Files
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

215 lines
6.3 KiB
LLVM

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
; RUN: llc < %s -mtriple=arm64-eabi -mattr=fullfp16,sve | FileCheck %s --check-prefixes=CHECK,CHECK-SD
; RUN: llc < %s -mtriple=arm64-eabi -mattr=fullfp16,sve -global-isel -global-isel-abort=2 2>&1 | FileCheck %s --check-prefixes=CHECK,CHECK-GI
; CHECK-GI: warning: Instruction selection used fallback path for divnxv4f32_2
define float @divf32_2(float %a) nounwind {
; CHECK-SD-LABEL: divf32_2:
; CHECK-SD: // %bb.0:
; CHECK-SD-NEXT: fmov s1, #0.50000000
; CHECK-SD-NEXT: fmul s0, s0, s1
; CHECK-SD-NEXT: ret
;
; CHECK-GI-LABEL: divf32_2:
; CHECK-GI: // %bb.0:
; CHECK-GI-NEXT: fmov s1, #2.00000000
; CHECK-GI-NEXT: fdiv s0, s0, s1
; CHECK-GI-NEXT: ret
%r = fdiv float %a, 2.0
ret float %r
}
define float @divf32_2_arcp(float %a) nounwind {
; CHECK-SD-LABEL: divf32_2_arcp:
; CHECK-SD: // %bb.0:
; CHECK-SD-NEXT: fmov s1, #0.50000000
; CHECK-SD-NEXT: fmul s0, s0, s1
; CHECK-SD-NEXT: ret
;
; CHECK-GI-LABEL: divf32_2_arcp:
; CHECK-GI: // %bb.0:
; CHECK-GI-NEXT: fmov s1, #2.00000000
; CHECK-GI-NEXT: fdiv s0, s0, s1
; CHECK-GI-NEXT: ret
%r = fdiv arcp float %a, 2.0
ret float %r
}
define float @divf32_p75(float %a) nounwind {
; CHECK-LABEL: divf32_p75:
; CHECK: // %bb.0:
; CHECK-NEXT: fmov s1, #0.75000000
; CHECK-NEXT: fdiv s0, s0, s1
; CHECK-NEXT: ret
%r = fdiv float %a, 0.75
ret float %r
}
define float @divf32_p75_arcp(float %a) nounwind {
; CHECK-SD-LABEL: divf32_p75_arcp:
; CHECK-SD: // %bb.0:
; CHECK-SD-NEXT: mov w8, #43691 // =0xaaab
; CHECK-SD-NEXT: movk w8, #16298, lsl #16
; CHECK-SD-NEXT: fmov s1, w8
; CHECK-SD-NEXT: fmul s0, s0, s1
; CHECK-SD-NEXT: ret
;
; CHECK-GI-LABEL: divf32_p75_arcp:
; CHECK-GI: // %bb.0:
; CHECK-GI-NEXT: fmov s1, #0.75000000
; CHECK-GI-NEXT: fdiv s0, s0, s1
; CHECK-GI-NEXT: ret
%r = fdiv arcp float %a, 0.75
ret float %r
}
define half @divf16_2(half %a) nounwind {
; CHECK-SD-LABEL: divf16_2:
; CHECK-SD: // %bb.0:
; CHECK-SD-NEXT: fmov h1, #0.50000000
; CHECK-SD-NEXT: fmul h0, h0, h1
; CHECK-SD-NEXT: ret
;
; CHECK-GI-LABEL: divf16_2:
; CHECK-GI: // %bb.0:
; CHECK-GI-NEXT: fmov h1, #2.00000000
; CHECK-GI-NEXT: fdiv h0, h0, h1
; CHECK-GI-NEXT: ret
%r = fdiv half %a, 2.0
ret half %r
}
define half @divf16_32768(half %a) nounwind {
; CHECK-SD-LABEL: divf16_32768:
; CHECK-SD: // %bb.0:
; CHECK-SD-NEXT: mov w8, #30720 // =0x7800
; CHECK-SD-NEXT: fmov h1, w8
; CHECK-SD-NEXT: fdiv h0, h0, h1
; CHECK-SD-NEXT: ret
;
; CHECK-GI-LABEL: divf16_32768:
; CHECK-GI: // %bb.0:
; CHECK-GI-NEXT: adrp x8, .LCPI5_0
; CHECK-GI-NEXT: ldr h1, [x8, :lo12:.LCPI5_0]
; CHECK-GI-NEXT: fdiv h0, h0, h1
; CHECK-GI-NEXT: ret
%r = fdiv half %a, 32768.0
ret half %r
}
define half @divf16_32768_arcp(half %a) nounwind {
; CHECK-SD-LABEL: divf16_32768_arcp:
; CHECK-SD: // %bb.0:
; CHECK-SD-NEXT: mov w8, #30720 // =0x7800
; CHECK-SD-NEXT: fmov h1, w8
; CHECK-SD-NEXT: fdiv h0, h0, h1
; CHECK-SD-NEXT: ret
;
; CHECK-GI-LABEL: divf16_32768_arcp:
; CHECK-GI: // %bb.0:
; CHECK-GI-NEXT: adrp x8, .LCPI6_0
; CHECK-GI-NEXT: ldr h1, [x8, :lo12:.LCPI6_0]
; CHECK-GI-NEXT: fdiv h0, h0, h1
; CHECK-GI-NEXT: ret
%r = fdiv arcp half %a, 32768.0
ret half %r
}
define double @divf64_2(double %a) nounwind {
; CHECK-SD-LABEL: divf64_2:
; CHECK-SD: // %bb.0:
; CHECK-SD-NEXT: fmov d1, #0.50000000
; CHECK-SD-NEXT: fmul d0, d0, d1
; CHECK-SD-NEXT: ret
;
; CHECK-GI-LABEL: divf64_2:
; CHECK-GI: // %bb.0:
; CHECK-GI-NEXT: fmov d1, #2.00000000
; CHECK-GI-NEXT: fdiv d0, d0, d1
; CHECK-GI-NEXT: ret
%r = fdiv double %a, 2.0
ret double %r
}
define <4 x float> @divv4f32_2(<4 x float> %a) nounwind {
; CHECK-SD-LABEL: divv4f32_2:
; CHECK-SD: // %bb.0:
; CHECK-SD-NEXT: movi v1.4s, #63, lsl #24
; CHECK-SD-NEXT: fmul v0.4s, v0.4s, v1.4s
; CHECK-SD-NEXT: ret
;
; CHECK-GI-LABEL: divv4f32_2:
; CHECK-GI: // %bb.0:
; CHECK-GI-NEXT: movi v1.4s, #64, lsl #24
; CHECK-GI-NEXT: fdiv v0.4s, v0.4s, v1.4s
; CHECK-GI-NEXT: ret
%r = fdiv <4 x float> %a, <float 2.0, float 2.0, float 2.0, float 2.0>
ret <4 x float> %r
}
define <4 x float> @divv4f32_2_arcp(<4 x float> %a) nounwind {
; CHECK-SD-LABEL: divv4f32_2_arcp:
; CHECK-SD: // %bb.0:
; CHECK-SD-NEXT: movi v1.4s, #63, lsl #24
; CHECK-SD-NEXT: fmul v0.4s, v0.4s, v1.4s
; CHECK-SD-NEXT: ret
;
; CHECK-GI-LABEL: divv4f32_2_arcp:
; CHECK-GI: // %bb.0:
; CHECK-GI-NEXT: movi v1.4s, #64, lsl #24
; CHECK-GI-NEXT: fdiv v0.4s, v0.4s, v1.4s
; CHECK-GI-NEXT: ret
%r = fdiv arcp <4 x float> %a, <float 2.0, float 2.0, float 2.0, float 2.0>
ret <4 x float> %r
}
define <4 x float> @divv4f32_3(<4 x float> %a) nounwind {
; CHECK-LABEL: divv4f32_3:
; CHECK: // %bb.0:
; CHECK-NEXT: fmov v1.4s, #3.00000000
; CHECK-NEXT: fdiv v0.4s, v0.4s, v1.4s
; CHECK-NEXT: ret
%r = fdiv <4 x float> %a, <float 3.0, float 3.0, float 3.0, float 3.0>
ret <4 x float> %r
}
define <4 x float> @divv4f32_3_arcp(<4 x float> %a) nounwind {
; CHECK-SD-LABEL: divv4f32_3_arcp:
; CHECK-SD: // %bb.0:
; CHECK-SD-NEXT: mov w8, #43691 // =0xaaab
; CHECK-SD-NEXT: movk w8, #16042, lsl #16
; CHECK-SD-NEXT: dup v1.4s, w8
; CHECK-SD-NEXT: fmul v0.4s, v0.4s, v1.4s
; CHECK-SD-NEXT: ret
;
; CHECK-GI-LABEL: divv4f32_3_arcp:
; CHECK-GI: // %bb.0:
; CHECK-GI-NEXT: fmov v1.4s, #3.00000000
; CHECK-GI-NEXT: fdiv v0.4s, v0.4s, v1.4s
; CHECK-GI-NEXT: ret
%r = fdiv arcp <4 x float> %a, <float 3.0, float 3.0, float 3.0, float 3.0>
ret <4 x float> %r
}
define <4 x float> @divv4f32_24816(<4 x float> %a) nounwind {
; CHECK-LABEL: divv4f32_24816:
; CHECK: // %bb.0:
; CHECK-NEXT: adrp x8, .LCPI12_0
; CHECK-NEXT: ldr q1, [x8, :lo12:.LCPI12_0]
; CHECK-NEXT: fdiv v0.4s, v0.4s, v1.4s
; CHECK-NEXT: ret
%r = fdiv <4 x float> %a, <float 2.0, float 4.0, float 8.0, float 16.0>
ret <4 x float> %r
}
define <vscale x 4 x float> @divnxv4f32_2(<vscale x 4 x float> %a) nounwind {
; CHECK-LABEL: divnxv4f32_2:
; CHECK: // %bb.0:
; CHECK-NEXT: ptrue p0.s
; CHECK-NEXT: fmul z0.s, p0/m, z0.s, #0.5
; CHECK-NEXT: ret
%r = fdiv <vscale x 4 x float> %a, splat (float 2.0)
ret <vscale x 4 x float> %r
}