On Tue, Feb 10, 2015 at 12:44:23AM +0530, supraja reddy wrote: > Hello, > > I have attached a patch to optimize the camellia block cipher. > I ran crypto_bench on the optimized version. Here are the results of > runtime using crypto_bench. > > Before the changes : > lavu CAMELLIA size: 1048576 runs: 1024 time: 32.541 +- 0.044 > > After the changes: > lavu CAMELLIA size: 1048576 runs: 1024 time: 24.589 +- 0.066 > > I ran this on a Linux x86_64 OS with Intel Core i5-3210M CPU. > > Please let me know if there are any changes you would like. > > > Thanks, > > Supraja
> camellia.c | 37 +++++++++++++++++++++++++++----------
> 1 file changed, 27 insertions(+), 10 deletions(-)
> 0aecabcca7c8404e628afc26ff632314d01f2c1c camopt1.patch
> From 9079c80bd5caa122ea1261b47e1fcb0e7d554af5 Mon Sep 17 00:00:00 2001
> From: Supraja Meedinti <supraja0...@gmail.com>
> Date: Tue, 10 Feb 2015 00:16:55 +0530
> Subject: [PATCH] libavutil: optimize camellia cipher
applied
[...]
> @@ -140,17 +142,11 @@ static void LR128(uint64_t d[2], const uint64_t K[2],
> int x)
>
> static uint64_t F(uint64_t F_IN, uint64_t KE)
> {
> - uint32_t Zl, Zr;
> + uint8_t y[8];
> KE ^= F_IN;
[...]
> + AV_WB64(y,KE);
> +
> KE=SP[0][y[0]]^SP[1][y[1]]^SP[2][y[2]]^SP[3][y[3]]^SP[4][y[4]]^SP[5][y[5]]^SP[6][y[6]]^SP[7][y[7]];
> + return KE;
have you tried it without the y[] array ?
storing and loading from memory may (or may not) be slower
thanks
[...]
--
Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB
The real ebay dictionary, page 3
"Rare item" - "Common item with rare defect or maybe just a lie"
"Professional" - "'Toy' made in china, not functional except as doorstop"
"Experts will know" - "The seller hopes you are not an expert"
signature.asc
Description: Digital signature
_______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
