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
233 lines
14 KiB
LLVM
233 lines
14 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: not --crash llc -global-isel=0 -mtriple=amdgcn -mcpu=gfx1100 < %s 2>&1 | FileCheck -check-prefix=ERR %s
|
|
; RUN: not llc -global-isel=1 -new-reg-bank-select -mtriple=amdgcn -mcpu=gfx1100 < %s 2>&1 | FileCheck -check-prefix=ERR %s
|
|
; RUN: llc -global-isel=0 -mtriple=amdgcn -mcpu=gfx1200 < %s | FileCheck -check-prefixes=GFX12-SDAG %s
|
|
; RUN: llc -global-isel=1 -new-reg-bank-select -mtriple=amdgcn -mcpu=gfx1200 < %s | FileCheck -check-prefixes=GFX12-GISEL %s
|
|
|
|
declare {<10 x i32>, <3 x float>, <3 x float>} @llvm.amdgcn.image.bvh.dual.intersect.ray(i64, float, i8, <3 x float>, <3 x float>, <2 x i32>, <4 x i32>)
|
|
|
|
; ERR: in function image_bvh_dual_intersect_ray{{.*}}intrinsic not supported on subtarget
|
|
define amdgpu_ps <10 x float> @image_bvh_dual_intersect_ray(i64 %node_ptr, float %ray_extent, float %ray_origin_x, float %ray_origin_y, float %ray_origin_z, float %ray_dir_x, float %ray_dir_y, float %ray_dir_z, <2 x i32> %offsets, <4 x i32> inreg %tdescr, ptr addrspace(1) %origin, ptr addrspace(1) %dir) {
|
|
; GFX12-SDAG-LABEL: image_bvh_dual_intersect_ray:
|
|
; GFX12-SDAG: ; %bb.0: ; %main_body
|
|
; GFX12-SDAG-NEXT: v_dual_mov_b32 v22, v8 :: v_dual_mov_b32 v21, v7
|
|
; GFX12-SDAG-NEXT: v_dual_mov_b32 v20, v6 :: v_dual_mov_b32 v19, v5
|
|
; GFX12-SDAG-NEXT: v_dual_mov_b32 v18, v4 :: v_dual_mov_b32 v17, v3
|
|
; GFX12-SDAG-NEXT: v_mov_b32_e32 v3, 0
|
|
; GFX12-SDAG-NEXT: image_bvh_dual_intersect_ray v[0:9], [v[0:1], v[2:3], v[17:19], v[20:22], v[9:10]], s[0:3]
|
|
; GFX12-SDAG-NEXT: s_wait_bvhcnt 0x0
|
|
; GFX12-SDAG-NEXT: global_store_b96 v[11:12], v[17:19], off
|
|
; GFX12-SDAG-NEXT: global_store_b96 v[13:14], v[20:22], off
|
|
; GFX12-SDAG-NEXT: ; return to shader part epilog
|
|
;
|
|
; GFX12-GISEL-LABEL: image_bvh_dual_intersect_ray:
|
|
; GFX12-GISEL: ; %bb.0: ; %main_body
|
|
; GFX12-GISEL-NEXT: v_dual_mov_b32 v15, v3 :: v_dual_mov_b32 v16, v4
|
|
; GFX12-GISEL-NEXT: v_dual_mov_b32 v17, v5 :: v_dual_mov_b32 v18, v6
|
|
; GFX12-GISEL-NEXT: v_dual_mov_b32 v19, v7 :: v_dual_mov_b32 v20, v8
|
|
; GFX12-GISEL-NEXT: v_mov_b32_e32 v3, 0
|
|
; GFX12-GISEL-NEXT: image_bvh_dual_intersect_ray v[0:9], [v[0:1], v[2:3], v[15:17], v[18:20], v[9:10]], s[0:3]
|
|
; GFX12-GISEL-NEXT: s_wait_bvhcnt 0x0
|
|
; GFX12-GISEL-NEXT: global_store_b96 v[11:12], v[15:17], off
|
|
; GFX12-GISEL-NEXT: global_store_b96 v[13:14], v[18:20], off
|
|
; GFX12-GISEL-NEXT: ; return to shader part epilog
|
|
main_body:
|
|
%ray_origin0 = insertelement <3 x float> poison, float %ray_origin_x, i32 0
|
|
%ray_origin1 = insertelement <3 x float> %ray_origin0, float %ray_origin_y, i32 1
|
|
%ray_origin = insertelement <3 x float> %ray_origin1, float %ray_origin_z, i32 2
|
|
%ray_dir0 = insertelement <3 x float> poison, float %ray_dir_x, i32 0
|
|
%ray_dir1 = insertelement <3 x float> %ray_dir0, float %ray_dir_y, i32 1
|
|
%ray_dir = insertelement <3 x float> %ray_dir1, float %ray_dir_z, i32 2
|
|
%v = call {<10 x i32>, <3 x float>, <3 x float>} @llvm.amdgcn.image.bvh.dual.intersect.ray(i64 %node_ptr, float %ray_extent, i8 0, <3 x float> %ray_origin, <3 x float> %ray_dir, <2 x i32> %offsets, <4 x i32> %tdescr)
|
|
%a = extractvalue {<10 x i32>, <3 x float>, <3 x float>} %v, 0
|
|
%r = bitcast <10 x i32> %a to <10 x float>
|
|
%o = extractvalue {<10 x i32>, <3 x float>, <3 x float>} %v, 1
|
|
store <3 x float> %o, ptr addrspace(1) %origin
|
|
%d = extractvalue {<10 x i32>, <3 x float>, <3 x float>} %v, 2
|
|
store <3 x float> %d, ptr addrspace(1) %dir
|
|
ret <10 x float> %r
|
|
}
|
|
|
|
define amdgpu_ps <10 x float> @image_bvh_dual_intersect_ray_1(i64 %node_ptr, float %ray_extent, float %ray_origin_x, float %ray_origin_y, float %ray_origin_z, float %ray_dir_x, float %ray_dir_y, float %ray_dir_z, <2 x i32> %offsets, <4 x i32> inreg %tdescr, ptr addrspace(1) %origin, ptr addrspace(1) %dir) {
|
|
; GFX12-SDAG-LABEL: image_bvh_dual_intersect_ray_1:
|
|
; GFX12-SDAG: ; %bb.0: ; %main_body
|
|
; GFX12-SDAG-NEXT: v_dual_mov_b32 v22, v8 :: v_dual_mov_b32 v21, v7
|
|
; GFX12-SDAG-NEXT: v_dual_mov_b32 v20, v6 :: v_dual_mov_b32 v19, v5
|
|
; GFX12-SDAG-NEXT: v_dual_mov_b32 v18, v4 :: v_dual_mov_b32 v17, v3
|
|
; GFX12-SDAG-NEXT: v_mov_b32_e32 v3, 1
|
|
; GFX12-SDAG-NEXT: image_bvh_dual_intersect_ray v[0:9], [v[0:1], v[2:3], v[17:19], v[20:22], v[9:10]], s[0:3]
|
|
; GFX12-SDAG-NEXT: s_wait_bvhcnt 0x0
|
|
; GFX12-SDAG-NEXT: global_store_b96 v[11:12], v[17:19], off
|
|
; GFX12-SDAG-NEXT: global_store_b96 v[13:14], v[20:22], off
|
|
; GFX12-SDAG-NEXT: ; return to shader part epilog
|
|
;
|
|
; GFX12-GISEL-LABEL: image_bvh_dual_intersect_ray_1:
|
|
; GFX12-GISEL: ; %bb.0: ; %main_body
|
|
; GFX12-GISEL-NEXT: v_dual_mov_b32 v15, v3 :: v_dual_mov_b32 v16, v4
|
|
; GFX12-GISEL-NEXT: v_dual_mov_b32 v17, v5 :: v_dual_mov_b32 v18, v6
|
|
; GFX12-GISEL-NEXT: v_dual_mov_b32 v19, v7 :: v_dual_mov_b32 v20, v8
|
|
; GFX12-GISEL-NEXT: v_mov_b32_e32 v3, 1
|
|
; GFX12-GISEL-NEXT: image_bvh_dual_intersect_ray v[0:9], [v[0:1], v[2:3], v[15:17], v[18:20], v[9:10]], s[0:3]
|
|
; GFX12-GISEL-NEXT: s_wait_bvhcnt 0x0
|
|
; GFX12-GISEL-NEXT: global_store_b96 v[11:12], v[15:17], off
|
|
; GFX12-GISEL-NEXT: global_store_b96 v[13:14], v[18:20], off
|
|
; GFX12-GISEL-NEXT: ; return to shader part epilog
|
|
main_body:
|
|
%ray_origin0 = insertelement <3 x float> poison, float %ray_origin_x, i32 0
|
|
%ray_origin1 = insertelement <3 x float> %ray_origin0, float %ray_origin_y, i32 1
|
|
%ray_origin = insertelement <3 x float> %ray_origin1, float %ray_origin_z, i32 2
|
|
%ray_dir0 = insertelement <3 x float> poison, float %ray_dir_x, i32 0
|
|
%ray_dir1 = insertelement <3 x float> %ray_dir0, float %ray_dir_y, i32 1
|
|
%ray_dir = insertelement <3 x float> %ray_dir1, float %ray_dir_z, i32 2
|
|
%v = call {<10 x i32>, <3 x float>, <3 x float>} @llvm.amdgcn.image.bvh.dual.intersect.ray(i64 %node_ptr, float %ray_extent, i8 1, <3 x float> %ray_origin, <3 x float> %ray_dir, <2 x i32> %offsets, <4 x i32> %tdescr)
|
|
%a = extractvalue {<10 x i32>, <3 x float>, <3 x float>} %v, 0
|
|
%r = bitcast <10 x i32> %a to <10 x float>
|
|
%o = extractvalue {<10 x i32>, <3 x float>, <3 x float>} %v, 1
|
|
store <3 x float> %o, ptr addrspace(1) %origin
|
|
%d = extractvalue {<10 x i32>, <3 x float>, <3 x float>} %v, 2
|
|
store <3 x float> %d, ptr addrspace(1) %dir
|
|
ret <10 x float> %r
|
|
}
|
|
|
|
define amdgpu_ps <10 x float> @image_bvh_dual_intersect_ray_ssssss(i64 inreg %node_ptr, float inreg %ray_extent, <3 x float> inreg %ray_origin, <3 x float> inreg %ray_dir, <2 x i32> inreg %offsets, <4 x i32> inreg %tdescr, ptr addrspace(1) %origin, ptr addrspace(1) %dir) {
|
|
; GFX12-SDAG-LABEL: image_bvh_dual_intersect_ray_ssssss:
|
|
; GFX12-SDAG: ; %bb.0:
|
|
; GFX12-SDAG-NEXT: v_dual_mov_b32 v11, v3 :: v_dual_mov_b32 v10, v2
|
|
; GFX12-SDAG-NEXT: v_dual_mov_b32 v13, v1 :: v_dual_mov_b32 v12, v0
|
|
; GFX12-SDAG-NEXT: v_dual_mov_b32 v14, s6 :: v_dual_mov_b32 v15, s7
|
|
; GFX12-SDAG-NEXT: v_dual_mov_b32 v16, s8 :: v_dual_mov_b32 v17, s3
|
|
; GFX12-SDAG-NEXT: v_dual_mov_b32 v18, s4 :: v_dual_mov_b32 v19, s5
|
|
; GFX12-SDAG-NEXT: v_dual_mov_b32 v0, s9 :: v_dual_mov_b32 v1, s10
|
|
; GFX12-SDAG-NEXT: v_dual_mov_b32 v2, s0 :: v_dual_mov_b32 v3, s1
|
|
; GFX12-SDAG-NEXT: v_dual_mov_b32 v4, s2 :: v_dual_mov_b32 v5, 0
|
|
; GFX12-SDAG-NEXT: s_mov_b32 s15, s14
|
|
; GFX12-SDAG-NEXT: s_mov_b32 s14, s13
|
|
; GFX12-SDAG-NEXT: s_mov_b32 s13, s12
|
|
; GFX12-SDAG-NEXT: s_mov_b32 s12, s11
|
|
; GFX12-SDAG-NEXT: image_bvh_dual_intersect_ray v[0:9], [v[2:3], v[4:5], v[17:19], v[14:16], v[0:1]], s[12:15]
|
|
; GFX12-SDAG-NEXT: s_wait_bvhcnt 0x0
|
|
; GFX12-SDAG-NEXT: global_store_b96 v[12:13], v[17:19], off
|
|
; GFX12-SDAG-NEXT: global_store_b96 v[10:11], v[14:16], off
|
|
; GFX12-SDAG-NEXT: ; return to shader part epilog
|
|
;
|
|
; GFX12-GISEL-LABEL: image_bvh_dual_intersect_ray_ssssss:
|
|
; GFX12-GISEL: ; %bb.0:
|
|
; GFX12-GISEL-NEXT: s_mov_b32 s16, s3
|
|
; GFX12-GISEL-NEXT: s_mov_b32 s18, s5
|
|
; GFX12-GISEL-NEXT: s_mov_b32 s20, s6
|
|
; GFX12-GISEL-NEXT: s_mov_b32 s17, s4
|
|
; GFX12-GISEL-NEXT: s_mov_b32 s21, s7
|
|
; GFX12-GISEL-NEXT: s_mov_b32 s22, s8
|
|
; GFX12-GISEL-NEXT: s_mov_b32 s8, s9
|
|
; GFX12-GISEL-NEXT: s_mov_b32 s9, s10
|
|
; GFX12-GISEL-NEXT: s_mov_b32 s3, 0
|
|
; GFX12-GISEL-NEXT: v_dual_mov_b32 v14, s16 :: v_dual_mov_b32 v15, s17
|
|
; GFX12-GISEL-NEXT: v_dual_mov_b32 v16, s18 :: v_dual_mov_b32 v17, s20
|
|
; GFX12-GISEL-NEXT: v_dual_mov_b32 v10, v0 :: v_dual_mov_b32 v11, v1
|
|
; GFX12-GISEL-NEXT: v_dual_mov_b32 v12, v2 :: v_dual_mov_b32 v13, v3
|
|
; GFX12-GISEL-NEXT: v_dual_mov_b32 v0, s0 :: v_dual_mov_b32 v1, s1
|
|
; GFX12-GISEL-NEXT: v_dual_mov_b32 v2, s2 :: v_dual_mov_b32 v3, s3
|
|
; GFX12-GISEL-NEXT: v_dual_mov_b32 v18, s21 :: v_dual_mov_b32 v19, s22
|
|
; GFX12-GISEL-NEXT: v_dual_mov_b32 v4, s8 :: v_dual_mov_b32 v5, s9
|
|
; GFX12-GISEL-NEXT: s_mov_b32 s4, s11
|
|
; GFX12-GISEL-NEXT: s_mov_b32 s5, s12
|
|
; GFX12-GISEL-NEXT: s_mov_b32 s6, s13
|
|
; GFX12-GISEL-NEXT: s_mov_b32 s7, s14
|
|
; GFX12-GISEL-NEXT: image_bvh_dual_intersect_ray v[0:9], [v[0:1], v[2:3], v[14:16], v[17:19], v[4:5]], s[4:7]
|
|
; GFX12-GISEL-NEXT: s_wait_bvhcnt 0x0
|
|
; GFX12-GISEL-NEXT: global_store_b96 v[10:11], v[14:16], off
|
|
; GFX12-GISEL-NEXT: global_store_b96 v[12:13], v[17:19], off
|
|
; GFX12-GISEL-NEXT: ; return to shader part epilog
|
|
%v = call {<10 x i32>, <3 x float>, <3 x float>} @llvm.amdgcn.image.bvh.dual.intersect.ray(i64 %node_ptr, float %ray_extent, i8 0, <3 x float> %ray_origin, <3 x float> %ray_dir, <2 x i32> %offsets, <4 x i32> %tdescr)
|
|
%a = extractvalue {<10 x i32>, <3 x float>, <3 x float>} %v, 0
|
|
%r = bitcast <10 x i32> %a to <10 x float>
|
|
%o = extractvalue {<10 x i32>, <3 x float>, <3 x float>} %v, 1
|
|
store <3 x float> %o, ptr addrspace(1) %origin
|
|
%d = extractvalue {<10 x i32>, <3 x float>, <3 x float>} %v, 2
|
|
store <3 x float> %d, ptr addrspace(1) %dir
|
|
ret <10 x float> %r
|
|
}
|
|
|
|
define amdgpu_ps <10 x float> @image_bvh_dual_intersect_ray_vvvvvv(i64 %node_ptr, float %ray_extent, <3 x float> %ray_origin, <3 x float> %ray_dir, <2 x i32> %offsets, <4 x i32> %tdescr, ptr addrspace(1) %origin, ptr addrspace(1) %dir) {
|
|
; GFX12-SDAG-LABEL: image_bvh_dual_intersect_ray_vvvvvv:
|
|
; GFX12-SDAG: ; %bb.0:
|
|
; GFX12-SDAG-NEXT: v_dual_mov_b32 v26, v10 :: v_dual_mov_b32 v25, v9
|
|
; GFX12-SDAG-NEXT: v_dual_mov_b32 v21, v8 :: v_dual_mov_b32 v20, v7
|
|
; GFX12-SDAG-NEXT: v_dual_mov_b32 v19, v6 :: v_dual_mov_b32 v24, v5
|
|
; GFX12-SDAG-NEXT: v_dual_mov_b32 v23, v4 :: v_dual_mov_b32 v22, v3
|
|
; GFX12-SDAG-NEXT: v_dual_mov_b32 v29, v2 :: v_dual_mov_b32 v28, v1
|
|
; GFX12-SDAG-NEXT: v_dual_mov_b32 v27, v0 :: v_dual_mov_b32 v30, 0
|
|
; GFX12-SDAG-NEXT: s_mov_b32 s1, exec_lo
|
|
; GFX12-SDAG-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
|
|
; GFX12-SDAG-NEXT: v_readfirstlane_b32 s4, v11
|
|
; GFX12-SDAG-NEXT: v_readfirstlane_b32 s5, v12
|
|
; GFX12-SDAG-NEXT: v_readfirstlane_b32 s6, v13
|
|
; GFX12-SDAG-NEXT: v_readfirstlane_b32 s7, v14
|
|
; GFX12-SDAG-NEXT: s_wait_alu depctr_va_sdst(0)
|
|
; GFX12-SDAG-NEXT: s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_2)
|
|
; GFX12-SDAG-NEXT: v_cmp_eq_u64_e32 vcc_lo, s[4:5], v[11:12]
|
|
; GFX12-SDAG-NEXT: v_cmp_eq_u64_e64 s0, s[6:7], v[13:14]
|
|
; GFX12-SDAG-NEXT: s_and_b32 s0, vcc_lo, s0
|
|
; GFX12-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
|
|
; GFX12-SDAG-NEXT: s_and_saveexec_b32 s0, s0
|
|
; GFX12-SDAG-NEXT: s_wait_bvhcnt 0x0
|
|
; GFX12-SDAG-NEXT: image_bvh_dual_intersect_ray v[0:9], [v[27:28], v[29:30], v[22:24], v[19:21], v[25:26]], s[4:7]
|
|
; GFX12-SDAG-NEXT: ; implicit-def: $vgpr11_vgpr12_vgpr13_vgpr14
|
|
; GFX12-SDAG-NEXT: ; implicit-def: $vgpr27_vgpr28
|
|
; GFX12-SDAG-NEXT: ; implicit-def: $vgpr29_vgpr30
|
|
; GFX12-SDAG-NEXT: ; implicit-def: $vgpr25_vgpr26
|
|
; GFX12-SDAG-NEXT: s_xor_b32 exec_lo, exec_lo, s0
|
|
; GFX12-SDAG-NEXT: s_cbranch_execnz .LBB3_1
|
|
; GFX12-SDAG-NEXT: ; %bb.2:
|
|
; GFX12-SDAG-NEXT: s_mov_b32 exec_lo, s1
|
|
; GFX12-SDAG-NEXT: s_wait_bvhcnt 0x0
|
|
; GFX12-SDAG-NEXT: global_store_b96 v[15:16], v[22:24], off
|
|
; GFX12-SDAG-NEXT: global_store_b96 v[17:18], v[19:21], off
|
|
; GFX12-SDAG-NEXT: ; return to shader part epilog
|
|
;
|
|
; GFX12-GISEL-LABEL: image_bvh_dual_intersect_ray_vvvvvv:
|
|
; GFX12-GISEL: ; %bb.0:
|
|
; GFX12-GISEL-NEXT: v_dual_mov_b32 v25, v0 :: v_dual_mov_b32 v26, v1
|
|
; GFX12-GISEL-NEXT: v_dual_mov_b32 v27, v2 :: v_dual_mov_b32 v20, v4
|
|
; GFX12-GISEL-NEXT: v_dual_mov_b32 v19, v3 :: v_dual_mov_b32 v22, v6
|
|
; GFX12-GISEL-NEXT: v_dual_mov_b32 v21, v5 :: v_dual_mov_b32 v24, v8
|
|
; GFX12-GISEL-NEXT: v_dual_mov_b32 v23, v7 :: v_dual_mov_b32 v30, v10
|
|
; GFX12-GISEL-NEXT: v_dual_mov_b32 v29, v9 :: v_dual_mov_b32 v28, 0
|
|
; GFX12-GISEL-NEXT: s_mov_b32 s1, exec_lo
|
|
; GFX12-GISEL-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
|
|
; GFX12-GISEL-NEXT: v_readfirstlane_b32 s4, v11
|
|
; GFX12-GISEL-NEXT: v_readfirstlane_b32 s5, v12
|
|
; GFX12-GISEL-NEXT: v_readfirstlane_b32 s6, v13
|
|
; GFX12-GISEL-NEXT: v_readfirstlane_b32 s7, v14
|
|
; GFX12-GISEL-NEXT: s_wait_alu depctr_va_sdst(0)
|
|
; GFX12-GISEL-NEXT: s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_2)
|
|
; GFX12-GISEL-NEXT: v_cmp_eq_u64_e32 vcc_lo, s[4:5], v[11:12]
|
|
; GFX12-GISEL-NEXT: v_cmp_eq_u64_e64 s0, s[6:7], v[13:14]
|
|
; GFX12-GISEL-NEXT: s_and_b32 s0, vcc_lo, s0
|
|
; GFX12-GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
|
|
; GFX12-GISEL-NEXT: s_and_saveexec_b32 s0, s0
|
|
; GFX12-GISEL-NEXT: s_wait_bvhcnt 0x0
|
|
; GFX12-GISEL-NEXT: image_bvh_dual_intersect_ray v[0:9], [v[25:26], v[27:28], v[19:21], v[22:24], v[29:30]], s[4:7]
|
|
; GFX12-GISEL-NEXT: ; implicit-def: $vgpr11
|
|
; GFX12-GISEL-NEXT: ; implicit-def: $vgpr25_vgpr26
|
|
; GFX12-GISEL-NEXT: ; implicit-def: $vgpr27_vgpr28
|
|
; GFX12-GISEL-NEXT: ; implicit-def: $vgpr29_vgpr30
|
|
; GFX12-GISEL-NEXT: ; implicit-def: $vgpr13_vgpr14
|
|
; GFX12-GISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s0
|
|
; GFX12-GISEL-NEXT: s_cbranch_execnz .LBB3_1
|
|
; GFX12-GISEL-NEXT: ; %bb.2:
|
|
; GFX12-GISEL-NEXT: s_mov_b32 exec_lo, s1
|
|
; GFX12-GISEL-NEXT: s_wait_bvhcnt 0x0
|
|
; GFX12-GISEL-NEXT: global_store_b96 v[15:16], v[19:21], off
|
|
; GFX12-GISEL-NEXT: global_store_b96 v[17:18], v[22:24], off
|
|
; GFX12-GISEL-NEXT: ; return to shader part epilog
|
|
%v = call {<10 x i32>, <3 x float>, <3 x float>} @llvm.amdgcn.image.bvh.dual.intersect.ray(i64 %node_ptr, float %ray_extent, i8 0, <3 x float> %ray_origin, <3 x float> %ray_dir, <2 x i32> %offsets, <4 x i32> %tdescr)
|
|
%a = extractvalue {<10 x i32>, <3 x float>, <3 x float>} %v, 0
|
|
%r = bitcast <10 x i32> %a to <10 x float>
|
|
%o = extractvalue {<10 x i32>, <3 x float>, <3 x float>} %v, 1
|
|
store <3 x float> %o, ptr addrspace(1) %origin
|
|
%d = extractvalue {<10 x i32>, <3 x float>, <3 x float>} %v, 2
|
|
store <3 x float> %d, ptr addrspace(1) %dir
|
|
ret <10 x float> %r
|
|
}
|