 8 x i1> %vb, i32 zeroext %evla, i32 zeroext %evlb) #0 {
|
|
; NOVLDEP-LABEL: test_vp_splice_nxv8i1_negative_offset:
|
|
; NOVLDEP: # %bb.0:
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m1, ta, ma
|
|
; NOVLDEP-NEXT: vmv1r.v v9, v0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v8
|
|
; NOVLDEP-NEXT: vmv.v.i v8, 0
|
|
; NOVLDEP-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a0, e8, m1, ta, ma
|
|
; NOVLDEP-NEXT: vmv.v.i v10, 0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v9
|
|
; NOVLDEP-NEXT: vmerge.vim v9, v10, 1, v0
|
|
; NOVLDEP-NEXT: addi a0, a0, -5
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m1, ta, ma
|
|
; NOVLDEP-NEXT: vslidedown.vx v9, v9, a0
|
|
; NOVLDEP-NEXT: vslideup.vi v9, v8, 5
|
|
; NOVLDEP-NEXT: vmsne.vi v0, v9, 0
|
|
; NOVLDEP-NEXT: ret
|
|
;
|
|
; VLDEP-LABEL: test_vp_splice_nxv8i1_negative_offset:
|
|
; VLDEP: # %bb.0:
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m1, ta, ma
|
|
; VLDEP-NEXT: vmv1r.v v9, v0
|
|
; VLDEP-NEXT: vmv1r.v v0, v8
|
|
; VLDEP-NEXT: vmv.v.i v8, 0
|
|
; VLDEP-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, m1, ta, ma
|
|
; VLDEP-NEXT: vmv.v.i v10, 0
|
|
; VLDEP-NEXT: vmv1r.v v0, v9
|
|
; VLDEP-NEXT: vmerge.vim v9, v10, 1, v0
|
|
; VLDEP-NEXT: addi a0, a0, -5
|
|
; VLDEP-NEXT: vsetivli zero, 5, e8, m1, ta, ma
|
|
; VLDEP-NEXT: vslidedown.vx v9, v9, a0
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m1, ta, ma
|
|
; VLDEP-NEXT: vslideup.vi v9, v8, 5
|
|
; VLDEP-NEXT: vmsne.vi v0, v9, 0
|
|
; VLDEP-NEXT: ret
|
|
|
|
%v = call <vscale x 8 x i1> @llvm.experimental.vp.splice.nxv8i1(<vscale x 8 x i1> %va, <vscale x 8 x i1> %vb, i32 -5, <vscale x 8 x i1> splat (i1 1), i32 %evla, i32 %evlb)
|
|
ret <vscale x 8 x i1> %v
|
|
}
|
|
|
|
define <vscale x 8 x i1> @test_vp_splice_nxv8i1_masked(<vscale x 8 x i1> %va, <vscale x 8 x i1> %vb, <vscale x 8 x i1> %mask, i32 zeroext %evla, i32 zeroext %evlb) #0 {
|
|
; NOVLDEP-LABEL: test_vp_splice_nxv8i1_masked:
|
|
; NOVLDEP: # %bb.0:
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m1, ta, ma
|
|
; NOVLDEP-NEXT: vmv1r.v v10, v0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v8
|
|
; NOVLDEP-NEXT: vmv.v.i v8, 0
|
|
; NOVLDEP-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a0, e8, m1, ta, ma
|
|
; NOVLDEP-NEXT: vmv.v.i v11, 0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v10
|
|
; NOVLDEP-NEXT: vmerge.vim v10, v11, 1, v0
|
|
; NOVLDEP-NEXT: addi a0, a0, -5
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v9
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m1, ta, ma
|
|
; NOVLDEP-NEXT: vslidedown.vi v10, v10, 5, v0.t
|
|
; NOVLDEP-NEXT: vsetvli zero, zero, e8, m1, ta, mu
|
|
; NOVLDEP-NEXT: vslideup.vx v10, v8, a0, v0.t
|
|
; NOVLDEP-NEXT: vsetvli zero, zero, e8, m1, ta, ma
|
|
; NOVLDEP-NEXT: vmsne.vi v0, v10, 0, v0.t
|
|
; NOVLDEP-NEXT: ret
|
|
;
|
|
; VLDEP-LABEL: test_vp_splice_nxv8i1_masked:
|
|
; VLDEP: # %bb.0:
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m1, ta, ma
|
|
; VLDEP-NEXT: vmv1r.v v10, v0
|
|
; VLDEP-NEXT: vmv1r.v v0, v8
|
|
; VLDEP-NEXT: vmv.v.i v8, 0
|
|
; VLDEP-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, m1, ta, ma
|
|
; VLDEP-NEXT: vmv.v.i v11, 0
|
|
; VLDEP-NEXT: vmv1r.v v0, v10
|
|
; VLDEP-NEXT: vmerge.vim v10, v11, 1, v0
|
|
; VLDEP-NEXT: addi a0, a0, -5
|
|
; VLDEP-NEXT: vmv1r.v v0, v9
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, m1, ta, ma
|
|
; VLDEP-NEXT: vslidedown.vi v10, v10, 5, v0.t
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m1, ta, mu
|
|
; VLDEP-NEXT: vslideup.vx v10, v8, a0, v0.t
|
|
; VLDEP-NEXT: vsetvli zero, zero, e8, m1, ta, ma
|
|
; VLDEP-NEXT: vmsne.vi v0, v10, 0, v0.t
|
|
; VLDEP-NEXT: ret
|
|
%v = call <vscale x 8 x i1> @llvm.experimental.vp.splice.nxv8i1(<vscale x 8 x i1> %va, <vscale x 8 x i1> %vb, i32 5, <vscale x 8 x i1> %mask, i32 %evla, i32 %evlb)
|
|
ret <vscale x 8 x i1> %v
|
|
}
|
|
|
|
define <vscale x 16 x i1> @test_vp_splice_nxv16i1(<vscale x 16 x i1> %va, <vscale x 16 x i1> %vb, i32 zeroext %evla, i32 zeroext %evlb) #0 {
|
|
; NOVLDEP-LABEL: test_vp_splice_nxv16i1:
|
|
; NOVLDEP: # %bb.0:
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m2, ta, ma
|
|
; NOVLDEP-NEXT: vmv1r.v v9, v0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v8
|
|
; NOVLDEP-NEXT: vmv.v.i v10, 0
|
|
; NOVLDEP-NEXT: vmerge.vim v10, v10, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a0, e8, m2, ta, ma
|
|
; NOVLDEP-NEXT: vmv.v.i v12, 0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v9
|
|
; NOVLDEP-NEXT: vmerge.vim v8, v12, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m2, ta, ma
|
|
; NOVLDEP-NEXT: vslidedown.vi v8, v8, 5
|
|
; NOVLDEP-NEXT: addi a0, a0, -5
|
|
; NOVLDEP-NEXT: vslideup.vx v8, v10, a0
|
|
; NOVLDEP-NEXT: vmsne.vi v0, v8, 0
|
|
; NOVLDEP-NEXT: ret
|
|
;
|
|
; VLDEP-LABEL: test_vp_splice_nxv16i1:
|
|
; VLDEP: # %bb.0:
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m2, ta, ma
|
|
; VLDEP-NEXT: vmv1r.v v9, v0
|
|
; VLDEP-NEXT: vmv1r.v v0, v8
|
|
; VLDEP-NEXT: vmv.v.i v10, 0
|
|
; VLDEP-NEXT: vmerge.vim v10, v10, 1, v0
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, m2, ta, ma
|
|
; VLDEP-NEXT: vmv.v.i v12, 0
|
|
; VLDEP-NEXT: vmv1r.v v0, v9
|
|
; VLDEP-NEXT: vmerge.vim v8, v12, 1, v0
|
|
; VLDEP-NEXT: addi a0, a0, -5
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, m2, ta, ma
|
|
; VLDEP-NEXT: vslidedown.vi v8, v8, 5
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m2, ta, ma
|
|
; VLDEP-NEXT: vslideup.vx v8, v10, a0
|
|
; VLDEP-NEXT: vmsne.vi v0, v8, 0
|
|
; VLDEP-NEXT: ret
|
|
|
|
%v = call <vscale x 16 x i1> @llvm.experimental.vp.splice.nxv16i1(<vscale x 16 x i1> %va, <vscale x 16 x i1> %vb, i32 5, <vscale x 16 x i1> splat (i1 1), i32 %evla, i32 %evlb)
|
|
ret <vscale x 16 x i1> %v
|
|
}
|
|
|
|
define <vscale x 16 x i1> @test_vp_splice_nxv16i1_negative_offset(<vscale x 16 x i1> %va, <vscale x 16 x i1> %vb, i32 zeroext %evla, i32 zeroext %evlb) #0 {
|
|
; NOVLDEP-LABEL: test_vp_splice_nxv16i1_negative_offset:
|
|
; NOVLDEP: # %bb.0:
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m2, ta, ma
|
|
; NOVLDEP-NEXT: vmv1r.v v9, v0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v8
|
|
; NOVLDEP-NEXT: vmv.v.i v10, 0
|
|
; NOVLDEP-NEXT: vmerge.vim v10, v10, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a0, e8, m2, ta, ma
|
|
; NOVLDEP-NEXT: vmv.v.i v12, 0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v9
|
|
; NOVLDEP-NEXT: vmerge.vim v8, v12, 1, v0
|
|
; NOVLDEP-NEXT: addi a0, a0, -5
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m2, ta, ma
|
|
; NOVLDEP-NEXT: vslidedown.vx v8, v8, a0
|
|
; NOVLDEP-NEXT: vslideup.vi v8, v10, 5
|
|
; NOVLDEP-NEXT: vmsne.vi v0, v8, 0
|
|
; NOVLDEP-NEXT: ret
|
|
;
|
|
; VLDEP-LABEL: test_vp_splice_nxv16i1_negative_offset:
|
|
; VLDEP: # %bb.0:
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m2, ta, ma
|
|
; VLDEP-NEXT: vmv1r.v v9, v0
|
|
; VLDEP-NEXT: vmv1r.v v0, v8
|
|
; VLDEP-NEXT: vmv.v.i v10, 0
|
|
; VLDEP-NEXT: vmerge.vim v10, v10, 1, v0
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, m2, ta, ma
|
|
; VLDEP-NEXT: vmv.v.i v12, 0
|
|
; VLDEP-NEXT: vmv1r.v v0, v9
|
|
; VLDEP-NEXT: vmerge.vim v8, v12, 1, v0
|
|
; VLDEP-NEXT: addi a0, a0, -5
|
|
; VLDEP-NEXT: vsetivli zero, 5, e8, m2, ta, ma
|
|
; VLDEP-NEXT: vslidedown.vx v8, v8, a0
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m2, ta, ma
|
|
; VLDEP-NEXT: vslideup.vi v8, v10, 5
|
|
; VLDEP-NEXT: vmsne.vi v0, v8, 0
|
|
; VLDEP-NEXT: ret
|
|
|
|
%v = call <vscale x 16 x i1> @llvm.experimental.vp.splice.nxv16i1(<vscale x 16 x i1> %va, <vscale x 16 x i1> %vb, i32 -5, <vscale x 16 x i1> splat (i1 1), i32 %evla, i32 %evlb)
|
|
ret <vscale x 16 x i1> %v
|
|
}
|
|
|
|
define <vscale x 16 x i1> @test_vp_splice_nxv16i1_masked(<vscale x 16 x i1> %va, <vscale x 16 x i1> %vb, <vscale x 16 x i1> %mask, i32 zeroext %evla, i32 zeroext %evlb) #0 {
|
|
; NOVLDEP-LABEL: test_vp_splice_nxv16i1_masked:
|
|
; NOVLDEP: # %bb.0:
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m2, ta, ma
|
|
; NOVLDEP-NEXT: vmv1r.v v10, v0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v8
|
|
; NOVLDEP-NEXT: vmv.v.i v12, 0
|
|
; NOVLDEP-NEXT: vmerge.vim v12, v12, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a0, e8, m2, ta, ma
|
|
; NOVLDEP-NEXT: vmv.v.i v14, 0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v10
|
|
; NOVLDEP-NEXT: vmerge.vim v10, v14, 1, v0
|
|
; NOVLDEP-NEXT: addi a0, a0, -5
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v9
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m2, ta, ma
|
|
; NOVLDEP-NEXT: vslidedown.vi v10, v10, 5, v0.t
|
|
; NOVLDEP-NEXT: vsetvli zero, zero, e8, m2, ta, mu
|
|
; NOVLDEP-NEXT: vslideup.vx v10, v12, a0, v0.t
|
|
; NOVLDEP-NEXT: vsetvli zero, zero, e8, m2, ta, ma
|
|
; NOVLDEP-NEXT: vmsne.vi v8, v10, 0, v0.t
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v8
|
|
; NOVLDEP-NEXT: ret
|
|
;
|
|
; VLDEP-LABEL: test_vp_splice_nxv16i1_masked:
|
|
; VLDEP: # %bb.0:
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m2, ta, ma
|
|
; VLDEP-NEXT: vmv1r.v v10, v0
|
|
; VLDEP-NEXT: vmv1r.v v0, v8
|
|
; VLDEP-NEXT: vmv.v.i v12, 0
|
|
; VLDEP-NEXT: vmerge.vim v12, v12, 1, v0
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, m2, ta, ma
|
|
; VLDEP-NEXT: vmv.v.i v14, 0
|
|
; VLDEP-NEXT: vmv1r.v v0, v10
|
|
; VLDEP-NEXT: vmerge.vim v10, v14, 1, v0
|
|
; VLDEP-NEXT: addi a0, a0, -5
|
|
; VLDEP-NEXT: vmv1r.v v0, v9
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, m2, ta, ma
|
|
; VLDEP-NEXT: vslidedown.vi v10, v10, 5, v0.t
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m2, ta, mu
|
|
; VLDEP-NEXT: vslideup.vx v10, v12, a0, v0.t
|
|
; VLDEP-NEXT: vsetvli zero, zero, e8, m2, ta, ma
|
|
; VLDEP-NEXT: vmsne.vi v8, v10, 0, v0.t
|
|
; VLDEP-NEXT: vmv1r.v v0, v8
|
|
; VLDEP-NEXT: ret
|
|
%v = call <vscale x 16 x i1> @llvm.experimental.vp.splice.nxv16i1(<vscale x 16 x i1> %va, <vscale x 16 x i1> %vb, i32 5, <vscale x 16 x i1> %mask, i32 %evla, i32 %evlb)
|
|
ret <vscale x 16 x i1> %v
|
|
}
|
|
|
|
define <vscale x 32 x i1> @test_vp_splice_nxv32i1(<vscale x 32 x i1> %va, <vscale x 32 x i1> %vb, i32 zeroext %evla, i32 zeroext %evlb) #0 {
|
|
; NOVLDEP-LABEL: test_vp_splice_nxv32i1:
|
|
; NOVLDEP: # %bb.0:
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m4, ta, ma
|
|
; NOVLDEP-NEXT: vmv1r.v v9, v0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v8
|
|
; NOVLDEP-NEXT: vmv.v.i v12, 0
|
|
; NOVLDEP-NEXT: vmerge.vim v12, v12, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a0, e8, m4, ta, ma
|
|
; NOVLDEP-NEXT: vmv.v.i v16, 0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v9
|
|
; NOVLDEP-NEXT: vmerge.vim v8, v16, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m4, ta, ma
|
|
; NOVLDEP-NEXT: vslidedown.vi v8, v8, 5
|
|
; NOVLDEP-NEXT: addi a0, a0, -5
|
|
; NOVLDEP-NEXT: vslideup.vx v8, v12, a0
|
|
; NOVLDEP-NEXT: vmsne.vi v0, v8, 0
|
|
; NOVLDEP-NEXT: ret
|
|
;
|
|
; VLDEP-LABEL: test_vp_splice_nxv32i1:
|
|
; VLDEP: # %bb.0:
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m4, ta, ma
|
|
; VLDEP-NEXT: vmv1r.v v9, v0
|
|
; VLDEP-NEXT: vmv1r.v v0, v8
|
|
; VLDEP-NEXT: vmv.v.i v12, 0
|
|
; VLDEP-NEXT: vmerge.vim v12, v12, 1, v0
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, m4, ta, ma
|
|
; VLDEP-NEXT: vmv.v.i v16, 0
|
|
; VLDEP-NEXT: vmv1r.v v0, v9
|
|
; VLDEP-NEXT: vmerge.vim v8, v16, 1, v0
|
|
; VLDEP-NEXT: addi a0, a0, -5
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, m4, ta, ma
|
|
; VLDEP-NEXT: vslidedown.vi v8, v8, 5
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m4, ta, ma
|
|
; VLDEP-NEXT: vslideup.vx v8, v12, a0
|
|
; VLDEP-NEXT: vmsne.vi v0, v8, 0
|
|
; VLDEP-NEXT: ret
|
|
|
|
%v = call <vscale x 32 x i1> @llvm.experimental.vp.splice.nxv32i1(<vscale x 32 x i1> %va, <vscale x 32 x i1> %vb, i32 5, <vscale x 32 x i1> splat (i1 1), i32 %evla, i32 %evlb)
|
|
ret <vscale x 32 x i1> %v
|
|
}
|
|
|
|
define <vscale x 32 x i1> @test_vp_splice_nxv32i1_negative_offset(<vscale x 32 x i1> %va, <vscale x 32 x i1> %vb, i32 zeroext %evla, i32 zeroext %evlb) #0 {
|
|
; NOVLDEP-LABEL: test_vp_splice_nxv32i1_negative_offset:
|
|
; NOVLDEP: # %bb.0:
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m4, ta, ma
|
|
; NOVLDEP-NEXT: vmv1r.v v9, v0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v8
|
|
; NOVLDEP-NEXT: vmv.v.i v12, 0
|
|
; NOVLDEP-NEXT: vmerge.vim v12, v12, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a0, e8, m4, ta, ma
|
|
; NOVLDEP-NEXT: vmv.v.i v16, 0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v9
|
|
; NOVLDEP-NEXT: vmerge.vim v8, v16, 1, v0
|
|
; NOVLDEP-NEXT: addi a0, a0, -5
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m4, ta, ma
|
|
; NOVLDEP-NEXT: vslidedown.vx v8, v8, a0
|
|
; NOVLDEP-NEXT: vslideup.vi v8, v12, 5
|
|
; NOVLDEP-NEXT: vmsne.vi v0, v8, 0
|
|
; NOVLDEP-NEXT: ret
|
|
;
|
|
; VLDEP-LABEL: test_vp_splice_nxv32i1_negative_offset:
|
|
; VLDEP: # %bb.0:
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m4, ta, ma
|
|
; VLDEP-NEXT: vmv1r.v v9, v0
|
|
; VLDEP-NEXT: vmv1r.v v0, v8
|
|
; VLDEP-NEXT: vmv.v.i v12, 0
|
|
; VLDEP-NEXT: vmerge.vim v12, v12, 1, v0
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, m4, ta, ma
|
|
; VLDEP-NEXT: vmv.v.i v16, 0
|
|
; VLDEP-NEXT: vmv1r.v v0, v9
|
|
; VLDEP-NEXT: vmerge.vim v8, v16, 1, v0
|
|
; VLDEP-NEXT: addi a0, a0, -5
|
|
; VLDEP-NEXT: vsetivli zero, 5, e8, m4, ta, ma
|
|
; VLDEP-NEXT: vslidedown.vx v8, v8, a0
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m4, ta, ma
|
|
; VLDEP-NEXT: vslideup.vi v8, v12, 5
|
|
; VLDEP-NEXT: vmsne.vi v0, v8, 0
|
|
; VLDEP-NEXT: ret
|
|
|
|
%v = call <vscale x 32 x i1> @llvm.experimental.vp.splice.nxv32i1(<vscale x 32 x i1> %va, <vscale x 32 x i1> %vb, i32 -5, <vscale x 32 x i1> splat (i1 1), i32 %evla, i32 %evlb)
|
|
ret <vscale x 32 x i1> %v
|
|
}
|
|
|
|
define <vscale x 32 x i1> @test_vp_splice_nxv32i1_masked(<vscale x 32 x i1> %va, <vscale x 32 x i1> %vb, <vscale x 32 x i1> %mask, i32 zeroext %evla, i32 zeroext %evlb) #0 {
|
|
; NOVLDEP-LABEL: test_vp_splice_nxv32i1_masked:
|
|
; NOVLDEP: # %bb.0:
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m4, ta, ma
|
|
; NOVLDEP-NEXT: vmv1r.v v10, v0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v8
|
|
; NOVLDEP-NEXT: vmv.v.i v12, 0
|
|
; NOVLDEP-NEXT: vmerge.vim v12, v12, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a0, e8, m4, ta, ma
|
|
; NOVLDEP-NEXT: vmv.v.i v16, 0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v10
|
|
; NOVLDEP-NEXT: vmerge.vim v16, v16, 1, v0
|
|
; NOVLDEP-NEXT: addi a0, a0, -5
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v9
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m4, ta, ma
|
|
; NOVLDEP-NEXT: vslidedown.vi v16, v16, 5, v0.t
|
|
; NOVLDEP-NEXT: vsetvli zero, zero, e8, m4, ta, mu
|
|
; NOVLDEP-NEXT: vslideup.vx v16, v12, a0, v0.t
|
|
; NOVLDEP-NEXT: vsetvli zero, zero, e8, m4, ta, ma
|
|
; NOVLDEP-NEXT: vmsne.vi v8, v16, 0, v0.t
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v8
|
|
; NOVLDEP-NEXT: ret
|
|
;
|
|
; VLDEP-LABEL: test_vp_splice_nxv32i1_masked:
|
|
; VLDEP: # %bb.0:
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m4, ta, ma
|
|
; VLDEP-NEXT: vmv1r.v v10, v0
|
|
; VLDEP-NEXT: vmv1r.v v0, v8
|
|
; VLDEP-NEXT: vmv.v.i v12, 0
|
|
; VLDEP-NEXT: vmerge.vim v12, v12, 1, v0
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, m4, ta, ma
|
|
; VLDEP-NEXT: vmv.v.i v16, 0
|
|
; VLDEP-NEXT: vmv1r.v v0, v10
|
|
; VLDEP-NEXT: vmerge.vim v16, v16, 1, v0
|
|
; VLDEP-NEXT: addi a0, a0, -5
|
|
; VLDEP-NEXT: vmv1r.v v0, v9
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, m4, ta, ma
|
|
; VLDEP-NEXT: vslidedown.vi v16, v16, 5, v0.t
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m4, ta, mu
|
|
; VLDEP-NEXT: vslideup.vx v16, v12, a0, v0.t
|
|
; VLDEP-NEXT: vsetvli zero, zero, e8, m4, ta, ma
|
|
; VLDEP-NEXT: vmsne.vi v8, v16, 0, v0.t
|
|
; VLDEP-NEXT: vmv1r.v v0, v8
|
|
; VLDEP-NEXT: ret
|
|
%v = call <vscale x 32 x i1> @llvm.experimental.vp.splice.nxv32i1(<vscale x 32 x i1> %va, <vscale x 32 x i1> %vb, i32 5, <vscale x 32 x i1> %mask, i32 %evla, i32 %evlb)
|
|
ret <vscale x 32 x i1> %v
|
|
}
|
|
|
|
define <vscale x 64 x i1> @test_vp_splice_nxv64i1(<vscale x 64 x i1> %va, <vscale x 64 x i1> %vb, i32 zeroext %evla, i32 zeroext %evlb) #0 {
|
|
; NOVLDEP-LABEL: test_vp_splice_nxv64i1:
|
|
; NOVLDEP: # %bb.0:
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vmv1r.v v9, v0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v8
|
|
; NOVLDEP-NEXT: vmv.v.i v16, 0
|
|
; NOVLDEP-NEXT: vmerge.vim v16, v16, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a0, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vmv.v.i v24, 0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v9
|
|
; NOVLDEP-NEXT: vmerge.vim v8, v24, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vslidedown.vi v8, v8, 5
|
|
; NOVLDEP-NEXT: addi a0, a0, -5
|
|
; NOVLDEP-NEXT: vslideup.vx v8, v16, a0
|
|
; NOVLDEP-NEXT: vmsne.vi v0, v8, 0
|
|
; NOVLDEP-NEXT: ret
|
|
;
|
|
; VLDEP-LABEL: test_vp_splice_nxv64i1:
|
|
; VLDEP: # %bb.0:
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vmv1r.v v9, v0
|
|
; VLDEP-NEXT: vmv1r.v v0, v8
|
|
; VLDEP-NEXT: vmv.v.i v16, 0
|
|
; VLDEP-NEXT: vmerge.vim v16, v16, 1, v0
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vmv.v.i v24, 0
|
|
; VLDEP-NEXT: vmv1r.v v0, v9
|
|
; VLDEP-NEXT: vmerge.vim v8, v24, 1, v0
|
|
; VLDEP-NEXT: addi a0, a0, -5
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vslidedown.vi v8, v8, 5
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vslideup.vx v8, v16, a0
|
|
; VLDEP-NEXT: vmsne.vi v0, v8, 0
|
|
; VLDEP-NEXT: ret
|
|
|
|
%v = call <vscale x 64 x i1> @llvm.experimental.vp.splice.nxv64i1(<vscale x 64 x i1> %va, <vscale x 64 x i1> %vb, i32 5, <vscale x 64 x i1> splat (i1 1), i32 %evla, i32 %evlb)
|
|
ret <vscale x 64 x i1> %v
|
|
}
|
|
|
|
define <vscale x 64 x i1> @test_vp_splice_nxv64i1_negative_offset(<vscale x 64 x i1> %va, <vscale x 64 x i1> %vb, i32 zeroext %evla, i32 zeroext %evlb) #0 {
|
|
; NOVLDEP-LABEL: test_vp_splice_nxv64i1_negative_offset:
|
|
; NOVLDEP: # %bb.0:
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vmv1r.v v9, v0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v8
|
|
; NOVLDEP-NEXT: vmv.v.i v16, 0
|
|
; NOVLDEP-NEXT: vmerge.vim v16, v16, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a0, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vmv.v.i v24, 0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v9
|
|
; NOVLDEP-NEXT: vmerge.vim v8, v24, 1, v0
|
|
; NOVLDEP-NEXT: addi a0, a0, -5
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vslidedown.vx v8, v8, a0
|
|
; NOVLDEP-NEXT: vslideup.vi v8, v16, 5
|
|
; NOVLDEP-NEXT: vmsne.vi v0, v8, 0
|
|
; NOVLDEP-NEXT: ret
|
|
;
|
|
; VLDEP-LABEL: test_vp_splice_nxv64i1_negative_offset:
|
|
; VLDEP: # %bb.0:
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vmv1r.v v9, v0
|
|
; VLDEP-NEXT: vmv1r.v v0, v8
|
|
; VLDEP-NEXT: vmv.v.i v16, 0
|
|
; VLDEP-NEXT: vmerge.vim v16, v16, 1, v0
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vmv.v.i v24, 0
|
|
; VLDEP-NEXT: vmv1r.v v0, v9
|
|
; VLDEP-NEXT: vmerge.vim v8, v24, 1, v0
|
|
; VLDEP-NEXT: addi a0, a0, -5
|
|
; VLDEP-NEXT: vsetivli zero, 5, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vslidedown.vx v8, v8, a0
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vslideup.vi v8, v16, 5
|
|
; VLDEP-NEXT: vmsne.vi v0, v8, 0
|
|
; VLDEP-NEXT: ret
|
|
|
|
%v = call <vscale x 64 x i1> @llvm.experimental.vp.splice.nxv64i1(<vscale x 64 x i1> %va, <vscale x 64 x i1> %vb, i32 -5, <vscale x 64 x i1> splat (i1 1), i32 %evla, i32 %evlb)
|
|
ret <vscale x 64 x i1> %v
|
|
}
|
|
|
|
define <vscale x 64 x i1> @test_vp_splice_nxv64i1_masked(<vscale x 64 x i1> %va, <vscale x 64 x i1> %vb, <vscale x 64 x i1> %mask, i32 zeroext %evla, i32 zeroext %evlb) #0 {
|
|
; NOVLDEP-LABEL: test_vp_splice_nxv64i1_masked:
|
|
; NOVLDEP: # %bb.0:
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vmv1r.v v10, v0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v8
|
|
; NOVLDEP-NEXT: vmv.v.i v16, 0
|
|
; NOVLDEP-NEXT: vmerge.vim v16, v16, 1, v0
|
|
; NOVLDEP-NEXT: vsetvli zero, a0, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vmv.v.i v24, 0
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v10
|
|
; NOVLDEP-NEXT: vmerge.vim v24, v24, 1, v0
|
|
; NOVLDEP-NEXT: addi a0, a0, -5
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v9
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vslidedown.vi v24, v24, 5, v0.t
|
|
; NOVLDEP-NEXT: vsetvli zero, zero, e8, m8, ta, mu
|
|
; NOVLDEP-NEXT: vslideup.vx v24, v16, a0, v0.t
|
|
; NOVLDEP-NEXT: vsetvli zero, zero, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vmsne.vi v8, v24, 0, v0.t
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v8
|
|
; NOVLDEP-NEXT: ret
|
|
;
|
|
; VLDEP-LABEL: test_vp_splice_nxv64i1_masked:
|
|
; VLDEP: # %bb.0:
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vmv1r.v v10, v0
|
|
; VLDEP-NEXT: vmv1r.v v0, v8
|
|
; VLDEP-NEXT: vmv.v.i v16, 0
|
|
; VLDEP-NEXT: vmerge.vim v16, v16, 1, v0
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vmv.v.i v24, 0
|
|
; VLDEP-NEXT: vmv1r.v v0, v10
|
|
; VLDEP-NEXT: vmerge.vim v24, v24, 1, v0
|
|
; VLDEP-NEXT: addi a0, a0, -5
|
|
; VLDEP-NEXT: vmv1r.v v0, v9
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vslidedown.vi v24, v24, 5, v0.t
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m8, ta, mu
|
|
; VLDEP-NEXT: vslideup.vx v24, v16, a0, v0.t
|
|
; VLDEP-NEXT: vsetvli zero, zero, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vmsne.vi v8, v24, 0, v0.t
|
|
; VLDEP-NEXT: vmv1r.v v0, v8
|
|
; VLDEP-NEXT: ret
|
|
%v = call <vscale x 64 x i1> @llvm.experimental.vp.splice.nxv64i1(<vscale x 64 x i1> %va, <vscale x 64 x i1> %vb, i32 5, <vscale x 64 x i1> %mask, i32 %evla, i32 %evlb)
|
|
ret <vscale x 64 x i1> %v
|
|
}
|
|
|
|
; nxv128i1 is split via the generic SplitVecRes_VP_SPLICE stack round-trip. The
|
|
; i1 elements must be widened to a byte type; otherwise getVectorElementPointer
|
|
; asserts and the negative-offset stride is 0.
|
|
define <vscale x 128 x i1> @test_vp_splice_nxv128i1(<vscale x 128 x i1> %va, <vscale x 128 x i1> %vb, i32 zeroext %evla, i32 zeroext %evlb) #0 {
|
|
; NOVLDEP-LABEL: test_vp_splice_nxv128i1:
|
|
; NOVLDEP: # %bb.0:
|
|
; NOVLDEP-NEXT: csrr a2, vlenb
|
|
; NOVLDEP-NEXT: slli a4, a2, 4
|
|
; NOVLDEP-NEXT: addi a4, a4, -1
|
|
; NOVLDEP-NEXT: mv a5, a0
|
|
; NOVLDEP-NEXT: mv a3, a0
|
|
; NOVLDEP-NEXT: bltu a0, a4, .LBB21_2
|
|
; NOVLDEP-NEXT: # %bb.1:
|
|
; NOVLDEP-NEXT: mv a3, a4
|
|
; NOVLDEP-NEXT: .LBB21_2:
|
|
; NOVLDEP-NEXT: addi sp, sp, -80
|
|
; NOVLDEP-NEXT: .cfi_def_cfa_offset 80
|
|
; NOVLDEP-NEXT: sd ra, 72(sp) # 8-byte Folded Spill
|
|
; NOVLDEP-NEXT: sd s0, 64(sp) # 8-byte Folded Spill
|
|
; NOVLDEP-NEXT: .cfi_offset ra, -8
|
|
; NOVLDEP-NEXT: .cfi_offset s0, -16
|
|
; NOVLDEP-NEXT: addi s0, sp, 80
|
|
; NOVLDEP-NEXT: .cfi_def_cfa s0, 0
|
|
; NOVLDEP-NEXT: csrr a4, vlenb
|
|
; NOVLDEP-NEXT: slli a4, a4, 5
|
|
; NOVLDEP-NEXT: sub sp, sp, a4
|
|
; NOVLDEP-NEXT: andi sp, sp, -64
|
|
; NOVLDEP-NEXT: addi a4, sp, 64
|
|
; NOVLDEP-NEXT: vsetvli a6, zero, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vmv.v.i v16, 0
|
|
; NOVLDEP-NEXT: slli a2, a2, 3
|
|
; NOVLDEP-NEXT: add a3, a4, a3
|
|
; NOVLDEP-NEXT: vmerge.vim v24, v16, 1, v0
|
|
; NOVLDEP-NEXT: bltu a5, a2, .LBB21_4
|
|
; NOVLDEP-NEXT: # %bb.3:
|
|
; NOVLDEP-NEXT: mv a5, a2
|
|
; NOVLDEP-NEXT: .LBB21_4:
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v8
|
|
; NOVLDEP-NEXT: vsetvli zero, a5, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vse8.v v24, (a4)
|
|
; NOVLDEP-NEXT: sub a5, a0, a2
|
|
; NOVLDEP-NEXT: add a4, a4, a2
|
|
; NOVLDEP-NEXT: sltu a0, a0, a5
|
|
; NOVLDEP-NEXT: addi a0, a0, -1
|
|
; NOVLDEP-NEXT: and a0, a0, a5
|
|
; NOVLDEP-NEXT: sub a5, a1, a2
|
|
; NOVLDEP-NEXT: vsetvli zero, a0, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vmerge.vim v24, v16, 1, v0
|
|
; NOVLDEP-NEXT: vse8.v v24, (a4)
|
|
; NOVLDEP-NEXT: add a4, a3, a2
|
|
; NOVLDEP-NEXT: sltu a0, a1, a5
|
|
; NOVLDEP-NEXT: addi a0, a0, -1
|
|
; NOVLDEP-NEXT: and a0, a0, a5
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v10
|
|
; NOVLDEP-NEXT: vsetvli zero, a0, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vmerge.vim v24, v16, 1, v0
|
|
; NOVLDEP-NEXT: vse8.v v24, (a4)
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v9
|
|
; NOVLDEP-NEXT: vsetvli a4, zero, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vmerge.vim v8, v16, 1, v0
|
|
; NOVLDEP-NEXT: bltu a1, a2, .LBB21_6
|
|
; NOVLDEP-NEXT: # %bb.5:
|
|
; NOVLDEP-NEXT: mv a1, a2
|
|
; NOVLDEP-NEXT: .LBB21_6:
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vse8.v v8, (a3)
|
|
; NOVLDEP-NEXT: addi a3, sp, 69
|
|
; NOVLDEP-NEXT: add a2, a3, a2
|
|
; NOVLDEP-NEXT: vsetvli zero, a0, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vle8.v v8, (a2)
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vle8.v v16, (a3)
|
|
; NOVLDEP-NEXT: vsetvli a0, zero, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vand.vi v24, v8, 1
|
|
; NOVLDEP-NEXT: vmsne.vi v8, v24, 0
|
|
; NOVLDEP-NEXT: vand.vi v16, v16, 1
|
|
; NOVLDEP-NEXT: vmsne.vi v0, v16, 0
|
|
; NOVLDEP-NEXT: addi sp, s0, -80
|
|
; NOVLDEP-NEXT: .cfi_def_cfa sp, 80
|
|
; NOVLDEP-NEXT: ld ra, 72(sp) # 8-byte Folded Reload
|
|
; NOVLDEP-NEXT: ld s0, 64(sp) # 8-byte Folded Reload
|
|
; NOVLDEP-NEXT: .cfi_restore ra
|
|
; NOVLDEP-NEXT: .cfi_restore s0
|
|
; NOVLDEP-NEXT: addi sp, sp, 80
|
|
; NOVLDEP-NEXT: .cfi_def_cfa_offset 0
|
|
; NOVLDEP-NEXT: ret
|
|
;
|
|
; VLDEP-LABEL: test_vp_splice_nxv128i1:
|
|
; VLDEP: # %bb.0:
|
|
; VLDEP-NEXT: csrr a2, vlenb
|
|
; VLDEP-NEXT: slli a4, a2, 4
|
|
; VLDEP-NEXT: addi a4, a4, -1
|
|
; VLDEP-NEXT: mv a5, a0
|
|
; VLDEP-NEXT: mv a3, a0
|
|
; VLDEP-NEXT: bltu a0, a4, .LBB21_2
|
|
; VLDEP-NEXT: # %bb.1:
|
|
; VLDEP-NEXT: mv a3, a4
|
|
; VLDEP-NEXT: .LBB21_2:
|
|
; VLDEP-NEXT: addi sp, sp, -80
|
|
; VLDEP-NEXT: .cfi_def_cfa_offset 80
|
|
; VLDEP-NEXT: sd ra, 72(sp) # 8-byte Folded Spill
|
|
; VLDEP-NEXT: sd s0, 64(sp) # 8-byte Folded Spill
|
|
; VLDEP-NEXT: .cfi_offset ra, -8
|
|
; VLDEP-NEXT: .cfi_offset s0, -16
|
|
; VLDEP-NEXT: addi s0, sp, 80
|
|
; VLDEP-NEXT: .cfi_def_cfa s0, 0
|
|
; VLDEP-NEXT: csrr a4, vlenb
|
|
; VLDEP-NEXT: slli a4, a4, 5
|
|
; VLDEP-NEXT: sub sp, sp, a4
|
|
; VLDEP-NEXT: andi sp, sp, -64
|
|
; VLDEP-NEXT: addi a4, sp, 64
|
|
; VLDEP-NEXT: vsetvli a6, zero, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vmv.v.i v16, 0
|
|
; VLDEP-NEXT: slli a2, a2, 3
|
|
; VLDEP-NEXT: add a3, a4, a3
|
|
; VLDEP-NEXT: vmerge.vim v24, v16, 1, v0
|
|
; VLDEP-NEXT: bltu a5, a2, .LBB21_4
|
|
; VLDEP-NEXT: # %bb.3:
|
|
; VLDEP-NEXT: mv a5, a2
|
|
; VLDEP-NEXT: .LBB21_4:
|
|
; VLDEP-NEXT: vmv1r.v v0, v8
|
|
; VLDEP-NEXT: vsetvli zero, a5, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vse8.v v24, (a4)
|
|
; VLDEP-NEXT: sub a5, a0, a2
|
|
; VLDEP-NEXT: add a4, a4, a2
|
|
; VLDEP-NEXT: sltu a0, a0, a5
|
|
; VLDEP-NEXT: addi a0, a0, -1
|
|
; VLDEP-NEXT: and a0, a0, a5
|
|
; VLDEP-NEXT: sub a5, a1, a2
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vmerge.vim v24, v16, 1, v0
|
|
; VLDEP-NEXT: vse8.v v24, (a4)
|
|
; VLDEP-NEXT: add a4, a3, a2
|
|
; VLDEP-NEXT: sltu a0, a1, a5
|
|
; VLDEP-NEXT: addi a0, a0, -1
|
|
; VLDEP-NEXT: and a0, a0, a5
|
|
; VLDEP-NEXT: vmv1r.v v0, v10
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vmerge.vim v24, v16, 1, v0
|
|
; VLDEP-NEXT: vse8.v v24, (a4)
|
|
; VLDEP-NEXT: vmv1r.v v0, v9
|
|
; VLDEP-NEXT: vsetvli a4, zero, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vmerge.vim v8, v16, 1, v0
|
|
; VLDEP-NEXT: bltu a1, a2, .LBB21_6
|
|
; VLDEP-NEXT: # %bb.5:
|
|
; VLDEP-NEXT: mv a1, a2
|
|
; VLDEP-NEXT: .LBB21_6:
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vse8.v v8, (a3)
|
|
; VLDEP-NEXT: addi a3, sp, 69
|
|
; VLDEP-NEXT: add a2, a3, a2
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vle8.v v8, (a2)
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vle8.v v16, (a3)
|
|
; VLDEP-NEXT: vsetvli a0, zero, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vand.vi v24, v8, 1
|
|
; VLDEP-NEXT: vmsne.vi v8, v24, 0
|
|
; VLDEP-NEXT: vand.vi v16, v16, 1
|
|
; VLDEP-NEXT: vmsne.vi v0, v16, 0
|
|
; VLDEP-NEXT: addi sp, s0, -80
|
|
; VLDEP-NEXT: .cfi_def_cfa sp, 80
|
|
; VLDEP-NEXT: ld ra, 72(sp) # 8-byte Folded Reload
|
|
; VLDEP-NEXT: ld s0, 64(sp) # 8-byte Folded Reload
|
|
; VLDEP-NEXT: .cfi_restore ra
|
|
; VLDEP-NEXT: .cfi_restore s0
|
|
; VLDEP-NEXT: addi sp, sp, 80
|
|
; VLDEP-NEXT: .cfi_def_cfa_offset 0
|
|
; VLDEP-NEXT: ret
|
|
%v = call <vscale x 128 x i1> @llvm.experimental.vp.splice.nxv128i1(<vscale x 128 x i1> %va, <vscale x 128 x i1> %vb, i32 5, <vscale x 128 x i1> splat (i1 1), i32 %evla, i32 %evlb)
|
|
ret <vscale x 128 x i1> %v
|
|
}
|
|
|
|
define <vscale x 128 x i1> @test_vp_splice_nxv128i1_negative_offset(<vscale x 128 x i1> %va, <vscale x 128 x i1> %vb, i32 zeroext %evla, i32 zeroext %evlb) #0 {
|
|
; NOVLDEP-LABEL: test_vp_splice_nxv128i1_negative_offset:
|
|
; NOVLDEP: # %bb.0:
|
|
; NOVLDEP-NEXT: csrr a3, vlenb
|
|
; NOVLDEP-NEXT: slli a4, a3, 4
|
|
; NOVLDEP-NEXT: addi a4, a4, -1
|
|
; NOVLDEP-NEXT: mv a6, a0
|
|
; NOVLDEP-NEXT: mv a2, a0
|
|
; NOVLDEP-NEXT: bltu a0, a4, .LBB22_2
|
|
; NOVLDEP-NEXT: # %bb.1:
|
|
; NOVLDEP-NEXT: mv a2, a4
|
|
; NOVLDEP-NEXT: .LBB22_2:
|
|
; NOVLDEP-NEXT: addi sp, sp, -80
|
|
; NOVLDEP-NEXT: .cfi_def_cfa_offset 80
|
|
; NOVLDEP-NEXT: sd ra, 72(sp) # 8-byte Folded Spill
|
|
; NOVLDEP-NEXT: sd s0, 64(sp) # 8-byte Folded Spill
|
|
; NOVLDEP-NEXT: .cfi_offset ra, -8
|
|
; NOVLDEP-NEXT: .cfi_offset s0, -16
|
|
; NOVLDEP-NEXT: addi s0, sp, 80
|
|
; NOVLDEP-NEXT: .cfi_def_cfa s0, 0
|
|
; NOVLDEP-NEXT: csrr a4, vlenb
|
|
; NOVLDEP-NEXT: slli a4, a4, 5
|
|
; NOVLDEP-NEXT: sub sp, sp, a4
|
|
; NOVLDEP-NEXT: andi sp, sp, -64
|
|
; NOVLDEP-NEXT: addi a5, sp, 64
|
|
; NOVLDEP-NEXT: vsetvli a4, zero, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vmv.v.i v16, 0
|
|
; NOVLDEP-NEXT: slli a3, a3, 3
|
|
; NOVLDEP-NEXT: add a4, a5, a2
|
|
; NOVLDEP-NEXT: vmerge.vim v24, v16, 1, v0
|
|
; NOVLDEP-NEXT: bltu a6, a3, .LBB22_4
|
|
; NOVLDEP-NEXT: # %bb.3:
|
|
; NOVLDEP-NEXT: mv a6, a3
|
|
; NOVLDEP-NEXT: .LBB22_4:
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v8
|
|
; NOVLDEP-NEXT: vsetvli zero, a6, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vse8.v v24, (a5)
|
|
; NOVLDEP-NEXT: sub a6, a0, a3
|
|
; NOVLDEP-NEXT: add a5, a5, a3
|
|
; NOVLDEP-NEXT: sltu a0, a0, a6
|
|
; NOVLDEP-NEXT: addi a0, a0, -1
|
|
; NOVLDEP-NEXT: and a0, a0, a6
|
|
; NOVLDEP-NEXT: sub a6, a1, a3
|
|
; NOVLDEP-NEXT: vsetvli zero, a0, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vmerge.vim v24, v16, 1, v0
|
|
; NOVLDEP-NEXT: vse8.v v24, (a5)
|
|
; NOVLDEP-NEXT: add a5, a4, a3
|
|
; NOVLDEP-NEXT: sltu a0, a1, a6
|
|
; NOVLDEP-NEXT: addi a0, a0, -1
|
|
; NOVLDEP-NEXT: and a0, a0, a6
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v10
|
|
; NOVLDEP-NEXT: vsetvli zero, a0, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vmerge.vim v24, v16, 1, v0
|
|
; NOVLDEP-NEXT: vse8.v v24, (a5)
|
|
; NOVLDEP-NEXT: vmv1r.v v0, v9
|
|
; NOVLDEP-NEXT: vsetvli a5, zero, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vmerge.vim v8, v16, 1, v0
|
|
; NOVLDEP-NEXT: bltu a1, a3, .LBB22_6
|
|
; NOVLDEP-NEXT: # %bb.5:
|
|
; NOVLDEP-NEXT: mv a1, a3
|
|
; NOVLDEP-NEXT: .LBB22_6:
|
|
; NOVLDEP-NEXT: li a5, 5
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vse8.v v8, (a4)
|
|
; NOVLDEP-NEXT: bltu a2, a5, .LBB22_8
|
|
; NOVLDEP-NEXT: # %bb.7:
|
|
; NOVLDEP-NEXT: li a2, 5
|
|
; NOVLDEP-NEXT: .LBB22_8:
|
|
; NOVLDEP-NEXT: sub a4, a4, a2
|
|
; NOVLDEP-NEXT: add a3, a4, a3
|
|
; NOVLDEP-NEXT: vsetvli zero, a0, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vle8.v v8, (a3)
|
|
; NOVLDEP-NEXT: vsetvli zero, a1, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vle8.v v16, (a4)
|
|
; NOVLDEP-NEXT: vsetvli a0, zero, e8, m8, ta, ma
|
|
; NOVLDEP-NEXT: vand.vi v24, v8, 1
|
|
; NOVLDEP-NEXT: vmsne.vi v8, v24, 0
|
|
; NOVLDEP-NEXT: vand.vi v16, v16, 1
|
|
; NOVLDEP-NEXT: vmsne.vi v0, v16, 0
|
|
; NOVLDEP-NEXT: addi sp, s0, -80
|
|
; NOVLDEP-NEXT: .cfi_def_cfa sp, 80
|
|
; NOVLDEP-NEXT: ld ra, 72(sp) # 8-byte Folded Reload
|
|
; NOVLDEP-NEXT: ld s0, 64(sp) # 8-byte Folded Reload
|
|
; NOVLDEP-NEXT: .cfi_restore ra
|
|
; NOVLDEP-NEXT: .cfi_restore s0
|
|
; NOVLDEP-NEXT: addi sp, sp, 80
|
|
; NOVLDEP-NEXT: .cfi_def_cfa_offset 0
|
|
; NOVLDEP-NEXT: ret
|
|
;
|
|
; VLDEP-LABEL: test_vp_splice_nxv128i1_negative_offset:
|
|
; VLDEP: # %bb.0:
|
|
; VLDEP-NEXT: csrr a3, vlenb
|
|
; VLDEP-NEXT: slli a4, a3, 4
|
|
; VLDEP-NEXT: addi a4, a4, -1
|
|
; VLDEP-NEXT: mv a6, a0
|
|
; VLDEP-NEXT: mv a2, a0
|
|
; VLDEP-NEXT: bltu a0, a4, .LBB22_2
|
|
; VLDEP-NEXT: # %bb.1:
|
|
; VLDEP-NEXT: mv a2, a4
|
|
; VLDEP-NEXT: .LBB22_2:
|
|
; VLDEP-NEXT: addi sp, sp, -80
|
|
; VLDEP-NEXT: .cfi_def_cfa_offset 80
|
|
; VLDEP-NEXT: sd ra, 72(sp) # 8-byte Folded Spill
|
|
; VLDEP-NEXT: sd s0, 64(sp) # 8-byte Folded Spill
|
|
; VLDEP-NEXT: .cfi_offset ra, -8
|
|
; VLDEP-NEXT: .cfi_offset s0, -16
|
|
; VLDEP-NEXT: addi s0, sp, 80
|
|
; VLDEP-NEXT: .cfi_def_cfa s0, 0
|
|
; VLDEP-NEXT: csrr a4, vlenb
|
|
; VLDEP-NEXT: slli a4, a4, 5
|
|
; VLDEP-NEXT: sub sp, sp, a4
|
|
; VLDEP-NEXT: andi sp, sp, -64
|
|
; VLDEP-NEXT: addi a5, sp, 64
|
|
; VLDEP-NEXT: vsetvli a4, zero, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vmv.v.i v16, 0
|
|
; VLDEP-NEXT: slli a3, a3, 3
|
|
; VLDEP-NEXT: add a4, a5, a2
|
|
; VLDEP-NEXT: vmerge.vim v24, v16, 1, v0
|
|
; VLDEP-NEXT: bltu a6, a3, .LBB22_4
|
|
; VLDEP-NEXT: # %bb.3:
|
|
; VLDEP-NEXT: mv a6, a3
|
|
; VLDEP-NEXT: .LBB22_4:
|
|
; VLDEP-NEXT: vmv1r.v v0, v8
|
|
; VLDEP-NEXT: vsetvli zero, a6, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vse8.v v24, (a5)
|
|
; VLDEP-NEXT: sub a6, a0, a3
|
|
; VLDEP-NEXT: add a5, a5, a3
|
|
; VLDEP-NEXT: sltu a0, a0, a6
|
|
; VLDEP-NEXT: addi a0, a0, -1
|
|
; VLDEP-NEXT: and a0, a0, a6
|
|
; VLDEP-NEXT: sub a6, a1, a3
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vmerge.vim v24, v16, 1, v0
|
|
; VLDEP-NEXT: vse8.v v24, (a5)
|
|
; VLDEP-NEXT: add a5, a4, a3
|
|
; VLDEP-NEXT: sltu a0, a1, a6
|
|
; VLDEP-NEXT: addi a0, a0, -1
|
|
; VLDEP-NEXT: and a0, a0, a6
|
|
; VLDEP-NEXT: vmv1r.v v0, v10
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vmerge.vim v24, v16, 1, v0
|
|
; VLDEP-NEXT: vse8.v v24, (a5)
|
|
; VLDEP-NEXT: vmv1r.v v0, v9
|
|
; VLDEP-NEXT: vsetvli a5, zero, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vmerge.vim v8, v16, 1, v0
|
|
; VLDEP-NEXT: bltu a1, a3, .LBB22_6
|
|
; VLDEP-NEXT: # %bb.5:
|
|
; VLDEP-NEXT: mv a1, a3
|
|
; VLDEP-NEXT: .LBB22_6:
|
|
; VLDEP-NEXT: li a5, 5
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vse8.v v8, (a4)
|
|
; VLDEP-NEXT: bltu a2, a5, .LBB22_8
|
|
; VLDEP-NEXT: # %bb.7:
|
|
; VLDEP-NEXT: li a2, 5
|
|
; VLDEP-NEXT: .LBB22_8:
|
|
; VLDEP-NEXT: sub a4, a4, a2
|
|
; VLDEP-NEXT: add a3, a4, a3
|
|
; VLDEP-NEXT: vsetvli zero, a0, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vle8.v v8, (a3)
|
|
; VLDEP-NEXT: vsetvli zero, a1, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vle8.v v16, (a4)
|
|
; VLDEP-NEXT: vsetvli a0, zero, e8, m8, ta, ma
|
|
; VLDEP-NEXT: vand.vi v24, v8, 1
|
|
; VLDEP-NEXT: vmsne.vi v8, v24, 0
|
|
; VLDEP-NEXT: vand.vi v16, v16, 1
|
|
; VLDEP-NEXT: vmsne.vi v0, v16, 0
|
|
; VLDEP-NEXT: addi sp, s0, -80
|
|
; VLDEP-NEXT: .cfi_def_cfa sp, 80
|
|
; VLDEP-NEXT: ld ra, 72(sp) # 8-byte Folded Reload
|
|
; VLDEP-NEXT: ld s0, 64(sp) # 8-byte Folded Reload
|
|
; VLDEP-NEXT: .cfi_restore ra
|
|
; VLDEP-NEXT: .cfi_restore s0
|
|
; VLDEP-NEXT: addi sp, sp, 80
|
|
; VLDEP-NEXT: .cfi_def_cfa_offset 0
|
|
; VLDEP-NEXT: ret
|
|
%v = call <vscale x 128 x i1> @llvm.experimental.vp.splice.nxv128i1(<vscale x 128 x i1> %va, <vscale x 128 x i1> %vb, i32 -5, <vscale x 128 x i1> splat (i1 1), i32 %evla, i32 %evlb)
|
|
ret <vscale x 128 x i1> %v
|
|
}
|
|
|
|
attributes #0 = { vscale_range(2,0) }
|