834cc88c5d
lowerBuildVectorAsBroadcast will not broadcast splat constants in all cases, resulting in a lot of situations where a full width vector load that has failed to fold but is loading splat constant values could use a broadcast load instruction just as cheaply, and save constant pool space. NOTE: SSE3 targets can use MOVDDUP but not all SSE era CPUs can perform this as cheaply as a vector load, we will need to add scheduler model checks if we want to pursue this. This is an updated commit of 98061013e01207444cfd3980cde17b5e75764fbe after being reverted at a279a09ab9524d1d74ef29b34618102d4b202e2f
277 lines
10 KiB
LLVM
277 lines
10 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+avx | FileCheck %s --check-prefix=X86 --check-prefix=X86-AVX
|
|
; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+avx512vl | FileCheck %s --check-prefix=X86 --check-prefix=X86-AVX512VL
|
|
; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+avx512dq,+avx512vl | FileCheck %s --check-prefix=X86 --check-prefix=X86-AVX512VLDQ
|
|
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s --check-prefix=X64 --check-prefix=X64-AVX
|
|
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512vl | FileCheck %s --check-prefix=X64 --check-prefix=X64-AVX512VL
|
|
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512dq,+avx512vl | FileCheck %s --check-prefix=X64 --check-prefix=X64-AVX512VLDQ
|
|
|
|
define <2 x double> @fabs_v2f64(<2 x double> %p) {
|
|
; X86-AVX-LABEL: fabs_v2f64:
|
|
; X86-AVX: # %bb.0:
|
|
; X86-AVX-NEXT: vandps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0, %xmm0
|
|
; X86-AVX-NEXT: retl
|
|
;
|
|
; X86-AVX512VL-LABEL: fabs_v2f64:
|
|
; X86-AVX512VL: # %bb.0:
|
|
; X86-AVX512VL-NEXT: vpandq {{\.?LCPI[0-9]+_[0-9]+}}{1to2}, %xmm0, %xmm0
|
|
; X86-AVX512VL-NEXT: retl
|
|
;
|
|
; X86-AVX512VLDQ-LABEL: fabs_v2f64:
|
|
; X86-AVX512VLDQ: # %bb.0:
|
|
; X86-AVX512VLDQ-NEXT: vandpd {{\.?LCPI[0-9]+_[0-9]+}}{1to2}, %xmm0, %xmm0
|
|
; X86-AVX512VLDQ-NEXT: retl
|
|
;
|
|
; X64-AVX-LABEL: fabs_v2f64:
|
|
; X64-AVX: # %bb.0:
|
|
; X64-AVX-NEXT: vandps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
|
|
; X64-AVX-NEXT: retq
|
|
;
|
|
; X64-AVX512VL-LABEL: fabs_v2f64:
|
|
; X64-AVX512VL: # %bb.0:
|
|
; X64-AVX512VL-NEXT: vpandq {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to2}, %xmm0, %xmm0
|
|
; X64-AVX512VL-NEXT: retq
|
|
;
|
|
; X64-AVX512VLDQ-LABEL: fabs_v2f64:
|
|
; X64-AVX512VLDQ: # %bb.0:
|
|
; X64-AVX512VLDQ-NEXT: vandpd {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to2}, %xmm0, %xmm0
|
|
; X64-AVX512VLDQ-NEXT: retq
|
|
%t = call <2 x double> @llvm.fabs.v2f64(<2 x double> %p)
|
|
ret <2 x double> %t
|
|
}
|
|
declare <2 x double> @llvm.fabs.v2f64(<2 x double> %p)
|
|
|
|
define <4 x float> @fabs_v4f32(<4 x float> %p) {
|
|
; X86-AVX-LABEL: fabs_v4f32:
|
|
; X86-AVX: # %bb.0:
|
|
; X86-AVX-NEXT: vandps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0, %xmm0
|
|
; X86-AVX-NEXT: retl
|
|
;
|
|
; X86-AVX512VL-LABEL: fabs_v4f32:
|
|
; X86-AVX512VL: # %bb.0:
|
|
; X86-AVX512VL-NEXT: vpandd {{\.?LCPI[0-9]+_[0-9]+}}{1to4}, %xmm0, %xmm0
|
|
; X86-AVX512VL-NEXT: retl
|
|
;
|
|
; X86-AVX512VLDQ-LABEL: fabs_v4f32:
|
|
; X86-AVX512VLDQ: # %bb.0:
|
|
; X86-AVX512VLDQ-NEXT: vandps {{\.?LCPI[0-9]+_[0-9]+}}{1to4}, %xmm0, %xmm0
|
|
; X86-AVX512VLDQ-NEXT: retl
|
|
;
|
|
; X64-AVX-LABEL: fabs_v4f32:
|
|
; X64-AVX: # %bb.0:
|
|
; X64-AVX-NEXT: vandps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
|
|
; X64-AVX-NEXT: retq
|
|
;
|
|
; X64-AVX512VL-LABEL: fabs_v4f32:
|
|
; X64-AVX512VL: # %bb.0:
|
|
; X64-AVX512VL-NEXT: vpandd {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to4}, %xmm0, %xmm0
|
|
; X64-AVX512VL-NEXT: retq
|
|
;
|
|
; X64-AVX512VLDQ-LABEL: fabs_v4f32:
|
|
; X64-AVX512VLDQ: # %bb.0:
|
|
; X64-AVX512VLDQ-NEXT: vandps {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to4}, %xmm0, %xmm0
|
|
; X64-AVX512VLDQ-NEXT: retq
|
|
%t = call <4 x float> @llvm.fabs.v4f32(<4 x float> %p)
|
|
ret <4 x float> %t
|
|
}
|
|
declare <4 x float> @llvm.fabs.v4f32(<4 x float> %p)
|
|
|
|
define <4 x double> @fabs_v4f64(<4 x double> %p) {
|
|
; X86-AVX-LABEL: fabs_v4f64:
|
|
; X86-AVX: # %bb.0:
|
|
; X86-AVX-NEXT: vandps {{\.?LCPI[0-9]+_[0-9]+}}, %ymm0, %ymm0
|
|
; X86-AVX-NEXT: retl
|
|
;
|
|
; X86-AVX512VL-LABEL: fabs_v4f64:
|
|
; X86-AVX512VL: # %bb.0:
|
|
; X86-AVX512VL-NEXT: vpandq {{\.?LCPI[0-9]+_[0-9]+}}{1to4}, %ymm0, %ymm0
|
|
; X86-AVX512VL-NEXT: retl
|
|
;
|
|
; X86-AVX512VLDQ-LABEL: fabs_v4f64:
|
|
; X86-AVX512VLDQ: # %bb.0:
|
|
; X86-AVX512VLDQ-NEXT: vandpd {{\.?LCPI[0-9]+_[0-9]+}}{1to4}, %ymm0, %ymm0
|
|
; X86-AVX512VLDQ-NEXT: retl
|
|
;
|
|
; X64-AVX-LABEL: fabs_v4f64:
|
|
; X64-AVX: # %bb.0:
|
|
; X64-AVX-NEXT: vandps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
|
|
; X64-AVX-NEXT: retq
|
|
;
|
|
; X64-AVX512VL-LABEL: fabs_v4f64:
|
|
; X64-AVX512VL: # %bb.0:
|
|
; X64-AVX512VL-NEXT: vpandq {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to4}, %ymm0, %ymm0
|
|
; X64-AVX512VL-NEXT: retq
|
|
;
|
|
; X64-AVX512VLDQ-LABEL: fabs_v4f64:
|
|
; X64-AVX512VLDQ: # %bb.0:
|
|
; X64-AVX512VLDQ-NEXT: vandpd {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to4}, %ymm0, %ymm0
|
|
; X64-AVX512VLDQ-NEXT: retq
|
|
%t = call <4 x double> @llvm.fabs.v4f64(<4 x double> %p)
|
|
ret <4 x double> %t
|
|
}
|
|
declare <4 x double> @llvm.fabs.v4f64(<4 x double> %p)
|
|
|
|
define <8 x float> @fabs_v8f32(<8 x float> %p) {
|
|
; X86-AVX-LABEL: fabs_v8f32:
|
|
; X86-AVX: # %bb.0:
|
|
; X86-AVX-NEXT: vandps {{\.?LCPI[0-9]+_[0-9]+}}, %ymm0, %ymm0
|
|
; X86-AVX-NEXT: retl
|
|
;
|
|
; X86-AVX512VL-LABEL: fabs_v8f32:
|
|
; X86-AVX512VL: # %bb.0:
|
|
; X86-AVX512VL-NEXT: vpandd {{\.?LCPI[0-9]+_[0-9]+}}{1to8}, %ymm0, %ymm0
|
|
; X86-AVX512VL-NEXT: retl
|
|
;
|
|
; X86-AVX512VLDQ-LABEL: fabs_v8f32:
|
|
; X86-AVX512VLDQ: # %bb.0:
|
|
; X86-AVX512VLDQ-NEXT: vandps {{\.?LCPI[0-9]+_[0-9]+}}{1to8}, %ymm0, %ymm0
|
|
; X86-AVX512VLDQ-NEXT: retl
|
|
;
|
|
; X64-AVX-LABEL: fabs_v8f32:
|
|
; X64-AVX: # %bb.0:
|
|
; X64-AVX-NEXT: vandps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
|
|
; X64-AVX-NEXT: retq
|
|
;
|
|
; X64-AVX512VL-LABEL: fabs_v8f32:
|
|
; X64-AVX512VL: # %bb.0:
|
|
; X64-AVX512VL-NEXT: vpandd {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to8}, %ymm0, %ymm0
|
|
; X64-AVX512VL-NEXT: retq
|
|
;
|
|
; X64-AVX512VLDQ-LABEL: fabs_v8f32:
|
|
; X64-AVX512VLDQ: # %bb.0:
|
|
; X64-AVX512VLDQ-NEXT: vandps {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to8}, %ymm0, %ymm0
|
|
; X64-AVX512VLDQ-NEXT: retq
|
|
%t = call <8 x float> @llvm.fabs.v8f32(<8 x float> %p)
|
|
ret <8 x float> %t
|
|
}
|
|
declare <8 x float> @llvm.fabs.v8f32(<8 x float> %p)
|
|
|
|
define <8 x double> @fabs_v8f64(<8 x double> %p) {
|
|
; X86-AVX-LABEL: fabs_v8f64:
|
|
; X86-AVX: # %bb.0:
|
|
; X86-AVX-NEXT: vbroadcastsd {{.*#+}} ymm2 = [NaN,NaN,NaN,NaN]
|
|
; X86-AVX-NEXT: vandps %ymm2, %ymm0, %ymm0
|
|
; X86-AVX-NEXT: vandps %ymm2, %ymm1, %ymm1
|
|
; X86-AVX-NEXT: retl
|
|
;
|
|
; X86-AVX512VL-LABEL: fabs_v8f64:
|
|
; X86-AVX512VL: # %bb.0:
|
|
; X86-AVX512VL-NEXT: vpandq {{\.?LCPI[0-9]+_[0-9]+}}{1to8}, %zmm0, %zmm0
|
|
; X86-AVX512VL-NEXT: retl
|
|
;
|
|
; X86-AVX512VLDQ-LABEL: fabs_v8f64:
|
|
; X86-AVX512VLDQ: # %bb.0:
|
|
; X86-AVX512VLDQ-NEXT: vandpd {{\.?LCPI[0-9]+_[0-9]+}}{1to8}, %zmm0, %zmm0
|
|
; X86-AVX512VLDQ-NEXT: retl
|
|
;
|
|
; X64-AVX-LABEL: fabs_v8f64:
|
|
; X64-AVX: # %bb.0:
|
|
; X64-AVX-NEXT: vbroadcastsd {{.*#+}} ymm2 = [NaN,NaN,NaN,NaN]
|
|
; X64-AVX-NEXT: vandps %ymm2, %ymm0, %ymm0
|
|
; X64-AVX-NEXT: vandps %ymm2, %ymm1, %ymm1
|
|
; X64-AVX-NEXT: retq
|
|
;
|
|
; X64-AVX512VL-LABEL: fabs_v8f64:
|
|
; X64-AVX512VL: # %bb.0:
|
|
; X64-AVX512VL-NEXT: vpandq {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to8}, %zmm0, %zmm0
|
|
; X64-AVX512VL-NEXT: retq
|
|
;
|
|
; X64-AVX512VLDQ-LABEL: fabs_v8f64:
|
|
; X64-AVX512VLDQ: # %bb.0:
|
|
; X64-AVX512VLDQ-NEXT: vandpd {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to8}, %zmm0, %zmm0
|
|
; X64-AVX512VLDQ-NEXT: retq
|
|
%t = call <8 x double> @llvm.fabs.v8f64(<8 x double> %p)
|
|
ret <8 x double> %t
|
|
}
|
|
declare <8 x double> @llvm.fabs.v8f64(<8 x double> %p)
|
|
|
|
define <16 x float> @fabs_v16f32(<16 x float> %p) {
|
|
; X86-AVX-LABEL: fabs_v16f32:
|
|
; X86-AVX: # %bb.0:
|
|
; X86-AVX-NEXT: vbroadcastss {{.*#+}} ymm2 = [NaN,NaN,NaN,NaN,NaN,NaN,NaN,NaN]
|
|
; X86-AVX-NEXT: vandps %ymm2, %ymm0, %ymm0
|
|
; X86-AVX-NEXT: vandps %ymm2, %ymm1, %ymm1
|
|
; X86-AVX-NEXT: retl
|
|
;
|
|
; X86-AVX512VL-LABEL: fabs_v16f32:
|
|
; X86-AVX512VL: # %bb.0:
|
|
; X86-AVX512VL-NEXT: vpandd {{\.?LCPI[0-9]+_[0-9]+}}{1to16}, %zmm0, %zmm0
|
|
; X86-AVX512VL-NEXT: retl
|
|
;
|
|
; X86-AVX512VLDQ-LABEL: fabs_v16f32:
|
|
; X86-AVX512VLDQ: # %bb.0:
|
|
; X86-AVX512VLDQ-NEXT: vandps {{\.?LCPI[0-9]+_[0-9]+}}{1to16}, %zmm0, %zmm0
|
|
; X86-AVX512VLDQ-NEXT: retl
|
|
;
|
|
; X64-AVX-LABEL: fabs_v16f32:
|
|
; X64-AVX: # %bb.0:
|
|
; X64-AVX-NEXT: vbroadcastss {{.*#+}} ymm2 = [NaN,NaN,NaN,NaN,NaN,NaN,NaN,NaN]
|
|
; X64-AVX-NEXT: vandps %ymm2, %ymm0, %ymm0
|
|
; X64-AVX-NEXT: vandps %ymm2, %ymm1, %ymm1
|
|
; X64-AVX-NEXT: retq
|
|
;
|
|
; X64-AVX512VL-LABEL: fabs_v16f32:
|
|
; X64-AVX512VL: # %bb.0:
|
|
; X64-AVX512VL-NEXT: vpandd {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to16}, %zmm0, %zmm0
|
|
; X64-AVX512VL-NEXT: retq
|
|
;
|
|
; X64-AVX512VLDQ-LABEL: fabs_v16f32:
|
|
; X64-AVX512VLDQ: # %bb.0:
|
|
; X64-AVX512VLDQ-NEXT: vandps {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to16}, %zmm0, %zmm0
|
|
; X64-AVX512VLDQ-NEXT: retq
|
|
%t = call <16 x float> @llvm.fabs.v16f32(<16 x float> %p)
|
|
ret <16 x float> %t
|
|
}
|
|
declare <16 x float> @llvm.fabs.v16f32(<16 x float> %p)
|
|
|
|
; PR20354: when generating code for a vector fabs op,
|
|
; make sure that we're only turning off the sign bit of each float value.
|
|
; No constant pool loads or vector ops are needed for the fabs of a
|
|
; bitcasted integer constant; we should just return an integer constant
|
|
; that has the sign bits turned off.
|
|
;
|
|
; So instead of something like this:
|
|
; movabsq (constant pool load of mask for sign bits)
|
|
; vmovq (move from integer register to vector/fp register)
|
|
; vandps (mask off sign bits)
|
|
; vmovq (move vector/fp register back to integer return register)
|
|
;
|
|
; We should generate:
|
|
; mov (put constant value in return register)
|
|
|
|
define i64 @fabs_v2f32_1() {
|
|
; X86-LABEL: fabs_v2f32_1:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: xorl %eax, %eax
|
|
; X86-NEXT: movl $2147483647, %edx # imm = 0x7FFFFFFF
|
|
; X86-NEXT: retl
|
|
;
|
|
; X64-LABEL: fabs_v2f32_1:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: movabsq $9223372032559808512, %rax # imm = 0x7FFFFFFF00000000
|
|
; X64-NEXT: retq
|
|
%bitcast = bitcast i64 18446744069414584320 to <2 x float> ; 0xFFFF_FFFF_0000_0000
|
|
%fabs = call <2 x float> @llvm.fabs.v2f32(<2 x float> %bitcast)
|
|
%ret = bitcast <2 x float> %fabs to i64
|
|
ret i64 %ret
|
|
}
|
|
|
|
define i64 @fabs_v2f32_2() {
|
|
; X86-LABEL: fabs_v2f32_2:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: movl $2147483647, %eax # imm = 0x7FFFFFFF
|
|
; X86-NEXT: xorl %edx, %edx
|
|
; X86-NEXT: retl
|
|
;
|
|
; X64-LABEL: fabs_v2f32_2:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: movl $2147483647, %eax # imm = 0x7FFFFFFF
|
|
; X64-NEXT: retq
|
|
%bitcast = bitcast i64 4294967295 to <2 x float> ; 0x0000_0000_FFFF_FFFF
|
|
%fabs = call <2 x float> @llvm.fabs.v2f32(<2 x float> %bitcast)
|
|
%ret = bitcast <2 x float> %fabs to i64
|
|
ret i64 %ret
|
|
}
|
|
|
|
declare <2 x float> @llvm.fabs.v2f32(<2 x float> %p)
|