diff options
author | Jussi Kivilinna <jussi.kivilinna@iki.fi> | 2013-12-17 15:35:38 +0200 |
---|---|---|
committer | Jussi Kivilinna <jussi.kivilinna@iki.fi> | 2013-12-18 17:00:24 +0200 |
commit | df629ba53a662427ebd3ddca90c3fe9ddd6511d3 (patch) | |
tree | 0e383e4186907a3f607343c4ef39116118bc363f /cipher/sha512.c | |
parent | a5c2bbfe0db515d739ab683297903c77b1eec124 (diff) | |
download | libgcrypt-df629ba53a662427ebd3ddca90c3fe9ddd6511d3.tar.gz |
Improve performance of SHA-512/ARM/NEON implementation
* cipher/sha512-armv7-neon.S (RT01q, RT23q, RT45q, RT67q): New.
(round_0_63, round_64_79): Remove.
(rounds2_0_63, rounds2_64_79): New.
(_gcry_sha512_transform_armv7_neon): Add 'nblks' input; Handle multiple
input blocks; Use new round macros.
* cipher/sha512.c [USE_ARM_NEON_ASM]
(_gcry_sha512_transform_armv7_neon): Add 'num_blks'.
(transform) [USE_ARM_NEON_ASM]: Pass nblks to assembly.
--
Benchmarks on ARM Cortex-A8:
C-language: 139.1 c/B
Old ARM/NEON: 34.30 c/B
New ARM/NEON: 24.46 c/B
New vs C: 5.68x
New vs Old: 1.40x
Signed-off-by: Jussi Kivilinna <jussi.kivilinna@iki.fi>
Diffstat (limited to 'cipher/sha512.c')
-rw-r--r-- | cipher/sha512.c | 9 |
1 files changed, 2 insertions, 7 deletions
diff --git a/cipher/sha512.c b/cipher/sha512.c index 215e8edf..3474694c 100644 --- a/cipher/sha512.c +++ b/cipher/sha512.c @@ -541,7 +541,7 @@ transform_blk (SHA512_STATE *hd, const unsigned char *data) #ifdef USE_ARM_NEON_ASM void _gcry_sha512_transform_armv7_neon (SHA512_STATE *hd, const unsigned char *data, - const u64 k[]); + const u64 k[], size_t num_blks); #endif #ifdef USE_SSSE3 @@ -587,12 +587,7 @@ transform (void *context, const unsigned char *data, size_t nblks) #ifdef USE_ARM_NEON_ASM if (ctx->use_neon) { - do - { - _gcry_sha512_transform_armv7_neon (&ctx->state, data, k); - data += 128; - } - while (--nblks); + _gcry_sha512_transform_armv7_neon (&ctx->state, data, k, nblks); /* _gcry_sha512_transform_armv7_neon does not store sensitive data * to stack. */ |