derek/TheRock
1
0 Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity
Files
1a182e2dac58879f215c0fad0f5f78c76dcafe56
TheRock/compiler/amd-llvm/llvm/test/CodeGen/AMDGPU/fabs-known-signbit-combine-fast-fdiv-lowering.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

320 lines
13 KiB
LLVM
Raw Blame History

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx900 < %s | FileCheck %s
; The fast fdiv path for 2.5 ulp fdiv with daz emits a fabs on the
; RHS, which should be droppable if we know the sign bit of the RHS is
; already 0 (or nan).
define float @fdiv_fast_daz_rhs_signbit_known_zero_fabs(float %x, float %y) #1 {
; CHECK-LABEL: fdiv_fast_daz_rhs_signbit_known_zero_fabs:
; CHECK: ; %bb.0: ; %entry
; CHECK-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; CHECK-NEXT: s_mov_b32 s4, 0x6f800000
; CHECK-NEXT: v_mov_b32_e32 v2, 0x2f800000
; CHECK-NEXT: v_cmp_gt_f32_e64 vcc, |v1|, s4
; CHECK-NEXT: v_cndmask_b32_e32 v2, 1.0, v2, vcc
; CHECK-NEXT: v_mul_f32_e64 v1, |v1|, v2
; CHECK-NEXT: v_rcp_f32_e32 v1, v1
; CHECK-NEXT: v_mul_f32_e32 v0, v0, v1
; CHECK-NEXT: v_mul_f32_e32 v0, v2, v0
; CHECK-NEXT: s_setpc_b64 s[30:31]
entry:
%fabs.y = call float @llvm.fabs.f32(float %y)
%div = fdiv float %x, %fabs.y, !fpmath !0
ret float %div
}
define float @fdiv_fast_daz_rhs_signbit_known_negative(float %x, float %y) #1 {
; CHECK-LABEL: fdiv_fast_daz_rhs_signbit_known_negative:
; CHECK: ; %bb.0: ; %entry
; CHECK-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; CHECK-NEXT: s_mov_b32 s4, 0x6f800000
; CHECK-NEXT: v_mov_b32_e32 v2, 0x2f800000
; CHECK-NEXT: v_cmp_gt_f32_e64 vcc, |v1|, s4
; CHECK-NEXT: v_cndmask_b32_e32 v2, 1.0, v2, vcc
; CHECK-NEXT: v_mul_f32_e64 v1, -|v1|, v2
; CHECK-NEXT: v_rcp_f32_e32 v1, v1
; CHECK-NEXT: v_mul_f32_e32 v0, v0, v1
; CHECK-NEXT: v_mul_f32_e32 v0, v2, v0
; CHECK-NEXT: s_setpc_b64 s[30:31]
entry:
%fabs.y = call float @llvm.fabs.f32(float %y)
%neg.fabs.y = fneg float %fabs.y
%div = fdiv float %x, %neg.fabs.y, !fpmath !0
ret float %div
}
define float @fdiv_fast_daz_rhs_signbit_known_zero_uitofp(float %x, i32 %y.i) #1 {
; CHECK-LABEL: fdiv_fast_daz_rhs_signbit_known_zero_uitofp:
; CHECK: ; %bb.0: ; %entry
; CHECK-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; CHECK-NEXT: v_cvt_f32_u32_e32 v1, v1
; CHECK-NEXT: s_mov_b32 s4, 0x6f800000
; CHECK-NEXT: v_mov_b32_e32 v2, 0x2f800000
; CHECK-NEXT: v_cmp_lt_f32_e32 vcc, s4, v1
; CHECK-NEXT: v_cndmask_b32_e32 v2, 1.0, v2, vcc
; CHECK-NEXT: v_mul_f32_e32 v1, v1, v2
; CHECK-NEXT: v_rcp_f32_e32 v1, v1
; CHECK-NEXT: v_mul_f32_e32 v0, v0, v1
; CHECK-NEXT: v_mul_f32_e32 v0, v2, v0
; CHECK-NEXT: s_setpc_b64 s[30:31]
entry:
%y.f = uitofp i32 %y.i to float
%div = fdiv float %x, %y.f, !fpmath !0
ret float %div
}
define float @fdiv_fast_daz_rhs_signbit_known_zero_maxnum_fabs(float %x, float %y, float %z) #1 {
; CHECK-LABEL: fdiv_fast_daz_rhs_signbit_known_zero_maxnum_fabs:
; CHECK: ; %bb.0: ; %entry
; CHECK-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; CHECK-NEXT: v_max_f32_e64 v2, |v2|, |v2|
; CHECK-NEXT: v_max_f32_e64 v1, |v1|, |v1|
; CHECK-NEXT: v_max_f32_e32 v1, v1, v2
; CHECK-NEXT: s_mov_b32 s4, 0x6f800000
; CHECK-NEXT: v_mov_b32_e32 v2, 0x2f800000
; CHECK-NEXT: v_cmp_lt_f32_e32 vcc, s4, v1
; CHECK-NEXT: v_cndmask_b32_e32 v2, 1.0, v2, vcc
; CHECK-NEXT: v_mul_f32_e32 v1, v1, v2
; CHECK-NEXT: v_rcp_f32_e32 v1, v1
; CHECK-NEXT: v_mul_f32_e32 v0, v0, v1
; CHECK-NEXT: v_mul_f32_e32 v0, v2, v0
; CHECK-NEXT: s_setpc_b64 s[30:31]
entry:
%y.abs = call float @llvm.fabs.f32(float %y)
%z.abs = call float @llvm.fabs.f32(float %z)
%max.abs = call float @llvm.maxnum.f32(float %y.abs, float %z.abs)
%div = fdiv float %x, %max.abs, !fpmath !0
ret float %div
}
define float @fdiv_fast_daz_rhs_signbit_known_zero_minnum_fabs(float %x, float %y, float %z) #1 {
; CHECK-LABEL: fdiv_fast_daz_rhs_signbit_known_zero_minnum_fabs:
; CHECK: ; %bb.0: ; %entry
; CHECK-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; CHECK-NEXT: v_max_f32_e64 v2, |v2|, |v2|
; CHECK-NEXT: v_max_f32_e64 v1, |v1|, |v1|
; CHECK-NEXT: v_min_f32_e32 v1, v1, v2
; CHECK-NEXT: s_mov_b32 s4, 0x6f800000
; CHECK-NEXT: v_mov_b32_e32 v2, 0x2f800000
; CHECK-NEXT: v_cmp_lt_f32_e32 vcc, s4, v1
; CHECK-NEXT: v_cndmask_b32_e32 v2, 1.0, v2, vcc
; CHECK-NEXT: v_mul_f32_e32 v1, v1, v2
; CHECK-NEXT: v_rcp_f32_e32 v1, v1
; CHECK-NEXT: v_mul_f32_e32 v0, v0, v1
; CHECK-NEXT: v_mul_f32_e32 v0, v2, v0
; CHECK-NEXT: s_setpc_b64 s[30:31]
entry:
%y.abs = call float @llvm.fabs.f32(float %y)
%z.abs = call float @llvm.fabs.f32(float %z)
%max.abs = call float @llvm.minnum.f32(float %y.abs, float %z.abs)
%div = fdiv float %x, %max.abs, !fpmath !0
ret float %div
}
define float @fdiv_fast_daz_rhs_signbit_known_zero_maximum_fabs(float %x, float %y, float %z) #1 {
; CHECK-LABEL: fdiv_fast_daz_rhs_signbit_known_zero_maximum_fabs:
; CHECK: ; %bb.0: ; %entry
; CHECK-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; CHECK-NEXT: v_max_f32_e64 v3, |v1|, |v2|
; CHECK-NEXT: v_mov_b32_e32 v4, 0x7fc00000
; CHECK-NEXT: v_cmp_o_f32_e64 vcc, |v1|, |v2|
; CHECK-NEXT: v_cndmask_b32_e32 v1, v4, v3, vcc
; CHECK-NEXT: s_mov_b32 s4, 0x6f800000
; CHECK-NEXT: v_mov_b32_e32 v2, 0x2f800000
; CHECK-NEXT: v_cmp_lt_f32_e32 vcc, s4, v1
; CHECK-NEXT: v_cndmask_b32_e32 v2, 1.0, v2, vcc
; CHECK-NEXT: v_mul_f32_e32 v1, v1, v2
; CHECK-NEXT: v_rcp_f32_e32 v1, v1
; CHECK-NEXT: v_mul_f32_e32 v0, v0, v1
; CHECK-NEXT: v_mul_f32_e32 v0, v2, v0
; CHECK-NEXT: s_setpc_b64 s[30:31]
entry:
%y.abs = call float @llvm.fabs.f32(float %y)
%z.abs = call float @llvm.fabs.f32(float %z)
%max.abs = call float @llvm.maximum.f32(float %y.abs, float %z.abs)
%div = fdiv float %x, %max.abs, !fpmath !0
ret float %div
}
define float @fdiv_fast_daz_rhs_signbit_known_zero_minimum_fabs(float %x, float %y, float %z) #1 {
; CHECK-LABEL: fdiv_fast_daz_rhs_signbit_known_zero_minimum_fabs:
; CHECK: ; %bb.0: ; %entry
; CHECK-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; CHECK-NEXT: v_min_f32_e64 v3, |v1|, |v2|
; CHECK-NEXT: v_mov_b32_e32 v4, 0x7fc00000
; CHECK-NEXT: v_cmp_o_f32_e64 vcc, |v1|, |v2|
; CHECK-NEXT: v_cndmask_b32_e32 v1, v4, v3, vcc
; CHECK-NEXT: s_mov_b32 s4, 0x6f800000
; CHECK-NEXT: v_mov_b32_e32 v2, 0x2f800000
; CHECK-NEXT: v_cmp_lt_f32_e32 vcc, s4, v1
; CHECK-NEXT: v_cndmask_b32_e32 v2, 1.0, v2, vcc
; CHECK-NEXT: v_mul_f32_e32 v1, v1, v2
; CHECK-NEXT: v_rcp_f32_e32 v1, v1
; CHECK-NEXT: v_mul_f32_e32 v0, v0, v1
; CHECK-NEXT: v_mul_f32_e32 v0, v2, v0
; CHECK-NEXT: s_setpc_b64 s[30:31]
entry:
%y.abs = call float @llvm.fabs.f32(float %y)
%z.abs = call float @llvm.fabs.f32(float %z)
%max.abs = call float @llvm.minimum.f32(float %y.abs, float %z.abs)
%div = fdiv float %x, %max.abs, !fpmath !0
ret float %div
}
define float @fdiv_fast_daz_rhs_signbit_known_zero_maximumnum_fabs(float %x, float %y, float %z) #1 {
; CHECK-LABEL: fdiv_fast_daz_rhs_signbit_known_zero_maximumnum_fabs:
; CHECK: ; %bb.0: ; %entry
; CHECK-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; CHECK-NEXT: v_max_f32_e64 v2, |v2|, |v2|
; CHECK-NEXT: v_max_f32_e64 v1, |v1|, |v1|
; CHECK-NEXT: v_max_f32_e32 v1, v1, v2
; CHECK-NEXT: s_mov_b32 s4, 0x6f800000
; CHECK-NEXT: v_mov_b32_e32 v2, 0x2f800000
; CHECK-NEXT: v_cmp_lt_f32_e32 vcc, s4, v1
; CHECK-NEXT: v_cndmask_b32_e32 v2, 1.0, v2, vcc
; CHECK-NEXT: v_mul_f32_e32 v1, v1, v2
; CHECK-NEXT: v_rcp_f32_e32 v1, v1
; CHECK-NEXT: v_mul_f32_e32 v0, v0, v1
; CHECK-NEXT: v_mul_f32_e32 v0, v2, v0
; CHECK-NEXT: s_setpc_b64 s[30:31]
entry:
%y.abs = call float @llvm.fabs.f32(float %y)
%z.abs = call float @llvm.fabs.f32(float %z)
%max.abs = call float @llvm.maximumnum.f32(float %y.abs, float %z.abs)
%div = fdiv float %x, %max.abs, !fpmath !0
ret float %div
}
define float @fdiv_fast_daz_rhs_signbit_known_zero_minimumnum_fabs(float %x, float %y, float %z) #1 {
; CHECK-LABEL: fdiv_fast_daz_rhs_signbit_known_zero_minimumnum_fabs:
; CHECK: ; %bb.0: ; %entry
; CHECK-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; CHECK-NEXT: v_max_f32_e64 v2, |v2|, |v2|
; CHECK-NEXT: v_max_f32_e64 v1, |v1|, |v1|
; CHECK-NEXT: v_min_f32_e32 v1, v1, v2
; CHECK-NEXT: s_mov_b32 s4, 0x6f800000
; CHECK-NEXT: v_mov_b32_e32 v2, 0x2f800000
; CHECK-NEXT: v_cmp_lt_f32_e32 vcc, s4, v1
; CHECK-NEXT: v_cndmask_b32_e32 v2, 1.0, v2, vcc
; CHECK-NEXT: v_mul_f32_e32 v1, v1, v2
; CHECK-NEXT: v_rcp_f32_e32 v1, v1
; CHECK-NEXT: v_mul_f32_e32 v0, v0, v1
; CHECK-NEXT: v_mul_f32_e32 v0, v2, v0
; CHECK-NEXT: s_setpc_b64 s[30:31]
entry:
%y.abs = call float @llvm.fabs.f32(float %y)
%z.abs = call float @llvm.fabs.f32(float %z)
%max.abs = call float @llvm.minimumnum.f32(float %y.abs, float %z.abs)
%div = fdiv float %x, %max.abs, !fpmath !0
ret float %div
}
; Negative test
define float @fdiv_fast_daz_rhs_signbit_maybe0_minimumnum_0(float %x, float %y, float %z) #1 {
; CHECK-LABEL: fdiv_fast_daz_rhs_signbit_maybe0_minimumnum_0:
; CHECK: ; %bb.0: ; %entry
; CHECK-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; CHECK-NEXT: v_max_f32_e64 v2, |v2|, |v2|
; CHECK-NEXT: v_max_f32_e32 v1, v1, v1
; CHECK-NEXT: v_min_f32_e32 v1, v1, v2
; CHECK-NEXT: s_mov_b32 s4, 0x6f800000
; CHECK-NEXT: v_mov_b32_e32 v2, 0x2f800000
; CHECK-NEXT: v_cmp_gt_f32_e64 vcc, |v1|, s4
; CHECK-NEXT: v_cndmask_b32_e32 v2, 1.0, v2, vcc
; CHECK-NEXT: v_mul_f32_e32 v1, v1, v2
; CHECK-NEXT: v_rcp_f32_e32 v1, v1
; CHECK-NEXT: v_mul_f32_e32 v0, v0, v1
; CHECK-NEXT: v_mul_f32_e32 v0, v2, v0
; CHECK-NEXT: s_setpc_b64 s[30:31]
entry:
%z.abs = call float @llvm.fabs.f32(float %z)
%max.abs = call float @llvm.minimumnum.f32(float %y, float %z.abs)
%div = fdiv float %x, %max.abs, !fpmath !0
ret float %div
}
; Negative test
define float @fdiv_fast_daz_rhs_signbit_maybe0_minimumnum_1(float %x, float %y, float %z) #1 {
; CHECK-LABEL: fdiv_fast_daz_rhs_signbit_maybe0_minimumnum_1:
; CHECK: ; %bb.0: ; %entry
; CHECK-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; CHECK-NEXT: v_max_f32_e32 v2, v2, v2
; CHECK-NEXT: v_max_f32_e64 v1, |v1|, |v1|
; CHECK-NEXT: v_min_f32_e32 v1, v1, v2
; CHECK-NEXT: s_mov_b32 s4, 0x6f800000
; CHECK-NEXT: v_mov_b32_e32 v2, 0x2f800000
; CHECK-NEXT: v_cmp_gt_f32_e64 vcc, |v1|, s4
; CHECK-NEXT: v_cndmask_b32_e32 v2, 1.0, v2, vcc
; CHECK-NEXT: v_mul_f32_e32 v1, v1, v2
; CHECK-NEXT: v_rcp_f32_e32 v1, v1
; CHECK-NEXT: v_mul_f32_e32 v0, v0, v1
; CHECK-NEXT: v_mul_f32_e32 v0, v2, v0
; CHECK-NEXT: s_setpc_b64 s[30:31]
entry:
%y.abs = call float @llvm.fabs.f32(float %y)
%max.abs = call float @llvm.minimumnum.f32(float %y.abs, float %z)
%div = fdiv float %x, %max.abs, !fpmath !0
ret float %div
}
; Positive operand with maximum implies both operands are positive
define float @fdiv_fast_daz_rhs_signbit_known0_maximumnum_one_fabs_0(float %x, float %y, float %z) #1 {
; CHECK-LABEL: fdiv_fast_daz_rhs_signbit_known0_maximumnum_one_fabs_0:
; CHECK: ; %bb.0: ; %entry
; CHECK-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; CHECK-NEXT: v_max_f32_e32 v1, v2, v2
; CHECK-NEXT: v_max_f32_e64 v2, |v2|, |v2|
; CHECK-NEXT: v_max_f32_e32 v1, v2, v1
; CHECK-NEXT: s_mov_b32 s4, 0x6f800000
; CHECK-NEXT: v_mov_b32_e32 v2, 0x2f800000
; CHECK-NEXT: v_cmp_gt_f32_e64 vcc, |v1|, s4
; CHECK-NEXT: v_cndmask_b32_e32 v2, 1.0, v2, vcc
; CHECK-NEXT: v_mul_f32_e32 v1, v1, v2
; CHECK-NEXT: v_rcp_f32_e32 v1, v1
; CHECK-NEXT: v_mul_f32_e32 v0, v0, v1
; CHECK-NEXT: v_mul_f32_e32 v0, v2, v0
; CHECK-NEXT: s_setpc_b64 s[30:31]
entry:
%y.abs = call float @llvm.fabs.f32(float %z)
%max.abs = call float @llvm.maximumnum.f32(float %y.abs, float %z)
%div = fdiv float %x, %max.abs, !fpmath !0
ret float %div
}
; Positive operand with maximum implies both operands are positive
define float @fdiv_fast_daz_rhs_signbit_known0_maximumnum_one_fabs_1(float %x, float %y, float %z) #1 {
; CHECK-LABEL: fdiv_fast_daz_rhs_signbit_known0_maximumnum_one_fabs_1:
; CHECK: ; %bb.0: ; %entry
; CHECK-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; CHECK-NEXT: v_max_f32_e64 v2, |v2|, |v2|
; CHECK-NEXT: v_max_f32_e32 v1, v1, v1
; CHECK-NEXT: v_max_f32_e32 v1, v1, v2
; CHECK-NEXT: s_mov_b32 s4, 0x6f800000
; CHECK-NEXT: v_mov_b32_e32 v2, 0x2f800000
; CHECK-NEXT: v_cmp_gt_f32_e64 vcc, |v1|, s4
; CHECK-NEXT: v_cndmask_b32_e32 v2, 1.0, v2, vcc
; CHECK-NEXT: v_mul_f32_e32 v1, v1, v2
; CHECK-NEXT: v_rcp_f32_e32 v1, v1
; CHECK-NEXT: v_mul_f32_e32 v0, v0, v1
; CHECK-NEXT: v_mul_f32_e32 v0, v2, v0
; CHECK-NEXT: s_setpc_b64 s[30:31]
entry:
%z.abs = call float @llvm.fabs.f32(float %z)
%max.abs = call float @llvm.maximumnum.f32(float %y, float %z.abs)
%div = fdiv float %x, %max.abs, !fpmath !0
ret float %div
}
declare float @llvm.maxnum.f32(float, float) #0
declare float @llvm.minnum.f32(float, float) #0
declare float @llvm.maximum.f32(float, float) #0
declare float @llvm.minimum.f32(float, float) #0
declare float @llvm.maximumnum.f32(float, float) #0
declare float @llvm.minimumnum.f32(float, float) #0
attributes #0 = { nocallback nocreateundeforpoison nofree nosync nounwind speculatable willreturn memory(none) }
attributes #1 = { denormal_fpenv(float: preservesign) }
!0 = !{float 2.500000e+00}
Reference in new issue View Git Blame Copy Permalink