578 lines
26 KiB
LLVM
578 lines
26 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc < %s -disable-peephole -mcpu=core-avx2 -show-mc-encoding | FileCheck %s --check-prefix=AVX2
|
|
; RUN: llc < %s -disable-peephole -mcpu=skx -show-mc-encoding | FileCheck %s --check-prefix=AVX512
|
|
|
|
target triple = "x86_64-unknown-unknown"
|
|
|
|
declare <4 x float> @llvm.x86.fma.vfmadd.ss(<4 x float>, <4 x float>, <4 x float>)
|
|
declare <4 x float> @llvm.x86.fma.vfmsub.ss(<4 x float>, <4 x float>, <4 x float>)
|
|
declare <4 x float> @llvm.x86.fma.vfnmadd.ss(<4 x float>, <4 x float>, <4 x float>)
|
|
declare <4 x float> @llvm.x86.fma.vfnmsub.ss(<4 x float>, <4 x float>, <4 x float>)
|
|
|
|
declare <2 x double> @llvm.x86.fma.vfmadd.sd(<2 x double>, <2 x double>, <2 x double>)
|
|
declare <2 x double> @llvm.x86.fma.vfmsub.sd(<2 x double>, <2 x double>, <2 x double>)
|
|
declare <2 x double> @llvm.x86.fma.vfnmadd.sd(<2 x double>, <2 x double>, <2 x double>)
|
|
declare <2 x double> @llvm.x86.fma.vfnmsub.sd(<2 x double>, <2 x double>, <2 x double>)
|
|
|
|
define void @fmadd_aab_ss(float* %a, float* %b) {
|
|
; AVX2-LABEL: fmadd_aab_ss:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vmovss (%rdi), %xmm0 # encoding: [0xc5,0xfa,0x10,0x07]
|
|
; AVX2-NEXT: # xmm0 = mem[0],zero,zero,zero
|
|
; AVX2-NEXT: vfmadd213ss (%rsi), %xmm0, %xmm0 # encoding: [0xc4,0xe2,0x79,0xa9,0x06]
|
|
; AVX2-NEXT: # xmm0 = (xmm0 * xmm0) + mem
|
|
; AVX2-NEXT: vmovss %xmm0, (%rdi) # encoding: [0xc5,0xfa,0x11,0x07]
|
|
; AVX2-NEXT: retq # encoding: [0xc3]
|
|
;
|
|
; AVX512-LABEL: fmadd_aab_ss:
|
|
; AVX512: # %bb.0:
|
|
; AVX512-NEXT: vmovss (%rdi), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x10,0x07]
|
|
; AVX512-NEXT: # xmm0 = mem[0],zero,zero,zero
|
|
; AVX512-NEXT: vfmadd213ss (%rsi), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0xa9,0x06]
|
|
; AVX512-NEXT: # xmm0 = (xmm0 * xmm0) + mem
|
|
; AVX512-NEXT: vmovss %xmm0, (%rdi) # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x11,0x07]
|
|
; AVX512-NEXT: retq # encoding: [0xc3]
|
|
%a.val = load float, float* %a
|
|
%av0 = insertelement <4 x float> undef, float %a.val, i32 0
|
|
%av1 = insertelement <4 x float> %av0, float 0.000000e+00, i32 1
|
|
%av2 = insertelement <4 x float> %av1, float 0.000000e+00, i32 2
|
|
%av = insertelement <4 x float> %av2, float 0.000000e+00, i32 3
|
|
|
|
%b.val = load float, float* %b
|
|
%bv0 = insertelement <4 x float> undef, float %b.val, i32 0
|
|
%bv1 = insertelement <4 x float> %bv0, float 0.000000e+00, i32 1
|
|
%bv2 = insertelement <4 x float> %bv1, float 0.000000e+00, i32 2
|
|
%bv = insertelement <4 x float> %bv2, float 0.000000e+00, i32 3
|
|
|
|
%vr = call <4 x float> @llvm.x86.fma.vfmadd.ss(<4 x float> %av, <4 x float> %av, <4 x float> %bv)
|
|
|
|
%sr = extractelement <4 x float> %vr, i32 0
|
|
store float %sr, float* %a
|
|
ret void
|
|
}
|
|
|
|
define void @fmadd_aba_ss(float* %a, float* %b) {
|
|
; AVX2-LABEL: fmadd_aba_ss:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vmovss (%rdi), %xmm0 # encoding: [0xc5,0xfa,0x10,0x07]
|
|
; AVX2-NEXT: # xmm0 = mem[0],zero,zero,zero
|
|
; AVX2-NEXT: vfmadd231ss (%rsi), %xmm0, %xmm0 # encoding: [0xc4,0xe2,0x79,0xb9,0x06]
|
|
; AVX2-NEXT: # xmm0 = (xmm0 * mem) + xmm0
|
|
; AVX2-NEXT: vmovss %xmm0, (%rdi) # encoding: [0xc5,0xfa,0x11,0x07]
|
|
; AVX2-NEXT: retq # encoding: [0xc3]
|
|
;
|
|
; AVX512-LABEL: fmadd_aba_ss:
|
|
; AVX512: # %bb.0:
|
|
; AVX512-NEXT: vmovss (%rdi), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x10,0x07]
|
|
; AVX512-NEXT: # xmm0 = mem[0],zero,zero,zero
|
|
; AVX512-NEXT: vfmadd231ss (%rsi), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0xb9,0x06]
|
|
; AVX512-NEXT: # xmm0 = (xmm0 * mem) + xmm0
|
|
; AVX512-NEXT: vmovss %xmm0, (%rdi) # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x11,0x07]
|
|
; AVX512-NEXT: retq # encoding: [0xc3]
|
|
%a.val = load float, float* %a
|
|
%av0 = insertelement <4 x float> undef, float %a.val, i32 0
|
|
%av1 = insertelement <4 x float> %av0, float 0.000000e+00, i32 1
|
|
%av2 = insertelement <4 x float> %av1, float 0.000000e+00, i32 2
|
|
%av = insertelement <4 x float> %av2, float 0.000000e+00, i32 3
|
|
|
|
%b.val = load float, float* %b
|
|
%bv0 = insertelement <4 x float> undef, float %b.val, i32 0
|
|
%bv1 = insertelement <4 x float> %bv0, float 0.000000e+00, i32 1
|
|
%bv2 = insertelement <4 x float> %bv1, float 0.000000e+00, i32 2
|
|
%bv = insertelement <4 x float> %bv2, float 0.000000e+00, i32 3
|
|
|
|
%vr = call <4 x float> @llvm.x86.fma.vfmadd.ss(<4 x float> %av, <4 x float> %bv, <4 x float> %av)
|
|
|
|
%sr = extractelement <4 x float> %vr, i32 0
|
|
store float %sr, float* %a
|
|
ret void
|
|
}
|
|
|
|
define void @fmsub_aab_ss(float* %a, float* %b) {
|
|
; AVX2-LABEL: fmsub_aab_ss:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vmovss (%rdi), %xmm0 # encoding: [0xc5,0xfa,0x10,0x07]
|
|
; AVX2-NEXT: # xmm0 = mem[0],zero,zero,zero
|
|
; AVX2-NEXT: vfmsub213ss (%rsi), %xmm0, %xmm0 # encoding: [0xc4,0xe2,0x79,0xab,0x06]
|
|
; AVX2-NEXT: # xmm0 = (xmm0 * xmm0) - mem
|
|
; AVX2-NEXT: vmovss %xmm0, (%rdi) # encoding: [0xc5,0xfa,0x11,0x07]
|
|
; AVX2-NEXT: retq # encoding: [0xc3]
|
|
;
|
|
; AVX512-LABEL: fmsub_aab_ss:
|
|
; AVX512: # %bb.0:
|
|
; AVX512-NEXT: vmovss (%rdi), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x10,0x07]
|
|
; AVX512-NEXT: # xmm0 = mem[0],zero,zero,zero
|
|
; AVX512-NEXT: vfmsub213ss (%rsi), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0xab,0x06]
|
|
; AVX512-NEXT: # xmm0 = (xmm0 * xmm0) - mem
|
|
; AVX512-NEXT: vmovss %xmm0, (%rdi) # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x11,0x07]
|
|
; AVX512-NEXT: retq # encoding: [0xc3]
|
|
%a.val = load float, float* %a
|
|
%av0 = insertelement <4 x float> undef, float %a.val, i32 0
|
|
%av1 = insertelement <4 x float> %av0, float 0.000000e+00, i32 1
|
|
%av2 = insertelement <4 x float> %av1, float 0.000000e+00, i32 2
|
|
%av = insertelement <4 x float> %av2, float 0.000000e+00, i32 3
|
|
|
|
%b.val = load float, float* %b
|
|
%bv0 = insertelement <4 x float> undef, float %b.val, i32 0
|
|
%bv1 = insertelement <4 x float> %bv0, float 0.000000e+00, i32 1
|
|
%bv2 = insertelement <4 x float> %bv1, float 0.000000e+00, i32 2
|
|
%bv = insertelement <4 x float> %bv2, float 0.000000e+00, i32 3
|
|
|
|
%vr = call <4 x float> @llvm.x86.fma.vfmsub.ss(<4 x float> %av, <4 x float> %av, <4 x float> %bv)
|
|
|
|
%sr = extractelement <4 x float> %vr, i32 0
|
|
store float %sr, float* %a
|
|
ret void
|
|
}
|
|
|
|
define void @fmsub_aba_ss(float* %a, float* %b) {
|
|
; AVX2-LABEL: fmsub_aba_ss:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vmovss (%rdi), %xmm0 # encoding: [0xc5,0xfa,0x10,0x07]
|
|
; AVX2-NEXT: # xmm0 = mem[0],zero,zero,zero
|
|
; AVX2-NEXT: vfmsub231ss (%rsi), %xmm0, %xmm0 # encoding: [0xc4,0xe2,0x79,0xbb,0x06]
|
|
; AVX2-NEXT: # xmm0 = (xmm0 * mem) - xmm0
|
|
; AVX2-NEXT: vmovss %xmm0, (%rdi) # encoding: [0xc5,0xfa,0x11,0x07]
|
|
; AVX2-NEXT: retq # encoding: [0xc3]
|
|
;
|
|
; AVX512-LABEL: fmsub_aba_ss:
|
|
; AVX512: # %bb.0:
|
|
; AVX512-NEXT: vmovss (%rdi), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x10,0x07]
|
|
; AVX512-NEXT: # xmm0 = mem[0],zero,zero,zero
|
|
; AVX512-NEXT: vfmsub231ss (%rsi), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0xbb,0x06]
|
|
; AVX512-NEXT: # xmm0 = (xmm0 * mem) - xmm0
|
|
; AVX512-NEXT: vmovss %xmm0, (%rdi) # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x11,0x07]
|
|
; AVX512-NEXT: retq # encoding: [0xc3]
|
|
%a.val = load float, float* %a
|
|
%av0 = insertelement <4 x float> undef, float %a.val, i32 0
|
|
%av1 = insertelement <4 x float> %av0, float 0.000000e+00, i32 1
|
|
%av2 = insertelement <4 x float> %av1, float 0.000000e+00, i32 2
|
|
%av = insertelement <4 x float> %av2, float 0.000000e+00, i32 3
|
|
|
|
%b.val = load float, float* %b
|
|
%bv0 = insertelement <4 x float> undef, float %b.val, i32 0
|
|
%bv1 = insertelement <4 x float> %bv0, float 0.000000e+00, i32 1
|
|
%bv2 = insertelement <4 x float> %bv1, float 0.000000e+00, i32 2
|
|
%bv = insertelement <4 x float> %bv2, float 0.000000e+00, i32 3
|
|
|
|
%vr = call <4 x float> @llvm.x86.fma.vfmsub.ss(<4 x float> %av, <4 x float> %bv, <4 x float> %av)
|
|
|
|
%sr = extractelement <4 x float> %vr, i32 0
|
|
store float %sr, float* %a
|
|
ret void
|
|
}
|
|
|
|
define void @fnmadd_aab_ss(float* %a, float* %b) {
|
|
; AVX2-LABEL: fnmadd_aab_ss:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vmovss (%rdi), %xmm0 # encoding: [0xc5,0xfa,0x10,0x07]
|
|
; AVX2-NEXT: # xmm0 = mem[0],zero,zero,zero
|
|
; AVX2-NEXT: vfnmadd213ss (%rsi), %xmm0, %xmm0 # encoding: [0xc4,0xe2,0x79,0xad,0x06]
|
|
; AVX2-NEXT: # xmm0 = -(xmm0 * xmm0) + mem
|
|
; AVX2-NEXT: vmovss %xmm0, (%rdi) # encoding: [0xc5,0xfa,0x11,0x07]
|
|
; AVX2-NEXT: retq # encoding: [0xc3]
|
|
;
|
|
; AVX512-LABEL: fnmadd_aab_ss:
|
|
; AVX512: # %bb.0:
|
|
; AVX512-NEXT: vmovss (%rdi), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x10,0x07]
|
|
; AVX512-NEXT: # xmm0 = mem[0],zero,zero,zero
|
|
; AVX512-NEXT: vfnmadd213ss (%rsi), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0xad,0x06]
|
|
; AVX512-NEXT: # xmm0 = -(xmm0 * xmm0) + mem
|
|
; AVX512-NEXT: vmovss %xmm0, (%rdi) # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x11,0x07]
|
|
; AVX512-NEXT: retq # encoding: [0xc3]
|
|
%a.val = load float, float* %a
|
|
%av0 = insertelement <4 x float> undef, float %a.val, i32 0
|
|
%av1 = insertelement <4 x float> %av0, float 0.000000e+00, i32 1
|
|
%av2 = insertelement <4 x float> %av1, float 0.000000e+00, i32 2
|
|
%av = insertelement <4 x float> %av2, float 0.000000e+00, i32 3
|
|
|
|
%b.val = load float, float* %b
|
|
%bv0 = insertelement <4 x float> undef, float %b.val, i32 0
|
|
%bv1 = insertelement <4 x float> %bv0, float 0.000000e+00, i32 1
|
|
%bv2 = insertelement <4 x float> %bv1, float 0.000000e+00, i32 2
|
|
%bv = insertelement <4 x float> %bv2, float 0.000000e+00, i32 3
|
|
|
|
%vr = call <4 x float> @llvm.x86.fma.vfnmadd.ss(<4 x float> %av, <4 x float> %av, <4 x float> %bv)
|
|
|
|
%sr = extractelement <4 x float> %vr, i32 0
|
|
store float %sr, float* %a
|
|
ret void
|
|
}
|
|
|
|
define void @fnmadd_aba_ss(float* %a, float* %b) {
|
|
; AVX2-LABEL: fnmadd_aba_ss:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vmovss (%rdi), %xmm0 # encoding: [0xc5,0xfa,0x10,0x07]
|
|
; AVX2-NEXT: # xmm0 = mem[0],zero,zero,zero
|
|
; AVX2-NEXT: vfnmadd231ss (%rsi), %xmm0, %xmm0 # encoding: [0xc4,0xe2,0x79,0xbd,0x06]
|
|
; AVX2-NEXT: # xmm0 = -(xmm0 * mem) + xmm0
|
|
; AVX2-NEXT: vmovss %xmm0, (%rdi) # encoding: [0xc5,0xfa,0x11,0x07]
|
|
; AVX2-NEXT: retq # encoding: [0xc3]
|
|
;
|
|
; AVX512-LABEL: fnmadd_aba_ss:
|
|
; AVX512: # %bb.0:
|
|
; AVX512-NEXT: vmovss (%rdi), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x10,0x07]
|
|
; AVX512-NEXT: # xmm0 = mem[0],zero,zero,zero
|
|
; AVX512-NEXT: vfnmadd231ss (%rsi), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0xbd,0x06]
|
|
; AVX512-NEXT: # xmm0 = -(xmm0 * mem) + xmm0
|
|
; AVX512-NEXT: vmovss %xmm0, (%rdi) # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x11,0x07]
|
|
; AVX512-NEXT: retq # encoding: [0xc3]
|
|
%a.val = load float, float* %a
|
|
%av0 = insertelement <4 x float> undef, float %a.val, i32 0
|
|
%av1 = insertelement <4 x float> %av0, float 0.000000e+00, i32 1
|
|
%av2 = insertelement <4 x float> %av1, float 0.000000e+00, i32 2
|
|
%av = insertelement <4 x float> %av2, float 0.000000e+00, i32 3
|
|
|
|
%b.val = load float, float* %b
|
|
%bv0 = insertelement <4 x float> undef, float %b.val, i32 0
|
|
%bv1 = insertelement <4 x float> %bv0, float 0.000000e+00, i32 1
|
|
%bv2 = insertelement <4 x float> %bv1, float 0.000000e+00, i32 2
|
|
%bv = insertelement <4 x float> %bv2, float 0.000000e+00, i32 3
|
|
|
|
%vr = call <4 x float> @llvm.x86.fma.vfnmadd.ss(<4 x float> %av, <4 x float> %bv, <4 x float> %av)
|
|
|
|
%sr = extractelement <4 x float> %vr, i32 0
|
|
store float %sr, float* %a
|
|
ret void
|
|
}
|
|
|
|
define void @fnmsub_aab_ss(float* %a, float* %b) {
|
|
; AVX2-LABEL: fnmsub_aab_ss:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vmovss (%rdi), %xmm0 # encoding: [0xc5,0xfa,0x10,0x07]
|
|
; AVX2-NEXT: # xmm0 = mem[0],zero,zero,zero
|
|
; AVX2-NEXT: vfnmsub213ss (%rsi), %xmm0, %xmm0 # encoding: [0xc4,0xe2,0x79,0xaf,0x06]
|
|
; AVX2-NEXT: # xmm0 = -(xmm0 * xmm0) - mem
|
|
; AVX2-NEXT: vmovss %xmm0, (%rdi) # encoding: [0xc5,0xfa,0x11,0x07]
|
|
; AVX2-NEXT: retq # encoding: [0xc3]
|
|
;
|
|
; AVX512-LABEL: fnmsub_aab_ss:
|
|
; AVX512: # %bb.0:
|
|
; AVX512-NEXT: vmovss (%rdi), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x10,0x07]
|
|
; AVX512-NEXT: # xmm0 = mem[0],zero,zero,zero
|
|
; AVX512-NEXT: vfnmsub213ss (%rsi), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0xaf,0x06]
|
|
; AVX512-NEXT: # xmm0 = -(xmm0 * xmm0) - mem
|
|
; AVX512-NEXT: vmovss %xmm0, (%rdi) # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x11,0x07]
|
|
; AVX512-NEXT: retq # encoding: [0xc3]
|
|
%a.val = load float, float* %a
|
|
%av0 = insertelement <4 x float> undef, float %a.val, i32 0
|
|
%av1 = insertelement <4 x float> %av0, float 0.000000e+00, i32 1
|
|
%av2 = insertelement <4 x float> %av1, float 0.000000e+00, i32 2
|
|
%av = insertelement <4 x float> %av2, float 0.000000e+00, i32 3
|
|
|
|
%b.val = load float, float* %b
|
|
%bv0 = insertelement <4 x float> undef, float %b.val, i32 0
|
|
%bv1 = insertelement <4 x float> %bv0, float 0.000000e+00, i32 1
|
|
%bv2 = insertelement <4 x float> %bv1, float 0.000000e+00, i32 2
|
|
%bv = insertelement <4 x float> %bv2, float 0.000000e+00, i32 3
|
|
|
|
%vr = call <4 x float> @llvm.x86.fma.vfnmsub.ss(<4 x float> %av, <4 x float> %av, <4 x float> %bv)
|
|
|
|
%sr = extractelement <4 x float> %vr, i32 0
|
|
store float %sr, float* %a
|
|
ret void
|
|
}
|
|
|
|
define void @fnmsub_aba_ss(float* %a, float* %b) {
|
|
; AVX2-LABEL: fnmsub_aba_ss:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vmovss (%rdi), %xmm0 # encoding: [0xc5,0xfa,0x10,0x07]
|
|
; AVX2-NEXT: # xmm0 = mem[0],zero,zero,zero
|
|
; AVX2-NEXT: vfnmsub231ss (%rsi), %xmm0, %xmm0 # encoding: [0xc4,0xe2,0x79,0xbf,0x06]
|
|
; AVX2-NEXT: # xmm0 = -(xmm0 * mem) - xmm0
|
|
; AVX2-NEXT: vmovss %xmm0, (%rdi) # encoding: [0xc5,0xfa,0x11,0x07]
|
|
; AVX2-NEXT: retq # encoding: [0xc3]
|
|
;
|
|
; AVX512-LABEL: fnmsub_aba_ss:
|
|
; AVX512: # %bb.0:
|
|
; AVX512-NEXT: vmovss (%rdi), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x10,0x07]
|
|
; AVX512-NEXT: # xmm0 = mem[0],zero,zero,zero
|
|
; AVX512-NEXT: vfnmsub231ss (%rsi), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0xbf,0x06]
|
|
; AVX512-NEXT: # xmm0 = -(xmm0 * mem) - xmm0
|
|
; AVX512-NEXT: vmovss %xmm0, (%rdi) # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x11,0x07]
|
|
; AVX512-NEXT: retq # encoding: [0xc3]
|
|
%a.val = load float, float* %a
|
|
%av0 = insertelement <4 x float> undef, float %a.val, i32 0
|
|
%av1 = insertelement <4 x float> %av0, float 0.000000e+00, i32 1
|
|
%av2 = insertelement <4 x float> %av1, float 0.000000e+00, i32 2
|
|
%av = insertelement <4 x float> %av2, float 0.000000e+00, i32 3
|
|
|
|
%b.val = load float, float* %b
|
|
%bv0 = insertelement <4 x float> undef, float %b.val, i32 0
|
|
%bv1 = insertelement <4 x float> %bv0, float 0.000000e+00, i32 1
|
|
%bv2 = insertelement <4 x float> %bv1, float 0.000000e+00, i32 2
|
|
%bv = insertelement <4 x float> %bv2, float 0.000000e+00, i32 3
|
|
|
|
%vr = call <4 x float> @llvm.x86.fma.vfnmsub.ss(<4 x float> %av, <4 x float> %bv, <4 x float> %av)
|
|
|
|
%sr = extractelement <4 x float> %vr, i32 0
|
|
store float %sr, float* %a
|
|
ret void
|
|
}
|
|
|
|
define void @fmadd_aab_sd(double* %a, double* %b) {
|
|
; AVX2-LABEL: fmadd_aab_sd:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vmovsd (%rdi), %xmm0 # encoding: [0xc5,0xfb,0x10,0x07]
|
|
; AVX2-NEXT: # xmm0 = mem[0],zero
|
|
; AVX2-NEXT: vfmadd213sd (%rsi), %xmm0, %xmm0 # encoding: [0xc4,0xe2,0xf9,0xa9,0x06]
|
|
; AVX2-NEXT: # xmm0 = (xmm0 * xmm0) + mem
|
|
; AVX2-NEXT: vmovsd %xmm0, (%rdi) # encoding: [0xc5,0xfb,0x11,0x07]
|
|
; AVX2-NEXT: retq # encoding: [0xc3]
|
|
;
|
|
; AVX512-LABEL: fmadd_aab_sd:
|
|
; AVX512: # %bb.0:
|
|
; AVX512-NEXT: vmovsd (%rdi), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfb,0x10,0x07]
|
|
; AVX512-NEXT: # xmm0 = mem[0],zero
|
|
; AVX512-NEXT: vfmadd213sd (%rsi), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0xf9,0xa9,0x06]
|
|
; AVX512-NEXT: # xmm0 = (xmm0 * xmm0) + mem
|
|
; AVX512-NEXT: vmovsd %xmm0, (%rdi) # EVEX TO VEX Compression encoding: [0xc5,0xfb,0x11,0x07]
|
|
; AVX512-NEXT: retq # encoding: [0xc3]
|
|
%a.val = load double, double* %a
|
|
%av0 = insertelement <2 x double> undef, double %a.val, i32 0
|
|
%av = insertelement <2 x double> %av0, double 0.000000e+00, i32 1
|
|
|
|
%b.val = load double, double* %b
|
|
%bv0 = insertelement <2 x double> undef, double %b.val, i32 0
|
|
%bv = insertelement <2 x double> %bv0, double 0.000000e+00, i32 1
|
|
|
|
%vr = call <2 x double> @llvm.x86.fma.vfmadd.sd(<2 x double> %av, <2 x double> %av, <2 x double> %bv)
|
|
|
|
%sr = extractelement <2 x double> %vr, i32 0
|
|
store double %sr, double* %a
|
|
ret void
|
|
}
|
|
|
|
define void @fmadd_aba_sd(double* %a, double* %b) {
|
|
; AVX2-LABEL: fmadd_aba_sd:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vmovsd (%rdi), %xmm0 # encoding: [0xc5,0xfb,0x10,0x07]
|
|
; AVX2-NEXT: # xmm0 = mem[0],zero
|
|
; AVX2-NEXT: vfmadd231sd (%rsi), %xmm0, %xmm0 # encoding: [0xc4,0xe2,0xf9,0xb9,0x06]
|
|
; AVX2-NEXT: # xmm0 = (xmm0 * mem) + xmm0
|
|
; AVX2-NEXT: vmovsd %xmm0, (%rdi) # encoding: [0xc5,0xfb,0x11,0x07]
|
|
; AVX2-NEXT: retq # encoding: [0xc3]
|
|
;
|
|
; AVX512-LABEL: fmadd_aba_sd:
|
|
; AVX512: # %bb.0:
|
|
; AVX512-NEXT: vmovsd (%rdi), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfb,0x10,0x07]
|
|
; AVX512-NEXT: # xmm0 = mem[0],zero
|
|
; AVX512-NEXT: vfmadd231sd (%rsi), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0xf9,0xb9,0x06]
|
|
; AVX512-NEXT: # xmm0 = (xmm0 * mem) + xmm0
|
|
; AVX512-NEXT: vmovsd %xmm0, (%rdi) # EVEX TO VEX Compression encoding: [0xc5,0xfb,0x11,0x07]
|
|
; AVX512-NEXT: retq # encoding: [0xc3]
|
|
%a.val = load double, double* %a
|
|
%av0 = insertelement <2 x double> undef, double %a.val, i32 0
|
|
%av = insertelement <2 x double> %av0, double 0.000000e+00, i32 1
|
|
|
|
%b.val = load double, double* %b
|
|
%bv0 = insertelement <2 x double> undef, double %b.val, i32 0
|
|
%bv = insertelement <2 x double> %bv0, double 0.000000e+00, i32 1
|
|
|
|
%vr = call <2 x double> @llvm.x86.fma.vfmadd.sd(<2 x double> %av, <2 x double> %bv, <2 x double> %av)
|
|
|
|
%sr = extractelement <2 x double> %vr, i32 0
|
|
store double %sr, double* %a
|
|
ret void
|
|
}
|
|
|
|
define void @fmsub_aab_sd(double* %a, double* %b) {
|
|
; AVX2-LABEL: fmsub_aab_sd:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vmovsd (%rdi), %xmm0 # encoding: [0xc5,0xfb,0x10,0x07]
|
|
; AVX2-NEXT: # xmm0 = mem[0],zero
|
|
; AVX2-NEXT: vfmsub213sd (%rsi), %xmm0, %xmm0 # encoding: [0xc4,0xe2,0xf9,0xab,0x06]
|
|
; AVX2-NEXT: # xmm0 = (xmm0 * xmm0) - mem
|
|
; AVX2-NEXT: vmovsd %xmm0, (%rdi) # encoding: [0xc5,0xfb,0x11,0x07]
|
|
; AVX2-NEXT: retq # encoding: [0xc3]
|
|
;
|
|
; AVX512-LABEL: fmsub_aab_sd:
|
|
; AVX512: # %bb.0:
|
|
; AVX512-NEXT: vmovsd (%rdi), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfb,0x10,0x07]
|
|
; AVX512-NEXT: # xmm0 = mem[0],zero
|
|
; AVX512-NEXT: vfmsub213sd (%rsi), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0xf9,0xab,0x06]
|
|
; AVX512-NEXT: # xmm0 = (xmm0 * xmm0) - mem
|
|
; AVX512-NEXT: vmovsd %xmm0, (%rdi) # EVEX TO VEX Compression encoding: [0xc5,0xfb,0x11,0x07]
|
|
; AVX512-NEXT: retq # encoding: [0xc3]
|
|
%a.val = load double, double* %a
|
|
%av0 = insertelement <2 x double> undef, double %a.val, i32 0
|
|
%av = insertelement <2 x double> %av0, double 0.000000e+00, i32 1
|
|
|
|
%b.val = load double, double* %b
|
|
%bv0 = insertelement <2 x double> undef, double %b.val, i32 0
|
|
%bv = insertelement <2 x double> %bv0, double 0.000000e+00, i32 1
|
|
|
|
%vr = call <2 x double> @llvm.x86.fma.vfmsub.sd(<2 x double> %av, <2 x double> %av, <2 x double> %bv)
|
|
|
|
%sr = extractelement <2 x double> %vr, i32 0
|
|
store double %sr, double* %a
|
|
ret void
|
|
}
|
|
|
|
define void @fmsub_aba_sd(double* %a, double* %b) {
|
|
; AVX2-LABEL: fmsub_aba_sd:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vmovsd (%rdi), %xmm0 # encoding: [0xc5,0xfb,0x10,0x07]
|
|
; AVX2-NEXT: # xmm0 = mem[0],zero
|
|
; AVX2-NEXT: vfmsub231sd (%rsi), %xmm0, %xmm0 # encoding: [0xc4,0xe2,0xf9,0xbb,0x06]
|
|
; AVX2-NEXT: # xmm0 = (xmm0 * mem) - xmm0
|
|
; AVX2-NEXT: vmovsd %xmm0, (%rdi) # encoding: [0xc5,0xfb,0x11,0x07]
|
|
; AVX2-NEXT: retq # encoding: [0xc3]
|
|
;
|
|
; AVX512-LABEL: fmsub_aba_sd:
|
|
; AVX512: # %bb.0:
|
|
; AVX512-NEXT: vmovsd (%rdi), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfb,0x10,0x07]
|
|
; AVX512-NEXT: # xmm0 = mem[0],zero
|
|
; AVX512-NEXT: vfmsub231sd (%rsi), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0xf9,0xbb,0x06]
|
|
; AVX512-NEXT: # xmm0 = (xmm0 * mem) - xmm0
|
|
; AVX512-NEXT: vmovsd %xmm0, (%rdi) # EVEX TO VEX Compression encoding: [0xc5,0xfb,0x11,0x07]
|
|
; AVX512-NEXT: retq # encoding: [0xc3]
|
|
%a.val = load double, double* %a
|
|
%av0 = insertelement <2 x double> undef, double %a.val, i32 0
|
|
%av = insertelement <2 x double> %av0, double 0.000000e+00, i32 1
|
|
|
|
%b.val = load double, double* %b
|
|
%bv0 = insertelement <2 x double> undef, double %b.val, i32 0
|
|
%bv = insertelement <2 x double> %bv0, double 0.000000e+00, i32 1
|
|
|
|
%vr = call <2 x double> @llvm.x86.fma.vfmsub.sd(<2 x double> %av, <2 x double> %bv, <2 x double> %av)
|
|
|
|
%sr = extractelement <2 x double> %vr, i32 0
|
|
store double %sr, double* %a
|
|
ret void
|
|
}
|
|
|
|
define void @fnmadd_aab_sd(double* %a, double* %b) {
|
|
; AVX2-LABEL: fnmadd_aab_sd:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vmovsd (%rdi), %xmm0 # encoding: [0xc5,0xfb,0x10,0x07]
|
|
; AVX2-NEXT: # xmm0 = mem[0],zero
|
|
; AVX2-NEXT: vfnmadd213sd (%rsi), %xmm0, %xmm0 # encoding: [0xc4,0xe2,0xf9,0xad,0x06]
|
|
; AVX2-NEXT: # xmm0 = -(xmm0 * xmm0) + mem
|
|
; AVX2-NEXT: vmovsd %xmm0, (%rdi) # encoding: [0xc5,0xfb,0x11,0x07]
|
|
; AVX2-NEXT: retq # encoding: [0xc3]
|
|
;
|
|
; AVX512-LABEL: fnmadd_aab_sd:
|
|
; AVX512: # %bb.0:
|
|
; AVX512-NEXT: vmovsd (%rdi), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfb,0x10,0x07]
|
|
; AVX512-NEXT: # xmm0 = mem[0],zero
|
|
; AVX512-NEXT: vfnmadd213sd (%rsi), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0xf9,0xad,0x06]
|
|
; AVX512-NEXT: # xmm0 = -(xmm0 * xmm0) + mem
|
|
; AVX512-NEXT: vmovsd %xmm0, (%rdi) # EVEX TO VEX Compression encoding: [0xc5,0xfb,0x11,0x07]
|
|
; AVX512-NEXT: retq # encoding: [0xc3]
|
|
%a.val = load double, double* %a
|
|
%av0 = insertelement <2 x double> undef, double %a.val, i32 0
|
|
%av = insertelement <2 x double> %av0, double 0.000000e+00, i32 1
|
|
|
|
%b.val = load double, double* %b
|
|
%bv0 = insertelement <2 x double> undef, double %b.val, i32 0
|
|
%bv = insertelement <2 x double> %bv0, double 0.000000e+00, i32 1
|
|
|
|
%vr = call <2 x double> @llvm.x86.fma.vfnmadd.sd(<2 x double> %av, <2 x double> %av, <2 x double> %bv)
|
|
|
|
%sr = extractelement <2 x double> %vr, i32 0
|
|
store double %sr, double* %a
|
|
ret void
|
|
}
|
|
|
|
define void @fnmadd_aba_sd(double* %a, double* %b) {
|
|
; AVX2-LABEL: fnmadd_aba_sd:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vmovsd (%rdi), %xmm0 # encoding: [0xc5,0xfb,0x10,0x07]
|
|
; AVX2-NEXT: # xmm0 = mem[0],zero
|
|
; AVX2-NEXT: vfnmadd231sd (%rsi), %xmm0, %xmm0 # encoding: [0xc4,0xe2,0xf9,0xbd,0x06]
|
|
; AVX2-NEXT: # xmm0 = -(xmm0 * mem) + xmm0
|
|
; AVX2-NEXT: vmovsd %xmm0, (%rdi) # encoding: [0xc5,0xfb,0x11,0x07]
|
|
; AVX2-NEXT: retq # encoding: [0xc3]
|
|
;
|
|
; AVX512-LABEL: fnmadd_aba_sd:
|
|
; AVX512: # %bb.0:
|
|
; AVX512-NEXT: vmovsd (%rdi), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfb,0x10,0x07]
|
|
; AVX512-NEXT: # xmm0 = mem[0],zero
|
|
; AVX512-NEXT: vfnmadd231sd (%rsi), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0xf9,0xbd,0x06]
|
|
; AVX512-NEXT: # xmm0 = -(xmm0 * mem) + xmm0
|
|
; AVX512-NEXT: vmovsd %xmm0, (%rdi) # EVEX TO VEX Compression encoding: [0xc5,0xfb,0x11,0x07]
|
|
; AVX512-NEXT: retq # encoding: [0xc3]
|
|
%a.val = load double, double* %a
|
|
%av0 = insertelement <2 x double> undef, double %a.val, i32 0
|
|
%av = insertelement <2 x double> %av0, double 0.000000e+00, i32 1
|
|
|
|
%b.val = load double, double* %b
|
|
%bv0 = insertelement <2 x double> undef, double %b.val, i32 0
|
|
%bv = insertelement <2 x double> %bv0, double 0.000000e+00, i32 1
|
|
|
|
%vr = call <2 x double> @llvm.x86.fma.vfnmadd.sd(<2 x double> %av, <2 x double> %bv, <2 x double> %av)
|
|
|
|
%sr = extractelement <2 x double> %vr, i32 0
|
|
store double %sr, double* %a
|
|
ret void
|
|
}
|
|
|
|
define void @fnmsub_aab_sd(double* %a, double* %b) {
|
|
; AVX2-LABEL: fnmsub_aab_sd:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vmovsd (%rdi), %xmm0 # encoding: [0xc5,0xfb,0x10,0x07]
|
|
; AVX2-NEXT: # xmm0 = mem[0],zero
|
|
; AVX2-NEXT: vfnmsub213sd (%rsi), %xmm0, %xmm0 # encoding: [0xc4,0xe2,0xf9,0xaf,0x06]
|
|
; AVX2-NEXT: # xmm0 = -(xmm0 * xmm0) - mem
|
|
; AVX2-NEXT: vmovsd %xmm0, (%rdi) # encoding: [0xc5,0xfb,0x11,0x07]
|
|
; AVX2-NEXT: retq # encoding: [0xc3]
|
|
;
|
|
; AVX512-LABEL: fnmsub_aab_sd:
|
|
; AVX512: # %bb.0:
|
|
; AVX512-NEXT: vmovsd (%rdi), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfb,0x10,0x07]
|
|
; AVX512-NEXT: # xmm0 = mem[0],zero
|
|
; AVX512-NEXT: vfnmsub213sd (%rsi), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0xf9,0xaf,0x06]
|
|
; AVX512-NEXT: # xmm0 = -(xmm0 * xmm0) - mem
|
|
; AVX512-NEXT: vmovsd %xmm0, (%rdi) # EVEX TO VEX Compression encoding: [0xc5,0xfb,0x11,0x07]
|
|
; AVX512-NEXT: retq # encoding: [0xc3]
|
|
%a.val = load double, double* %a
|
|
%av0 = insertelement <2 x double> undef, double %a.val, i32 0
|
|
%av = insertelement <2 x double> %av0, double 0.000000e+00, i32 1
|
|
|
|
%b.val = load double, double* %b
|
|
%bv0 = insertelement <2 x double> undef, double %b.val, i32 0
|
|
%bv = insertelement <2 x double> %bv0, double 0.000000e+00, i32 1
|
|
|
|
%vr = call <2 x double> @llvm.x86.fma.vfnmsub.sd(<2 x double> %av, <2 x double> %av, <2 x double> %bv)
|
|
|
|
%sr = extractelement <2 x double> %vr, i32 0
|
|
store double %sr, double* %a
|
|
ret void
|
|
}
|
|
|
|
define void @fnmsub_aba_sd(double* %a, double* %b) {
|
|
; AVX2-LABEL: fnmsub_aba_sd:
|
|
; AVX2: # %bb.0:
|
|
; AVX2-NEXT: vmovsd (%rdi), %xmm0 # encoding: [0xc5,0xfb,0x10,0x07]
|
|
; AVX2-NEXT: # xmm0 = mem[0],zero
|
|
; AVX2-NEXT: vfnmsub231sd (%rsi), %xmm0, %xmm0 # encoding: [0xc4,0xe2,0xf9,0xbf,0x06]
|
|
; AVX2-NEXT: # xmm0 = -(xmm0 * mem) - xmm0
|
|
; AVX2-NEXT: vmovsd %xmm0, (%rdi) # encoding: [0xc5,0xfb,0x11,0x07]
|
|
; AVX2-NEXT: retq # encoding: [0xc3]
|
|
;
|
|
; AVX512-LABEL: fnmsub_aba_sd:
|
|
; AVX512: # %bb.0:
|
|
; AVX512-NEXT: vmovsd (%rdi), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfb,0x10,0x07]
|
|
; AVX512-NEXT: # xmm0 = mem[0],zero
|
|
; AVX512-NEXT: vfnmsub231sd (%rsi), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0xf9,0xbf,0x06]
|
|
; AVX512-NEXT: # xmm0 = -(xmm0 * mem) - xmm0
|
|
; AVX512-NEXT: vmovsd %xmm0, (%rdi) # EVEX TO VEX Compression encoding: [0xc5,0xfb,0x11,0x07]
|
|
; AVX512-NEXT: retq # encoding: [0xc3]
|
|
%a.val = load double, double* %a
|
|
%av0 = insertelement <2 x double> undef, double %a.val, i32 0
|
|
%av = insertelement <2 x double> %av0, double 0.000000e+00, i32 1
|
|
|
|
%b.val = load double, double* %b
|
|
%bv0 = insertelement <2 x double> undef, double %b.val, i32 0
|
|
%bv = insertelement <2 x double> %bv0, double 0.000000e+00, i32 1
|
|
|
|
%vr = call <2 x double> @llvm.x86.fma.vfnmsub.sd(<2 x double> %av, <2 x double> %bv, <2 x double> %av)
|
|
|
|
%sr = extractelement <2 x double> %vr, i32 0
|
|
store double %sr, double* %a
|
|
ret void
|
|
}
|
|
|
|
|