derek/TheRock
1
0 Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity
Files
main
TheRock/compiler/amd-llvm/llvm/test/CodeGen/AMDGPU/permlane16_opsel.ll
T
derek.holloway 7502bf64a4
CodeQL / Analyze (actions) (push) Canceled after 0s
Details
CodeQL / Analyze (python) (push) Canceled after 0s
Details
pre-commit / pre-commit (push) Canceled after 0s
Details
Multi-Arch CI / setup (push) Canceled after 0s
Details
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Details
Setup (Multi-Arch) / setup (push) Canceled after 0s
Details
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Details
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Details
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Details
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Details
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Details
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
Details
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Details
Multi-Arch CI / CI Summary (push) Canceled after 0s
Details
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Details
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Details
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Details
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Details
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Details
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Details
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Details
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Details
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Details
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Details
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Details
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Details
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
Details
Flatten subprojects into a single base monorepo
2026-09-02 16:34:47 -07:00

299 lines
19 KiB
LLVM
Raw Permalink Blame History

; RUN: llc -global-isel=0 -mtriple=amdgcn -mcpu=gfx1010 --stop-after=amdgpu-isel < %s | FileCheck -check-prefixes=SDAG,SDAG-GFX10 %s
; RUN: llc -global-isel=1 -new-reg-bank-select -mtriple=amdgcn -mcpu=gfx1010 --stop-after=instruction-select < %s | FileCheck -check-prefixes=GISEL %s
; RUN: llc -global-isel=0 -mtriple=amdgcn -mcpu=gfx1100 --stop-after=amdgpu-isel < %s | FileCheck -check-prefixes=SDAG,SDAG-GFX11 %s
; RUN: llc -global-isel=1 -new-reg-bank-select -mtriple=amdgcn -mcpu=gfx1100 --stop-after=instruction-select < %s | FileCheck -check-prefixes=GISEL %s
declare i32 @llvm.amdgcn.permlane16(i32, i32, i32, i32, i1, i1)
declare i32 @llvm.amdgcn.permlanex16(i32, i32, i32, i32, i1, i1)
declare i32 @llvm.amdgcn.workitem.id.x()
declare i32 @llvm.amdgcn.workitem.id.y()
define amdgpu_kernel void @v_permlane16_b32_vss(ptr addrspace(1) %out, i32 %src0, i32 %src1, i32 %src2) {
; SDAG: V_PERMLANE16_B32_e64 0, {{%[0-9]+}}, 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANE16_B32_e64 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%v = call i32 @llvm.amdgcn.permlane16(i32 %src0, i32 %src0, i32 %src1, i32 %src2, i1 false, i1 false)
store i32 %v, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @v_permlane16_b32_vii(ptr addrspace(1) %out, i32 %src0) {
; SDAG: V_PERMLANE16_B32_e64 0, {{%[0-9]+}}, 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANE16_B32_e64 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%v = call i32 @llvm.amdgcn.permlane16(i32 %src0, i32 %src0, i32 1, i32 2, i1 false, i1 false)
store i32 %v, ptr addrspace(1) %out
ret void
}
; FIXME-GFX10PLUS: It is allowed to have both immediates as literals
define amdgpu_kernel void @v_permlane16_b32_vll(ptr addrspace(1) %out, i32 %src0) {
; SDAG: V_PERMLANE16_B32_e64 0, {{%[0-9]+}}, 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANE16_B32_e64 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%v = call i32 @llvm.amdgcn.permlane16(i32 %src0, i32 %src0, i32 4660, i32 49617, i1 false, i1 false)
store i32 %v, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @v_permlane16_b32_vvv(ptr addrspace(1) %out, i32 %src0) {
; SDAG-GFX10: V_PERMLANE16_B32_e64 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; SDAG-GFX11: V_PERMLANE16_B32_e64 0, {{%[0-9]+}}, 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANE16_B32_e64 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%tidx = call i32 @llvm.amdgcn.workitem.id.x()
%tidy = call i32 @llvm.amdgcn.workitem.id.y()
%v = call i32 @llvm.amdgcn.permlane16(i32 %src0, i32 %src0, i32 %tidx, i32 %tidy, i1 false, i1 false)
store i32 %v, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @v_permlane16_b32_vvs(ptr addrspace(1) %out, i32 %src0, i32 %src2) {
; SDAG-GFX10: V_PERMLANE16_B32_e64 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; SDAG-GFX11: V_PERMLANE16_B32_e64 0, {{%[0-9]+}}, 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANE16_B32_e64 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%tidx = call i32 @llvm.amdgcn.workitem.id.x()
%v = call i32 @llvm.amdgcn.permlane16(i32 %src0, i32 %src0, i32 %tidx, i32 %src2, i1 false, i1 false)
store i32 %v, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @v_permlane16_b32_vsv(ptr addrspace(1) %out, i32 %src0, i32 %src1) {
; SDAG-GFX10: V_PERMLANE16_B32_e64 0, {{%[0-9]+}}, 0, killed {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; SDAG-GFX11: V_PERMLANE16_B32_e64 0, {{%[0-9]+}}, 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANE16_B32_e64 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%tidy = call i32 @llvm.amdgcn.workitem.id.y()
%v = call i32 @llvm.amdgcn.permlane16(i32 %src0, i32 %src0, i32 %src1, i32 %tidy, i1 false, i1 false)
store i32 %v, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @v_permlane16_b32_vss_fi(ptr addrspace(1) %out, i32 %src0, i32 %src1, i32 %src2) {
; SDAG: V_PERMLANE16_B32_e64 4, {{%[0-9]+}}, 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANE16_B32_e64 4, {{%[0-9]+}}, 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%v = call i32 @llvm.amdgcn.permlane16(i32 %src0, i32 %src0, i32 %src1, i32 %src2, i1 true, i1 false)
store i32 %v, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @v_permlane16_b32_vss_bc(ptr addrspace(1) %out, i32 %src0, i32 %src1, i32 %src2) {
; SDAG: V_PERMLANE16_B32_e64 0, {{%[0-9]+}}, 4, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANE16_B32_e64 0, {{%[0-9]+}}, 4, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%v = call i32 @llvm.amdgcn.permlane16(i32 %src0, i32 %src0, i32 %src1, i32 %src2, i1 false, i1 true)
store i32 %v, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @v_permlane16_b32_vss_fi_bc(ptr addrspace(1) %out, i32 %src0, i32 %src1, i32 %src2) {
; SDAG: V_PERMLANE16_B32_e64 4, {{%[0-9]+}}, 4, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANE16_B32_e64 4, {{%[0-9]+}}, 4, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%v = call i32 @llvm.amdgcn.permlane16(i32 %src0, i32 %src0, i32 %src1, i32 %src2, i1 true, i1 true)
store i32 %v, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @v_permlanex16_b32_vss(ptr addrspace(1) %out, i32 %src0, i32 %src1, i32 %src2) {
; SDAG: V_PERMLANEX16_B32_e64 0, {{%[0-9]+}}, 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANEX16_B32_e64 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%v = call i32 @llvm.amdgcn.permlanex16(i32 %src0, i32 %src0, i32 %src1, i32 %src2, i1 false, i1 false)
store i32 %v, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @v_permlanex16_b32_vii(ptr addrspace(1) %out, i32 %src0) {
; SDAG: V_PERMLANEX16_B32_e64 0, {{%[0-9]+}}, 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANEX16_B32_e64 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%v = call i32 @llvm.amdgcn.permlanex16(i32 %src0, i32 %src0, i32 1, i32 2, i1 false, i1 false)
store i32 %v, ptr addrspace(1) %out
ret void
}
; FIXME-GFX10PLUS: It is allowed to have both immediates as literals
define amdgpu_kernel void @v_permlanex16_b32_vll(ptr addrspace(1) %out, i32 %src0) {
; SDAG: V_PERMLANEX16_B32_e64 0, {{%[0-9]+}}, 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANEX16_B32_e64 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%v = call i32 @llvm.amdgcn.permlanex16(i32 %src0, i32 %src0, i32 4660, i32 49617, i1 false, i1 false)
store i32 %v, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @v_permlanex16_b32_vvv(ptr addrspace(1) %out, i32 %src0) {
; SDAG-GFX10: V_PERMLANEX16_B32_e64 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; SDAG-GFX11: V_PERMLANEX16_B32_e64 0, {{%[0-9]+}}, 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANEX16_B32_e64 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%tidx = call i32 @llvm.amdgcn.workitem.id.x()
%tidy = call i32 @llvm.amdgcn.workitem.id.y()
%v = call i32 @llvm.amdgcn.permlanex16(i32 %src0, i32 %src0, i32 %tidx, i32 %tidy, i1 false, i1 false)
store i32 %v, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @v_permlanex16_b32_vvs(ptr addrspace(1) %out, i32 %src0, i32 %src2) {
; SDAG-GFX10: V_PERMLANEX16_B32_e64 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; SDAG-GFX11: V_PERMLANEX16_B32_e64 0, {{%[0-9]+}}, 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANEX16_B32_e64 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%tidx = call i32 @llvm.amdgcn.workitem.id.x()
%v = call i32 @llvm.amdgcn.permlanex16(i32 %src0, i32 %src0, i32 %tidx, i32 %src2, i1 false, i1 false)
store i32 %v, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @v_permlanex16_b32_vsv(ptr addrspace(1) %out, i32 %src0, i32 %src1) {
; SDAG-GFX10: V_PERMLANEX16_B32_e64 0, {{%[0-9]+}}, 0, killed {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; SDAG-GFX11: V_PERMLANEX16_B32_e64 0, {{%[0-9]+}}, 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANEX16_B32_e64 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%tidy = call i32 @llvm.amdgcn.workitem.id.y()
%v = call i32 @llvm.amdgcn.permlanex16(i32 %src0, i32 %src0, i32 %src1, i32 %tidy, i1 false, i1 false)
store i32 %v, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @v_permlanex16_b32_vss_fi(ptr addrspace(1) %out, i32 %src0, i32 %src1, i32 %src2) {
; SDAG: V_PERMLANEX16_B32_e64 4, {{%[0-9]+}}, 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANEX16_B32_e64 4, {{%[0-9]+}}, 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%v = call i32 @llvm.amdgcn.permlanex16(i32 %src0, i32 %src0, i32 %src1, i32 %src2, i1 true, i1 false)
store i32 %v, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @v_permlanex16_b32_vss_bc(ptr addrspace(1) %out, i32 %src0, i32 %src1, i32 %src2) {
; SDAG: V_PERMLANEX16_B32_e64 0, {{%[0-9]+}}, 4, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANEX16_B32_e64 0, {{%[0-9]+}}, 4, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%v = call i32 @llvm.amdgcn.permlanex16(i32 %src0, i32 %src0, i32 %src1, i32 %src2, i1 false, i1 true)
store i32 %v, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @v_permlanex16_b32_vss_fi_bc(ptr addrspace(1) %out, i32 %src0, i32 %src1, i32 %src2) {
; SDAG: V_PERMLANEX16_B32_e64 4, {{%[0-9]+}}, 4, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANEX16_B32_e64 4, {{%[0-9]+}}, 4, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%v = call i32 @llvm.amdgcn.permlanex16(i32 %src0, i32 %src0, i32 %src1, i32 %src2, i1 true, i1 true)
store i32 %v, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @v_permlane16_b32_tid_tid(ptr addrspace(1) %out, i32 %src0, i32 %src1, i32 %src2) {
; SDAG-GFX10: V_PERMLANE16_B32_e64 0, {{%[0-9]+}}(s32), 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}(s32), 0, implicit $exec
; SDAG-GFX11: V_PERMLANE16_B32_e64 0, {{%[0-9]+}}, 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANE16_B32_e64 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%tidx = call i32 @llvm.amdgcn.workitem.id.x()
%v = call i32 @llvm.amdgcn.permlane16(i32 %tidx, i32 %tidx, i32 %src1, i32 %src2, i1 false, i1 false)
store i32 %v, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @v_permlane16_b32_undef_tid(ptr addrspace(1) %out, i32 %src0, i32 %src1, i32 %src2) {
; SDAG-GFX10: V_PERMLANE16_B32_e64 0, {{%[0-9]+}}(s32), 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; SDAG-GFX11: V_PERMLANE16_B32_e64 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANE16_B32_e64 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%tidx = call i32 @llvm.amdgcn.workitem.id.x()
%undef = freeze i32 poison
%v = call i32 @llvm.amdgcn.permlane16(i32 %undef, i32 %tidx, i32 %src1, i32 %src2, i1 false, i1 false)
store i32 %v, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @v_permlane16_b32_i_tid(ptr addrspace(1) %out, i32 %src0, i32 %src1, i32 %src2) {
; SDAG-GFX10: V_PERMLANE16_B32_e64 0, {{%[0-9]+}}(s32), 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; SDAG-GFX11: V_PERMLANE16_B32_e64 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANE16_B32_e64 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%tidx = call i32 @llvm.amdgcn.workitem.id.x()
%v = call i32 @llvm.amdgcn.permlane16(i32 12345, i32 %tidx, i32 %src1, i32 %src2, i1 false, i1 false)
store i32 %v, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @v_permlane16_b32_i_tid_fi(ptr addrspace(1) %out, i32 %src0, i32 %src1, i32 %src2) {
; SDAG-GFX10: V_PERMLANE16_B32_e64 4, {{%[0-9]+}}(s32), 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; SDAG-GFX11: V_PERMLANE16_B32_e64 4, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANE16_B32_e64 4, {{%[0-9]+}}, 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%tidx = call i32 @llvm.amdgcn.workitem.id.x()
%undef = freeze i32 poison
%v = call i32 @llvm.amdgcn.permlane16(i32 %undef, i32 %tidx, i32 %src1, i32 %src2, i1 true, i1 false)
store i32 %v, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @v_permlane16_b32_i_tid_bc(ptr addrspace(1) %out, i32 %src0, i32 %src1, i32 %src2) {
; SDAG-GFX10: V_PERMLANE16_B32_e64 0, {{%[0-9]+}}(s32), 4, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; SDAG-GFX11: V_PERMLANE16_B32_e64 0, killed {{%[0-9]+}}, 4, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANE16_B32_e64 0, {{%[0-9]+}}, 4, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%tidx = call i32 @llvm.amdgcn.workitem.id.x()
%undef = freeze i32 poison
%v = call i32 @llvm.amdgcn.permlane16(i32 %undef, i32 %tidx, i32 %src1, i32 %src2, i1 false, i1 true)
store i32 %v, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @v_permlane16_b32_i_tid_fi_bc(ptr addrspace(1) %out, i32 %src0, i32 %src1, i32 %src2) {
; SDAG-GFX10: V_PERMLANE16_B32_e64 4, {{%[0-9]+}}(s32), 4, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; SDAG-GFX11: V_PERMLANE16_B32_e64 4, killed {{%[0-9]+}}, 4, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANE16_B32_e64 4, {{%[0-9]+}}, 4, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%tidx = call i32 @llvm.amdgcn.workitem.id.x()
%undef = freeze i32 poison
%v = call i32 @llvm.amdgcn.permlane16(i32 %undef, i32 %tidx, i32 %src1, i32 %src2, i1 true, i1 true)
store i32 %v, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @v_permlanex16_b32_tid_tid(ptr addrspace(1) %out, i32 %src0, i32 %src1, i32 %src2) {
; SDAG-GFX10: V_PERMLANEX16_B32_e64 0, {{%[0-9]+}}(s32), 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}(s32), 0, implicit $exec
; SDAG-GFX11: V_PERMLANEX16_B32_e64 0, {{%[0-9]+}}, 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANEX16_B32_e64 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%tidx = call i32 @llvm.amdgcn.workitem.id.x()
%v = call i32 @llvm.amdgcn.permlanex16(i32 %tidx, i32 %tidx, i32 %src1, i32 %src2, i1 false, i1 false)
store i32 %v, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @v_permlanex16_b32_undef_tid(ptr addrspace(1) %out, i32 %src0, i32 %src1, i32 %src2) {
; SDAG-GFX10: V_PERMLANEX16_B32_e64 0, {{%[0-9]+}}(s32), 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; SDAG-GFX11: V_PERMLANEX16_B32_e64 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANEX16_B32_e64 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%tidx = call i32 @llvm.amdgcn.workitem.id.x()
%undef = freeze i32 poison
%v = call i32 @llvm.amdgcn.permlanex16(i32 %undef, i32 %tidx, i32 %src1, i32 %src2, i1 false, i1 false)
store i32 %v, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @v_permlanex16_b32_i_tid(ptr addrspace(1) %out, i32 %src0, i32 %src1, i32 %src2) {
; SDAG-GFX10: V_PERMLANEX16_B32_e64 0, {{%[0-9]+}}(s32), 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; SDAG-GFX11: V_PERMLANEX16_B32_e64 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANEX16_B32_e64 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%tidx = call i32 @llvm.amdgcn.workitem.id.x()
%v = call i32 @llvm.amdgcn.permlanex16(i32 12345, i32 %tidx, i32 %src1, i32 %src2, i1 false, i1 false)
store i32 %v, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @v_permlanex16_b32_i_tid_fi(ptr addrspace(1) %out, i32 %src0, i32 %src1, i32 %src2) {
; SDAG-GFX10: V_PERMLANEX16_B32_e64 4, {{%[0-9]+}}(s32), 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; SDAG-GFX11: V_PERMLANEX16_B32_e64 4, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANEX16_B32_e64 4, {{%[0-9]+}}, 0, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%tidx = call i32 @llvm.amdgcn.workitem.id.x()
%undef = freeze i32 poison
%v = call i32 @llvm.amdgcn.permlanex16(i32 %undef, i32 %tidx, i32 %src1, i32 %src2, i1 true, i1 false)
store i32 %v, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @v_permlanex16_b32_i_tid_bc(ptr addrspace(1) %out, i32 %src0, i32 %src1, i32 %src2) {
; SDAG-GFX10: V_PERMLANEX16_B32_e64 0, {{%[0-9]+}}(s32), 4, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; SDAG-GFX11: V_PERMLANEX16_B32_e64 0, killed {{%[0-9]+}}, 4, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANEX16_B32_e64 0, {{%[0-9]+}}, 4, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%tidx = call i32 @llvm.amdgcn.workitem.id.x()
%undef = freeze i32 poison
%v = call i32 @llvm.amdgcn.permlanex16(i32 %undef, i32 %tidx, i32 %src1, i32 %src2, i1 false, i1 true)
store i32 %v, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @v_permlanex16_b32_i_tid_fi_bc(ptr addrspace(1) %out, i32 %src0, i32 %src1, i32 %src2) {
; SDAG-GFX10: V_PERMLANEX16_B32_e64 4, {{%[0-9]+}}(s32), 4, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; SDAG-GFX11: V_PERMLANEX16_B32_e64 4, killed {{%[0-9]+}}, 4, killed {{%[0-9]+}}, 0, killed {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
; GISEL: V_PERMLANEX16_B32_e64 4, {{%[0-9]+}}, 4, {{%[0-9]+}}, 0, {{%[0-9]+}}, {{%[0-9]+}}, 0, implicit $exec
%tidx = call i32 @llvm.amdgcn.workitem.id.x()
%undef = freeze i32 poison
%v = call i32 @llvm.amdgcn.permlanex16(i32 %undef, i32 %tidx, i32 %src1, i32 %src2, i1 true, i1 true)
store i32 %v, ptr addrspace(1) %out
ret void
}
Reference in new issue View Git Blame Copy Permalink