Added siphash24x2and siphash24x4variants with SSE2 instruction optimization. Also replaced ROT16 in the SSE2 variant with shuffle instead of shift instructions.
There are two build examples, lean30x2and lean30x4in the Makefile, with a flag to disable AVX2.
Tested on the lean miner against verify30with nounces 63 and 140902 (on a single thread).
Added
siphash24x2
andsiphash24x4
variants with SSE2 instruction optimization. Also replaced ROT16 in the SSE2 variant with shuffle instead of shift instructions.There are two build examples,
lean30x2
andlean30x4
in theMakefile
, with a flag to disable AVX2.Tested on the lean miner against
verify30
with nounces 63 and 140902 (on a single thread).