mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-11-26 04:32:44 +01:00
c955b9f411
This patch adds codegen for the following BFloat operations to the ARM backend: * concatenation of bf16 vectors * bf16 vector element extraction * bf16 vector element insertion * duplication of a bf16 value into each lane of a vector * duplication of a bf16 vector lane into each lane Differential Revision: https://reviews.llvm.org/D81411
46 lines
1.4 KiB
LLVM
46 lines
1.4 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc -mtriple=armv8.6a-arm-none-eabi -mattr=+bf16,+neon,+fullfp16 < %s | FileCheck %s
|
|
|
|
target datalayout = "e-m:e-p:32:32-Fi8-i64:64-v128:64:128-a:0:32-n32-S64"
|
|
target triple = "armv8.6a-arm-none-eabi"
|
|
|
|
define arm_aapcs_vfpcc bfloat @test_vgetq_lane_bf16_even(<8 x bfloat> %v) {
|
|
; CHECK-LABEL: test_vgetq_lane_bf16_even:
|
|
; CHECK: @ %bb.0: @ %entry
|
|
; CHECK-NEXT: vmov.f32 s0, s3
|
|
; CHECK-NEXT: bx lr
|
|
entry:
|
|
%0 = extractelement <8 x bfloat> %v, i32 6
|
|
ret bfloat %0
|
|
}
|
|
|
|
define arm_aapcs_vfpcc bfloat @test_vgetq_lane_bf16_odd(<8 x bfloat> %v) {
|
|
; CHECK-LABEL: test_vgetq_lane_bf16_odd:
|
|
; CHECK: @ %bb.0: @ %entry
|
|
; CHECK-NEXT: vmovx.f16 s0, s3
|
|
; CHECK-NEXT: bx lr
|
|
entry:
|
|
%0 = extractelement <8 x bfloat> %v, i32 7
|
|
ret bfloat %0
|
|
}
|
|
|
|
define arm_aapcs_vfpcc bfloat @test_vget_lane_bf16_even(<4 x bfloat> %v) {
|
|
; CHECK-LABEL: test_vget_lane_bf16_even:
|
|
; CHECK: @ %bb.0: @ %entry
|
|
; CHECK-NEXT: vmov.f32 s0, s1
|
|
; CHECK-NEXT: bx lr
|
|
entry:
|
|
%0 = extractelement <4 x bfloat> %v, i32 2
|
|
ret bfloat %0
|
|
}
|
|
|
|
define arm_aapcs_vfpcc bfloat @test_vget_lane_bf16_odd(<4 x bfloat> %v) {
|
|
; CHECK-LABEL: test_vget_lane_bf16_odd:
|
|
; CHECK: @ %bb.0: @ %entry
|
|
; CHECK-NEXT: vmovx.f16 s0, s0
|
|
; CHECK-NEXT: bx lr
|
|
entry:
|
|
%0 = extractelement <4 x bfloat> %v, i32 1
|
|
ret bfloat %0
|
|
}
|