08:52:01 https://st.overclockers.ru/images/news/2020/03/06/260_cores.png 08:52:23 260 million cores is 150-300 GH/s RandomX total, depending on how many of them are Zen2 09:45:00 hyc tevador hmm, this is the first scatter graph where CN/R ASICs are clearly visible: https://i.imgur.com/HJTTILf.png 09:46:45 highlighted: https://i.imgur.com/kU2EgST.png 09:48:15 They appeared around block 1948000 09:48:39 October 19th, 2019 09:48:46 More than a month before RandomX fork 09:49:20 or ~7 months since CN/R fork 09:51:57 But since there were only ~30 MH/s of them, overall hashrate didn't jump and they just replaced some GPU miners 09:55:09 They earned $360-400k in total (if selling mined XMR immediately) 10:52:33 Has anyone gone public with CN/R asics? 10:52:50 or is it all based on nonce-interpretation? 12:21:25 i think nonce interpretation Inge- . was that just done by visual inspection sech1 ? 12:26:32 and sech1, can you resolve between asics vs. FPGAs? 12:28:57 damn, nanopool saw a jump from 139 to 252 Mh/s recently 12:30:04 and it looks like the hashrate didn't come from another pool, nor did it go back to another pool 12:34:04 I already knew what to look for base on Sumo nonce analysis: https://www.reddit.com/r/Monero/comments/eqecqn/cryptonightr_is_mined_by_asics_or_fpgas_now/ 12:34:28 It was just not visible on my own graphs, but that one with thicker points makes it clear 12:35:22 It's ASICs. It doesn't make sense for FPGAs to start mining 7 months after algo change, they could've done it in a few weeks. 12:54:11 30MH is kind of a small number compared to earlier ASIC inflow 12:57:49 Maybe it was just a small batch or something that used cheap 32nm or 65nm node (i.e. less efficient) 12:58:19 plus CN/R was more asic resistant compared to previous variants 13:05:35 still a pretty cool find 13:05:51 And AFAIK nobody has gone public with these ASICs - so the only trace is from nonces? 13:13:24 The trace comes from nonces and the fact that Sumo is not profitable for GPUs (at all) 17:50:14 Isthmus, i forget... have u tried statistical magical on nonce distribution to detect asics? if so, can u confirm sech1 's finding? 17:50:35 sorry, statistical magic 19:22:14 do zen, zen2 and zen 3 all use the same socket for board reuse? 19:41:20 kinghat: yes, with zen 3 being 2nd generation 19:48:12 👌 19:53:51 though not all older boards get sufficient BIOS updates to handle the newer processors 22:12:58 some A320 boards can run Zen2, and most all B450 boards, X470 boards .. 22:13:22 problem with A320 was not enough bios space for gui + support for old and new CPU's iirc 22:13:50 so some had specific non-gui BIOS to support 3xxx cpu's