Add Aarch64 assembly implementation of Twofish
5418d9ca4c0eUnpublished
Actions

Unpublished Commit · Learn More

Not On Permanent Ref: This commit is not an ancestor of any permanent ref.

Description

Add Aarch64 assembly implementation of Twofish

* cipher/Makefile.am: Add 'twofish-aarch64.S'.
* cipher/twofish-aarch64.S: New.
* cipher/twofish.c: Enable USE_ARM_ASM if __AARCH64EL__ and
HAVE_COMPATIBLE_GCC_AARCH64_PLATFORM_AS defined.
* configure.ac [host=aarch64]: Add 'twofish-aarch64.lo'.

Patch adds ARMv8/Aarch64 implementation of Twofish.

Benchmark on Cortex-A53 (1152 Mhz):

Before:
TWOFISH | nanosecs/byte mebibytes/sec cycles/byte

 ECB enc |     27.51 ns/B     34.67 MiB/s     31.69 c/B
 ECB dec |     26.37 ns/B     36.17 MiB/s     30.38 c/B
 CBC enc |     28.64 ns/B     33.29 MiB/s     33.00 c/B
 CBC dec |     26.21 ns/B     36.39 MiB/s     30.19 c/B
 CFB enc |     28.54 ns/B     33.42 MiB/s     32.88 c/B
 CFB dec |     27.40 ns/B     34.81 MiB/s     31.56 c/B
 OFB enc |     28.38 ns/B     33.61 MiB/s     32.69 c/B
 OFB dec |     28.37 ns/B     33.61 MiB/s     32.69 c/B
 CTR enc |     27.57 ns/B     34.60 MiB/s     31.76 c/B
 CTR dec |     27.57 ns/B     34.60 MiB/s     31.76 c/B
 CCM enc |     55.28 ns/B     17.25 MiB/s     63.69 c/B
 CCM dec |     55.29 ns/B     17.25 MiB/s     63.70 c/B
CCM auth |     27.83 ns/B     34.27 MiB/s     32.06 c/B
 GCM enc |     28.86 ns/B     33.04 MiB/s     33.25 c/B
 GCM dec |     28.87 ns/B     33.04 MiB/s     33.25 c/B
GCM auth |      1.30 ns/B     731.9 MiB/s      1.50 c/B
 OCB enc |     29.69 ns/B     32.12 MiB/s     34.20 c/B
 OCB dec |     28.50 ns/B     33.47 MiB/s     32.83 c/B
OCB auth |     29.04 ns/B     32.84 MiB/s     33.45 c/B
         =

After (~1.3x faster):
TWOFISH | nanosecs/byte mebibytes/sec cycles/byte

 ECB enc |     19.97 ns/B     47.77 MiB/s     23.00 c/B
 ECB dec |     18.29 ns/B     52.16 MiB/s     21.06 c/B
 CBC enc |     20.94 ns/B     45.54 MiB/s     24.13 c/B
 CBC dec |     18.34 ns/B     52.00 MiB/s     21.13 c/B
 CFB enc |     20.83 ns/B     45.77 MiB/s     24.00 c/B
 CFB dec |     19.97 ns/B     47.76 MiB/s     23.00 c/B
 OFB enc |     20.94 ns/B     45.54 MiB/s     24.13 c/B
 OFB dec |     20.94 ns/B     45.54 MiB/s     24.13 c/B
 CTR enc |     20.19 ns/B     47.24 MiB/s     23.26 c/B
 CTR dec |     20.19 ns/B     47.24 MiB/s     23.26 c/B
 CCM enc |     40.53 ns/B     23.53 MiB/s     46.69 c/B
 CCM dec |     40.53 ns/B     23.53 MiB/s     46.69 c/B
CCM auth |     20.40 ns/B     46.74 MiB/s     23.50 c/B
 GCM enc |     21.49 ns/B     44.39 MiB/s     24.75 c/B
 GCM dec |     21.48 ns/B     44.39 MiB/s     24.75 c/B
GCM auth |      1.30 ns/B     731.8 MiB/s      1.50 c/B
 OCB enc |     22.15 ns/B     43.05 MiB/s     25.52 c/B
 OCB dec |     20.47 ns/B     46.58 MiB/s     23.59 c/B
OCB auth |     21.64 ns/B     44.07 MiB/s     24.93 c/B
         =