12:23:15 <sech1> hyc https://github.com/tevador/RandomX/pull/246
13:41:01 <hyc> nice catch
13:44:03 <sech1> I actually found it a long time ago when memory leak debugging p2pool, just merging it upstream now
14:50:55 <hyc> you actually ran into it? pretty rare for these failures to happen
14:51:16 <hyc> ironically once they happen, leaks make it worse
15:00:59 <sech1> It's easy to run into it when you don't have enough huge pages
15:01:27 <sech1> RandomX then switches to regular allocator, but JIT compiler from the first call leaks
15:05:15 <hyc> ahh
15:05:55 <hyc> then this could have been leaking in monerod too
15:09:28 <sech1> most likely
15:09:37 <sech1> but it's a small one-time leak
15:09:52 <sech1> because monerod creates cache only once
18:49:43 <QuickBASIC_> Offtopic, but does anyone even like c#, holy crap, I was writing a plugin for an app and the example plugin was in c# and I hate my life. It's like the bastard child of PowerShell and c++.
19:03:43 <sech1> It's Microsoft's creation. nuff said
19:03:58 <sech1> On topic: https://youtu.be/BRtBB2VnF8M?t=1070
19:04:07 <sech1> 70 million Ryzen desktop CPUs sold so far
19:04:54 <sech1> Assuming they are all at or above Ryzen 5 1600 hashrate (4.5 kh/s), it's 315 GH/s of potential RandomX hashrate
19:05:12 <sech1> or more than 100x current network hashrate
19:07:21 <gingeropolous> then what are they doin
19:08:44 <sech1> netflixing and chilling
19:10:04 <gingeropolous> can't believe they are going pins on mobo. finally
19:11:31 <sech1> and if we add Intel desktop CPUs sold in the last 10 years, it's probably more than 1000 GH/s (lower hashrates, but sheer numbers compensate it)
19:12:04 <sech1> so Monero can easily pump 500x and still stays barely profitable to mine :D
19:16:36 <gingeropolous> lol
19:53:54 <merope> Does that include Epyc? Or just desktop ryzen?
20:05:48 <sech1> "AM4 socket CPUs"
20:05:53 <sech1> so only desktop
20:29:53 <merope> I wonder what the production numbers for server cpus are
21:28:58 <hyc> smaller.