News for hackers. Stuff that matters.

A comment on BLAKE3 1.0. Parent comment

Re:BLAKE3 1.0 by SAI_Peregrinus on Monday July 26, 2021 @11:12AM (#27960405)
And if you're designing hardware (or on an FPGA) ChaCha (and BLAKE2/3) are actually really freakin' fast for the die area they take. IIRC ChaCha20 beats AES in hardware. It's just extremely rare (FPGA only) to find it in hardware, since there's not enough demand. I expect that to eventually change.
[ Parent | on HN ]
Re:BLAKE3 1.0 by api on Monday July 26, 2021 @11:16AM (#27960443)
Given that all these are ARX cores, I wonder if a fused ARX instruction could cover a wide range of them?
[ Parent | on HN ]
Re:BLAKE3 1.0 by brandmeyer on Monday July 26, 2021 @07:56PM (#27967005)
ARMv8.2 has rotate-and-xor and xor-and-rotate so that the (extremely cheap) xor can be saved.
[ Parent | on HN ]
Re:BLAKE3 1.0 by SAI_Peregrinus on Monday July 26, 2021 @11:28AM (#27960635)
Probably. With a barrel shifter easily. Barrel shifters are a bit slower than wired shifts though, so for ultimate speed you'd end up with hardware shift amounts.
[ Parent | on HN ]
Re:BLAKE3 1.0 by api on Monday July 26, 2021 @11:34AM (#27960710)
The advantage would be a single instruction that implemented the core of a wide range of things including BLAKE2, BLAKE3, Salsa, ChaCha, Speck, and more.
[ Parent | on HN ]