Dec 7, 2024

These are numbers drawn directly out of our libbitcoin-system performance tests. There are two distinct scenarios, simple sha256 hashes and sha256 Merkle root computations. I don't have a avx512 platform yet, but I have projected the results in both scenarios based on the trend between sse4 (4x32) and avx2 (8x32) to avx512 (16x32). The third plot is base2 logarithmic scale.