diff options
| author | Romain Dolbeau <[email protected]> | 2016-10-21 19:55:49 +0200 |
|---|---|---|
| committer | Brian Behlendorf <[email protected]> | 2016-10-21 10:55:49 -0700 |
| commit | 24cdeaf12e9e546621902449699fc6d664aeac2b (patch) | |
| tree | 12af696407ba15b157d591f44f23de6d5eee2882 /module/zcommon/zfs_fletcher.c | |
| parent | e4ffa98dcaf2208c742609f0ab2bdb343071446a (diff) | |
Fletcher4 algorithm implemented in pure NEON for Aarch64 / ARMv8 64 bits
This is not useful on micro-architecture with a weak NEON
implementation (only 64 bits); the native version is slower &
the byteswap barely faster than scalar. On A53 or A57, it's
a small improvement on scalar but OK for byteswap.
Results from an A53 system:
0 0 0x01 -1 0 1499068294333000 1499101101878000
implementation native byteswap
scalar 1008227510 755880264
aarch64_neon 1198098720 1044818671
fastest aarch64_neon aarch64_neon
Results from a A57 system:
0 0 0x01 -1 0 4407214734807033 4407233933777404
implementation native byteswap
scalar 2302071241 1124873346
aarch64_neon 2542214946 2245570352
fastest aarch64_neon aarch64_neon
Reviewed-by: Gvozden Neskovic <[email protected]>
Reviewed-by: Brian Behlendorf <[email protected]>
Signed-off-by: Romain Dolbeau <[email protected]>
Closes #5248
Diffstat (limited to 'module/zcommon/zfs_fletcher.c')
| -rw-r--r-- | module/zcommon/zfs_fletcher.c | 3 |
1 files changed, 3 insertions, 0 deletions
diff --git a/module/zcommon/zfs_fletcher.c b/module/zcommon/zfs_fletcher.c index 355384f50..508ac4a1d 100644 --- a/module/zcommon/zfs_fletcher.c +++ b/module/zcommon/zfs_fletcher.c @@ -176,6 +176,9 @@ static const fletcher_4_ops_t *fletcher_4_impls[] = { #if defined(__x86_64) && defined(HAVE_AVX512F) &fletcher_4_avx512f_ops, #endif +#if defined(__aarch64__) + &fletcher_4_aarch64_neon_ops, +#endif }; /* Hold all supported implementations */ |