Frank Denis 6669885aa2
Faster BLAKE3 implementation (#25574)
This is a rewrite of the BLAKE3 implementation, with vectorization.

On Apple Silicon, the new implementation is about twice as fast as the previous one.

With AVX2, it is more than 4 times faster.

With AVX512, it is more than 7.5x faster than the previous implementation (from 678 MB/s to 5086 MB/s).
2025-10-15 14:03:56 +02:00
..
2025-07-07 22:43:51 -07:00
2025-09-18 22:39:33 -07:00
2025-09-30 13:44:54 +01:00
2025-10-06 06:54:52 +02:00
2025-09-03 21:46:01 -07:00
2025-09-18 22:39:33 -07:00
2025-08-29 17:14:26 -07:00
2025-09-30 13:44:54 +01:00
2025-09-18 22:39:33 -07:00
2025-08-31 12:49:18 -07:00
2025-08-28 18:30:57 -07:00
2025-10-10 22:47:47 -07:00
2025-09-20 14:34:18 -07:00
2024-08-22 08:44:08 +02:00
2025-09-30 13:44:53 +01:00
2025-09-11 00:18:37 -07:00
2025-07-31 22:10:11 -07:00
2025-08-29 17:14:26 -07:00
2025-08-29 17:14:26 -07:00
2025-08-29 17:14:26 -07:00
2025-09-18 22:39:33 -07:00
2025-09-30 13:44:51 +01:00
2025-10-08 17:00:15 -07:00
2025-08-29 17:14:26 -07:00
2025-09-30 13:44:55 +01:00
2025-07-22 09:41:44 -07:00
2025-10-06 01:28:56 +02:00
2025-08-28 18:30:57 -07:00
2025-08-31 12:49:18 -07:00
2025-08-31 12:49:18 -07:00
2025-07-16 10:27:39 -07:00
2025-09-24 20:01:18 -07:00
2025-08-29 17:14:26 -07:00