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
138 lines
4.1 KiB
LLVM
138 lines
4.1 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc -mtriple=armv8.6a-none-none-eabi < %s | FileCheck %s --check-prefixes=CHECK
|
|
; RUN: llc -mtriple=armv8.6a-none-none-eabi -mattr=+bf16,+neon < %s | FileCheck %s --check-prefixes=CHECK
|
|
; RUN: llc -mtriple=armv8.6a-none-none-eabi -mattr=+bf16,+neon,+fullfp16 < %s | FileCheck %s --check-prefix=CHECK-FP16
|
|
; RUN: llc -mtriple=armv8.6a-none-none-eabi -mattr=+bf16,+neon -float-abi=hard < %s | FileCheck %s --check-prefix=CHECK-HARD
|
|
|
|
define bfloat @bitcast_zero() {
|
|
; CHECK-LABEL: bitcast_zero:
|
|
; CHECK: @ %bb.0:
|
|
; CHECK-NEXT: .pad #4
|
|
; CHECK-NEXT: sub sp, sp, #4
|
|
; CHECK-NEXT: mov r0, #0
|
|
; CHECK-NEXT: strh r0, [sp, #2]
|
|
; CHECK-NEXT: ldrh r0, [sp, #2]
|
|
; CHECK-NEXT: add sp, sp, #4
|
|
; CHECK-NEXT: bx lr
|
|
;
|
|
; CHECK-FP16-LABEL: bitcast_zero:
|
|
; CHECK-FP16: @ %bb.0:
|
|
; CHECK-FP16-NEXT: mov r0, #0
|
|
; CHECK-FP16-NEXT: vmov.f16 s0, r0
|
|
; CHECK-FP16-NEXT: vmov r0, s0
|
|
; CHECK-FP16-NEXT: bx lr
|
|
;
|
|
; CHECK-HARD-LABEL: bitcast_zero:
|
|
; CHECK-HARD: @ %bb.0:
|
|
; CHECK-HARD-NEXT: .pad #4
|
|
; CHECK-HARD-NEXT: sub sp, sp, #4
|
|
; CHECK-HARD-NEXT: mov r0, #0
|
|
; CHECK-HARD-NEXT: strh r0, [sp, #2]
|
|
; CHECK-HARD-NEXT: ldrh r0, [sp, #2]
|
|
; CHECK-HARD-NEXT: vmov s0, r0
|
|
; CHECK-HARD-NEXT: add sp, sp, #4
|
|
; CHECK-HARD-NEXT: bx lr
|
|
%z = bitcast i16 0 to bfloat
|
|
ret bfloat %z
|
|
}
|
|
|
|
define bfloat @zero() {
|
|
; CHECK-LABEL: zero:
|
|
; CHECK: @ %bb.0:
|
|
; CHECK-NEXT: adr r0, .LCPI1_0
|
|
; CHECK-NEXT: ldrh r0, [r0]
|
|
; CHECK-NEXT: bx lr
|
|
; CHECK-NEXT: .p2align 1
|
|
; CHECK-NEXT: @ %bb.1:
|
|
; CHECK-NEXT: .LCPI1_0:
|
|
; CHECK-NEXT: .short 0x0000 @ bfloat 0
|
|
;
|
|
; CHECK-FP16-LABEL: zero:
|
|
; CHECK-FP16: @ %bb.0:
|
|
; CHECK-FP16-NEXT: vldr.16 s0, .LCPI1_0
|
|
; CHECK-FP16-NEXT: vmov r0, s0
|
|
; CHECK-FP16-NEXT: bx lr
|
|
; CHECK-FP16-NEXT: .p2align 1
|
|
; CHECK-FP16-NEXT: @ %bb.1:
|
|
; CHECK-FP16-NEXT: .LCPI1_0:
|
|
; CHECK-FP16-NEXT: .short 0x0000 @ bfloat 0
|
|
;
|
|
; CHECK-HARD-LABEL: zero:
|
|
; CHECK-HARD: @ %bb.0:
|
|
; CHECK-HARD-NEXT: adr r0, .LCPI1_0
|
|
; CHECK-HARD-NEXT: ldrh r0, [r0]
|
|
; CHECK-HARD-NEXT: vmov s0, r0
|
|
; CHECK-HARD-NEXT: bx lr
|
|
; CHECK-HARD-NEXT: .p2align 1
|
|
; CHECK-HARD-NEXT: @ %bb.1:
|
|
; CHECK-HARD-NEXT: .LCPI1_0:
|
|
; CHECK-HARD-NEXT: .short 0x0000 @ bfloat 0
|
|
ret bfloat 0xR0000
|
|
}
|
|
|
|
define bfloat @bitcast_tenk() {
|
|
; CHECK-LABEL: bitcast_tenk:
|
|
; CHECK: @ %bb.0:
|
|
; CHECK-NEXT: .pad #4
|
|
; CHECK-NEXT: sub sp, sp, #4
|
|
; CHECK-NEXT: movw r0, #10000
|
|
; CHECK-NEXT: strh r0, [sp, #2]
|
|
; CHECK-NEXT: ldrh r0, [sp, #2]
|
|
; CHECK-NEXT: add sp, sp, #4
|
|
; CHECK-NEXT: bx lr
|
|
;
|
|
; CHECK-FP16-LABEL: bitcast_tenk:
|
|
; CHECK-FP16: @ %bb.0:
|
|
; CHECK-FP16-NEXT: movw r0, #10000
|
|
; CHECK-FP16-NEXT: vmov.f16 s0, r0
|
|
; CHECK-FP16-NEXT: vmov r0, s0
|
|
; CHECK-FP16-NEXT: bx lr
|
|
;
|
|
; CHECK-HARD-LABEL: bitcast_tenk:
|
|
; CHECK-HARD: @ %bb.0:
|
|
; CHECK-HARD-NEXT: .pad #4
|
|
; CHECK-HARD-NEXT: sub sp, sp, #4
|
|
; CHECK-HARD-NEXT: movw r0, #10000
|
|
; CHECK-HARD-NEXT: strh r0, [sp, #2]
|
|
; CHECK-HARD-NEXT: ldrh r0, [sp, #2]
|
|
; CHECK-HARD-NEXT: vmov s0, r0
|
|
; CHECK-HARD-NEXT: add sp, sp, #4
|
|
; CHECK-HARD-NEXT: bx lr
|
|
%z = bitcast i16 10000 to bfloat
|
|
ret bfloat %z
|
|
}
|
|
|
|
define bfloat @minus0() {
|
|
; CHECK-LABEL: minus0:
|
|
; CHECK: @ %bb.0:
|
|
; CHECK-NEXT: adr r0, .LCPI3_0
|
|
; CHECK-NEXT: ldrh r0, [r0]
|
|
; CHECK-NEXT: bx lr
|
|
; CHECK-NEXT: .p2align 1
|
|
; CHECK-NEXT: @ %bb.1:
|
|
; CHECK-NEXT: .LCPI3_0:
|
|
; CHECK-NEXT: .short 0x8000 @ bfloat -0
|
|
;
|
|
; CHECK-FP16-LABEL: minus0:
|
|
; CHECK-FP16: @ %bb.0:
|
|
; CHECK-FP16-NEXT: vldr.16 s0, .LCPI3_0
|
|
; CHECK-FP16-NEXT: vmov r0, s0
|
|
; CHECK-FP16-NEXT: bx lr
|
|
; CHECK-FP16-NEXT: .p2align 1
|
|
; CHECK-FP16-NEXT: @ %bb.1:
|
|
; CHECK-FP16-NEXT: .LCPI3_0:
|
|
; CHECK-FP16-NEXT: .short 0x8000 @ bfloat -0
|
|
;
|
|
; CHECK-HARD-LABEL: minus0:
|
|
; CHECK-HARD: @ %bb.0:
|
|
; CHECK-HARD-NEXT: adr r0, .LCPI3_0
|
|
; CHECK-HARD-NEXT: ldrh r0, [r0]
|
|
; CHECK-HARD-NEXT: vmov s0, r0
|
|
; CHECK-HARD-NEXT: bx lr
|
|
; CHECK-HARD-NEXT: .p2align 1
|
|
; CHECK-HARD-NEXT: @ %bb.1:
|
|
; CHECK-HARD-NEXT: .LCPI3_0:
|
|
; CHECK-HARD-NEXT: .short 0x8000 @ bfloat -0
|
|
ret bfloat 0xR8000
|
|
}
|