#6 sha3 performance patch

[And Sch]

This patch increases performance by 1.5 times, from 91 MB/s to 137 MB/s on my i5 system.

Changes are unrolling the loops using macros. You may want to change the macro names, I didn't know what to call them.

That's all the patches I have for now, for the best hash functions in my opinion.

[Aleksey]

Thank you very much for improving the algorithm. Your version is indeed faster on the Core i7 GCC-4.7.0 64-bit. I'll verify the productuctivity with other compilation settings latter.

But please, please, please send the next patches in the more standart and convenient *.patch format [1]! ;)

[1] http://docs.moodle.org/dev/How_to_create_a_patch

[And Sch]

Ok, I'll do that.

[Aleksey]

Confirmed speed improvement on Intel Core i7 2.8GHz

• using i686-linux-gcc6.3.0 (SHA3-512 CPB: 102,17 -> 98,67; SHA3-256 CPB: 54,28 -> 52,57)
• using x86_64-linux-gcc6.3.0 (SHA3-512 CPB: 42,66 -> 27,99; SHA3-256 CPB: 22,64 -> 14,90)
• using x86_64-linux-gcc5.4.0 (SHA3-512 CPB: 44,00 -> 27.47; SHA3-256 CPB: 22,64 -> 14,63)

Tests were run as:

sudo LD_LIBRARY_PATH=librhash/ nice -n -20 /bin/bash -c './rhash -B --sha3-512; ./rhash -B --sha3-256; ./rhash -B --sha3-512; ./rhash -B --sha3-256'

[Aleksey]

Included into RHash v1.3.6.

Exact commit: cb44..16ee.