A branch of sgminer optimized with GCN cross lane instructions on AMD (ethash, phi2, lyra2Z[z], allium, x25x, lyra2REv2/v3, argon2d, yescrypt, neoscrypt, 0x10)
I did a benchmark of this 2 patterns and the old one seems to be faster, but not sure why. Maybe this info will be useful for further x22i performance improvements.
There is 2 known patterns that used for keccak kernel. The old one that is used in x22i now and newer one that was implemented by wolf I think. The second one used by avermore in x16r algo: https://github.com/brian112358/avermore-miner/blob/43e36dff5c21ad1ba98ef4b0efa07592780028d8/kernel/x16.cl#L993 This pattern is used by lycl miner: https://github.com/CryptoGraphics/lyclMiner/blob/master/kernels/keccakF1600/keccakF1600.cl And also it used by Ethash kernel: https://github.com/ethereum-mining/ethminer/blob/16e214428ddfc0954df10ba3a728149d68409606/libethash-cl/kernels/cl/ethash.cl#L87
I did a benchmark of this 2 patterns and the old one seems to be faster, but not sure why. Maybe this info will be useful for further x22i performance improvements.