coffee/libyuv
1
0
Fork 0
mirror of https://chromium.googlesource.com/libyuv/libyuv synced 2025-12-06 16:56:55 +08:00
Code Issues Packages Projects Releases Wiki Activity

Port HammingDistance_NEON 32 bit code to 64 bit

Browse Source 
The 32 bit version of HammingDistance_NEON accumulates
using vertical add and paired adds, which takes 3 instructions
instead of 4.
The instructions are also portable between 32 and 64 bit.

Was BenchmarkHammingDistance_Opt (105 ms)
Now BenchmarkHammingDistance_Opt (90 ms)

TBR=kjellander@chromium.org
BUG=libyuv:701
TEST=BenchmarkHammingDistance

BenchmarkHammingDistance_Opt (90 ms)

Change-Id: If9e621e0bd2fe2492a1532056f8a1b451ba53d7e
Reviewed-on: https://chromium-review.googlesource.com/526365
Reviewed-by: Frank Barchard <fbarchard@google.com>
Commit-Queue: Frank Barchard <fbarchard@google.com>
This commit is contained in:
Frank Barchard 2017-06-06 17:50:32 -07:00
parent 47d6eaa377
commit 790e0634a8
1 changed files with 7 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
source/compare_neon64.cc
View File

@ -24,24 +24,22 @@ extern "C" {
uint32 HammingDistance_NEON(const uint8* src_a, const uint8* src_b, int count) {
uint32 diff;
asm volatile (
"movi d4, #0 \n"
"movi v4.4s, #0 \n"
"1: \n"
MEMACCESS(0)
"ld1 {v0.16b, v1.16b}, [%0], #32 \n"
MEMACCESS(1)
"ld1 {v2.16b, v3.16b}, [%1], #32 \n"
"subs %w2, %w2, #32 \n"
"eor v0.16b, v0.16b, v2.16b \n"
"eor v1.16b, v1.16b, v3.16b \n"
"eor v1.16b, v1.16b, v3.16b \n"
"cnt v0.16b, v0.16b \n"
"cnt v1.16b, v1.16b \n"
"uaddlv h0, v0.16b \n"
"uaddlv h1, v1.16b \n"
"add d4, d4, d0 \n"
"add d4, d4, d1 \n"
"subs %w2, %w2, #32 \n"
"add v0.16b, v0.16b, v1.16b \n"
"uaddlp v0.8h, v0.16b \n"
"uadalp v4.4s, v0.8h \n"
"b.gt 1b \n"
"addv s4, v4.4s \n"
"fmov %w3, s4 \n"
: "+r"(src_a),
"+r"(src_b),