18:27:22 <sech1> hyc I think I know why Bitmain X5 can't mine Zephyr even if it's the same RandomX. The reason is simple and stupid, lol. Zephyr's mining blob is 196 bytes, while Monero's mining blob is 76 bytes. So X5 crashes when it tries to mine Zephyr - probably a buffer overflow. The overflow can be fixed with a firmware patch, but if they did an ASIC circuit
18:27:22 <sech1> for the initial Blake2b hash, they might've limited to 128 bytes (Blake2b block size), and this can't be fixed in firmware.
18:51:26 <hyc> lol
18:55:32 <hyc> a hard limit to hash block size seems unlikely, it's such common practice to use hashers on inputs different from the block size
18:57:53 <sech1> They might've "optimized" the Blake2 hash to be limited to only 1 block (128 bytes)
18:59:41 <sech1> because that's what I did too in CUDA code, it's faster :D