TrueCloudLab/tzhash
15
0
Fork 6
Code Issues 3 Pull requests Releases 2 Activity

Use unaligned move in AVX2 implementation

Browse source 
Signed-off-by: Evgenii Stratonikov <evgeniy@nspcc.ru>
This commit is contained in:
Evgenii Stratonikov 2021-12-29 09:48:28 +03:00 committed by Alex Vanin
parent c8a32b25ec
commit bbbcf3fa5c
1 changed files with 9 additions and 9 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

18
tz/avx2_amd64.s
View file

@ -18,14 +18,14 @@
VPXOR Y3, Y8, Y3 \
VPAND Y3, Y2, Y4 \
VPXOR Y4, Y0, Y8 \
VMOVDQA Y3, Y0
VMOVDQU Y3, Y0
// func mulByteRightx2(c00c10, c01c11 *[4]uint64, b byte)
TEXT ·mulByteRightx2(SB),NOSPLIT,$0
MOVQ c00c10+0(FP), AX
VMOVDQA (AX), Y0
VMOVDQU (AX), Y0
MOVQ c01c11+8(FP), BX
VMOVDQA (BX), Y8
VMOVDQU (BX), Y8
VPXOR Y13, Y13, Y13 // Y13 = 0x0000...
VPCMPEQB Y12, Y12, Y12 // Y12 = 0xFFFF...
@ -43,17 +43,17 @@ TEXT ·mulByteRightx2(SB),NOSPLIT,$0
mulBit($1)
mulBit($0)
VMOVDQA Y8, (BX)
VMOVDQA Y0, (AX)
VMOVDQU Y8, (BX)
VMOVDQU Y0, (AX)
RET
// func mulBitRightx2(c00c10, c01c11 *[4]uint64, e *[2]uint64)
TEXT ·mulBitRightx2(SB),NOSPLIT,$0
MOVQ c00c10+0(FP), AX
VMOVDQA (AX), Y0
VMOVDQU (AX), Y0
MOVQ c01c11+8(FP), BX
VMOVDQA (BX), Y8
VMOVDQU (BX), Y8
VPSLLQ $1, Y0, Y1
VPALIGNR $8, Y1, Y0, Y2
@ -70,6 +70,6 @@ TEXT ·mulBitRightx2(SB),NOSPLIT,$0
VPXOR Y3, Y8, Y3
VPAND Y3, Y2, Y4
VPXOR Y4, Y0, Y8
VMOVDQA Y8, (BX)
VMOVDQA Y3, (AX)
VMOVDQU Y8, (BX)
VMOVDQU Y3, (AX)
RET