fumiama/blake2b-simd
1
0
Fork 0
mirror of https://github.com/fumiama/blake2b-simd.git synced 2026-06-19 10:26:45 +08:00
Code Activity

Small optimization of using VPALINGR instead of 2 VPUNPCK?QDQ's

Browse Source
This commit is contained in:
frankw
2016-06-26 23:55:49 +02:00
parent b7b013f822
commit 95d67c9edb
1 changed files with 2 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
compress_amd64.s
View File

@@ -299,8 +299,7 @@ TEXT ·compressSSE(SB), 7, $0
MOVOU 0(DX), X12 // X12 = m[0]+ m[1]
MOVOU 32(DX), X13 // X13 = m[4]+ m[5]
MOVOU 80(DX), X14 // X14 = m[10]+m[11]
BYTE $0xc4; BYTE $0x41; BYTE $0x19; BYTE $0x6d; BYTE $0xc4 // VPUNPCKHQDQ XMM8, XMM12, XMM12 /* m[1], ___ */
BYTE $0xc4; BYTE $0x41; BYTE $0x39; BYTE $0x6c; BYTE $0xc4 // VPUNPCKLQDQ XMM8, XMM8, XMM12 /* ___, m[0] */
BYTE $0xc4; BYTE $0x43; BYTE $0x19; BYTE $0x0f; BYTE $0xc4; BYTE $0x08 // VPALIGNR XMM8, XMM12, XMM12, 0x8 /* m[1], m[0] */
BYTE $0xc4; BYTE $0x41; BYTE $0x09; BYTE $0x6d; BYTE $0xcd // VPUNPCKHQDQ XMM9, XMM14, XMM13 /* m[11], m[5] */
MOVOU 16(DX), X12 // X12 = m[2]+ m[3]
MOVOU 48(DX), X13 // X13 = m[6]+ m[7]
@@ -744,8 +743,7 @@ TEXT ·compressSSE(SB), 7, $0
MOVOU 0(DX), X12 // X12 = m[0]+ m[1]
MOVOU 32(DX), X13 // X13 = m[4]+ m[5]
MOVOU 80(DX), X14 // X14 = m[10]+m[11]
BYTE $0xc4; BYTE $0x41; BYTE $0x19; BYTE $0x6d; BYTE $0xc4 // VPUNPCKHQDQ XMM8, XMM12, XMM12 /* m[1], ___ */
BYTE $0xc4; BYTE $0x41; BYTE $0x39; BYTE $0x6c; BYTE $0xc4 // VPUNPCKLQDQ XMM8, XMM8, XMM12 /* ___, m[0] */
BYTE $0xc4; BYTE $0x43; BYTE $0x19; BYTE $0x0f; BYTE $0xc4; BYTE $0x08 // VPALIGNR XMM8, XMM12, XMM12, 0x8 /* m[1], m[0] */
BYTE $0xc4; BYTE $0x41; BYTE $0x09; BYTE $0x6d; BYTE $0xcd // VPUNPCKHQDQ XMM9, XMM14, XMM13 /* m[11], m[5] */
MOVOU 16(DX), X12 // X12 = m[2]+ m[3]
MOVOU 48(DX), X13 // X13 = m[6]+ m[7]