Files
TheRock/compiler/amd-llvm/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.s.prefetch.data.ll
T
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
Flatten subprojects into a single base monorepo
2026-09-02 16:34:47 -07:00

164 lines
6.2 KiB
LLVM

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
; RUN: llc -global-isel=0 -mtriple=amdgcn -mcpu=gfx1200 < %s | FileCheck --check-prefixes=GCN,SDAG %s
; RUN: llc -global-isel=1 -new-reg-bank-select -mtriple=amdgcn -mcpu=gfx1200 < %s | FileCheck --check-prefixes=GCN,GISEL %s
define amdgpu_ps void @prefetch_data_sgpr_base_sgpr_len(ptr addrspace(4) inreg %ptr, i32 inreg %len) {
; GCN-LABEL: prefetch_data_sgpr_base_sgpr_len:
; GCN: ; %bb.0: ; %entry
; GCN-NEXT: s_prefetch_data s[0:1], 0x0, s2, 0
; GCN-NEXT: s_endpgm
entry:
tail call void @llvm.amdgcn.s.prefetch.data.p4(ptr addrspace(4) %ptr, i32 %len)
ret void
}
define amdgpu_ps void @prefetch_data_sgpr_imm_base_sgpr_len(ptr addrspace(4) inreg %ptr, i32 inreg %len) {
; GCN-LABEL: prefetch_data_sgpr_imm_base_sgpr_len:
; GCN: ; %bb.0: ; %entry
; GCN-NEXT: s_prefetch_data s[0:1], 0x200, s2, 0
; GCN-NEXT: s_endpgm
entry:
%gep = getelementptr i32, ptr addrspace(4) %ptr, i32 128
tail call void @llvm.amdgcn.s.prefetch.data.p4(ptr addrspace(4) %gep, i32 %len)
ret void
}
define amdgpu_ps void @prefetch_data_sgpr_base_imm_len(ptr addrspace(4) inreg %ptr) {
; GCN-LABEL: prefetch_data_sgpr_base_imm_len:
; GCN: ; %bb.0: ; %entry
; GCN-NEXT: s_prefetch_data s[0:1], 0x0, null, 31
; GCN-NEXT: s_endpgm
entry:
tail call void @llvm.amdgcn.s.prefetch.data.p4(ptr addrspace(4) %ptr, i32 31)
ret void
}
define amdgpu_ps void @prefetch_data_sgpr_imm_base_imm_len(ptr addrspace(4) inreg %ptr) {
; GCN-LABEL: prefetch_data_sgpr_imm_base_imm_len:
; GCN: ; %bb.0: ; %entry
; GCN-NEXT: s_prefetch_data s[0:1], 0x200, null, 31
; GCN-NEXT: s_endpgm
entry:
%gep = getelementptr i32, ptr addrspace(4) %ptr, i32 128
tail call void @llvm.amdgcn.s.prefetch.data.p4(ptr addrspace(4) %gep, i32 31)
ret void
}
define amdgpu_ps void @prefetch_data_vgpr_base_sgpr_len(ptr addrspace(4) %ptr, i32 inreg %len) {
; GCN-LABEL: prefetch_data_vgpr_base_sgpr_len:
; GCN: ; %bb.0: ; %entry
; GCN-NEXT: v_readfirstlane_b32 s2, v0
; GCN-NEXT: v_readfirstlane_b32 s3, v1
; GCN-NEXT: s_prefetch_data s[2:3], 0x0, s0, 0
; GCN-NEXT: s_endpgm
entry:
tail call void @llvm.amdgcn.s.prefetch.data.p4(ptr addrspace(4) %ptr, i32 %len)
ret void
}
define amdgpu_ps void @prefetch_data_vgpr_imm_base_sgpr_len(ptr addrspace(4) %ptr, i32 inreg %len) {
; SDAG-LABEL: prefetch_data_vgpr_imm_base_sgpr_len:
; SDAG: ; %bb.0: ; %entry
; SDAG-NEXT: v_readfirstlane_b32 s2, v0
; SDAG-NEXT: v_readfirstlane_b32 s3, v1
; SDAG-NEXT: s_prefetch_data s[2:3], 0x200, s0, 0
; SDAG-NEXT: s_endpgm
;
; GISEL-LABEL: prefetch_data_vgpr_imm_base_sgpr_len:
; GISEL: ; %bb.0: ; %entry
; GISEL-NEXT: v_add_co_u32 v0, vcc_lo, 0x200, v0
; GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2)
; GISEL-NEXT: v_add_co_ci_u32_e64 v1, null, 0, v1, vcc_lo
; GISEL-NEXT: v_readfirstlane_b32 s2, v0
; GISEL-NEXT: s_delay_alu instid0(VALU_DEP_2)
; GISEL-NEXT: v_readfirstlane_b32 s3, v1
; GISEL-NEXT: s_prefetch_data s[2:3], 0x0, s0, 0
; GISEL-NEXT: s_endpgm
entry:
%gep = getelementptr i32, ptr addrspace(4) %ptr, i32 128
tail call void @llvm.amdgcn.s.prefetch.data.p4(ptr addrspace(4) %gep, i32 %len)
ret void
}
define amdgpu_ps void @prefetch_data_sgpr_base_vgpr_len(ptr addrspace(4) inreg %ptr, i32 %len) {
; GCN-LABEL: prefetch_data_sgpr_base_vgpr_len:
; GCN: ; %bb.0: ; %entry
; GCN-NEXT: v_readfirstlane_b32 s2, v0
; GCN-NEXT: s_prefetch_data s[0:1], 0x0, s2, 0
; GCN-NEXT: s_endpgm
entry:
tail call void @llvm.amdgcn.s.prefetch.data.p4(ptr addrspace(4) %ptr, i32 %len)
ret void
}
define amdgpu_ps void @prefetch_data_sgpr_base_imm_len_global(ptr addrspace(1) inreg %ptr) {
; GCN-LABEL: prefetch_data_sgpr_base_imm_len_global:
; GCN: ; %bb.0: ; %entry
; GCN-NEXT: s_prefetch_data s[0:1], 0x0, null, 31
; GCN-NEXT: s_endpgm
entry:
tail call void @llvm.amdgcn.s.prefetch.data.p1(ptr addrspace(1) %ptr, i32 31)
ret void
}
define amdgpu_ps void @prefetch_data_sgpr_base_imm_len_flat(ptr inreg %ptr) {
; GCN-LABEL: prefetch_data_sgpr_base_imm_len_flat:
; GCN: ; %bb.0: ; %entry
; GCN-NEXT: s_prefetch_data s[0:1], 0x0, null, 31
; GCN-NEXT: s_endpgm
entry:
tail call void @llvm.amdgcn.s.prefetch.data.p0(ptr %ptr, i32 31)
ret void
}
define amdgpu_ps void @prefetch_data_vgpr_base_imm_len(ptr addrspace(4) %ptr) {
; GCN-LABEL: prefetch_data_vgpr_base_imm_len:
; GCN: ; %bb.0: ; %entry
; GCN-NEXT: v_readfirstlane_b32 s0, v0
; GCN-NEXT: v_readfirstlane_b32 s1, v1
; GCN-NEXT: s_prefetch_data s[0:1], 0x0, null, 0
; GCN-NEXT: s_endpgm
entry:
tail call void @llvm.amdgcn.s.prefetch.data.p4(ptr addrspace(4) %ptr, i32 0)
ret void
}
define amdgpu_ps void @prefetch_data_vgpr_base_vgpr_len(ptr addrspace(4) %ptr, i32 %len) {
; SDAG-LABEL: prefetch_data_vgpr_base_vgpr_len:
; SDAG: ; %bb.0: ; %entry
; SDAG-NEXT: v_readfirstlane_b32 s2, v2
; SDAG-NEXT: v_readfirstlane_b32 s0, v0
; SDAG-NEXT: v_readfirstlane_b32 s1, v1
; SDAG-NEXT: s_prefetch_data s[0:1], 0x0, s2, 0
; SDAG-NEXT: s_endpgm
;
; GISEL-LABEL: prefetch_data_vgpr_base_vgpr_len:
; GISEL: ; %bb.0: ; %entry
; GISEL-NEXT: v_readfirstlane_b32 s0, v0
; GISEL-NEXT: v_readfirstlane_b32 s1, v1
; GISEL-NEXT: v_readfirstlane_b32 s2, v2
; GISEL-NEXT: s_prefetch_data s[0:1], 0x0, s2, 0
; GISEL-NEXT: s_endpgm
entry:
tail call void @llvm.amdgcn.s.prefetch.data.p4(ptr addrspace(4) %ptr, i32 %len)
ret void
}
define amdgpu_ps float @prefetch_and_load_b32(ptr addrspace(4) align 4 inreg %p) {
; GCN-LABEL: prefetch_and_load_b32:
; GCN: ; %bb.0: ; %entry
; GCN-NEXT: s_prefetch_data s[0:1], 0x0, null, 0
; GCN-NEXT: s_load_b32 s0, s[0:1], 0x0
; GCN-NEXT: s_wait_kmcnt 0x0
; GCN-NEXT: v_mov_b32_e32 v0, s0
; GCN-NEXT: ; return to shader part epilog
entry:
tail call void @llvm.amdgcn.s.prefetch.data.p4(ptr addrspace(4) %p, i32 0)
%ret = load float, ptr addrspace(4) %p, align 4
ret float %ret
}
declare void @llvm.amdgcn.s.prefetch.data.p4(ptr addrspace(4) %ptr, i32 %len)
declare void @llvm.amdgcn.s.prefetch.data.p1(ptr addrspace(1) %ptr, i32 %len)
declare void @llvm.amdgcn.s.prefetch.data.p0(ptr %ptr, i32 %len)