03:39:52 https://hardware.slashdot.org/story/20/03/01/0012226/stealth-startup-plans-fundamentally-new-kind-of-computer-with-circuit-rearranging-processor 18:44:07 sounds highly specialized. expecting a chip to reconfigure itself for entire programs is a pipe dream 18:44:18 imagine a chip trying to reconfigure itself to run an Oracle database 18:44:32 you'd need quadrillions of logic gates 18:45:10 just to run a simple operating system would be overwhelming 18:46:23 if this isn't pure snake oil, it's just an FPGA_style coprocessor that's activated for particular jobs 18:47:53 but indeed, if it's real, it could be the RandomX chip we've hypothesized... 18:48:14 and if it gets adopted into every smartphone and other personal computing device, big win for us 18:52:12 can you explain why a RX chip would be beneficial? 18:52:57 if it were easily available, ubiquitous? that's a solid win because anyone and everyone could mine in their idle time 18:53:08 which is what they can do now with a CPU 18:53:20 but if this new chip is more efficient than a CPU, hey, why not use it 18:53:47 *and is not a single-source proprietary system 18:54:30 like, would anyone have a problem if ASICs if they were open? 18:54:48 with* 18:55:28 with the present generation of asic designs, probably yes. 18:55:44 and then, if they were open, but still only produced by a couple large players, is that still a problem? 18:55:46 e.g., see the antminer E3s being obsoleted on ETH 18:56:25 if they aren't inherently limited designs with planned obsolescence, then I think they would be acceptable. 18:56:43 hmm ya. 18:56:54 just like x86 chips only come from a couple of large players; everyone can get them, and the don't instantly become unusable as tech marches forward 18:58:35 i think a bad scenario would be that these new chips are better at RX than general CPUs, so they wouldnt be mass adopted which is one of the advantages we have targeting CPUs, and then they would only have a niche for RX. 18:59:02 are only better* 18:59:35 clearly they're being pitched for machine learning, which is a much bigger market than Monero 18:59:51 sorry thats a garbled mess. my point is that maybe they cant beat CPUs generally but they can at RX, so they are not useful to the mass market. 18:59:57 true 19:00:37 and the current trend is that every new smartphone processor design includes an AI/neural net processor 19:00:55 so the potential for mass adoption is already assured, if it actually works well. 19:01:41 are they pitching it to also be used generally or as a piggyback chip? 19:02:09 that seems a bit vague to me. they talk about it being the core of a ground-up redesign of entire computing systems 19:02:24 but as I already noted, the idea is ludicrous for a lot of common computing tasks 19:02:39 so I believe it will only pan out for some specialized tasks 22:56:36 https://cdn.discordapp.com/attachments/645553582292074526/683785570081570877/unknown.png 22:56:41 From -pools 22:56:56 Apparently randomX is being used to benchmark/stress test CPUs