GCM: Add bulk processing for ARMv8/AArch64 implementation
authorJussi Kivilinna <jussi.kivilinna@iki.fi>
Sun, 9 Oct 2016 09:53:48 +0000 (12:53 +0300)
committerJussi Kivilinna <jussi.kivilinna@iki.fi>
Sun, 9 Oct 2016 14:57:03 +0000 (17:57 +0300)
commitbfd732f53a9b5dfe14217a68a0fa289bf6913ec0
treeff692b729e5dcbf9ff143af7acc6a8dc99263425
parent27747921cb1dfced83c5666cd1c474764724c52b
GCM: Add bulk processing for ARMv8/AArch64 implementation

* cipher/cipher-gcm-armv8-aarch64-ce.S: Add 6 blocks bulk processing.
--

Benchmark on Cortex-A53 (1152 Mhz):

Before:
                    |  nanosecs/byte   mebibytes/sec   cycles/byte
 GMAC_AES           |      1.30 ns/B     731.6 MiB/s      1.50 c/B

After (1.49x faster):
                    |  nanosecs/byte   mebibytes/sec   cycles/byte
 GMAC_AES           |     0.873 ns/B    1092.1 MiB/s      1.01 c/B

Signed-off-by: Jussi Kivilinna <jussi.kivilinna@iki.fi>
cipher/cipher-gcm-armv8-aarch64-ce.S