08:46:40 <hyc> yeah, we have M1 results now, ~450H/s 1 thread
08:48:14 <hyc> bench is detecting things weirdly tho, --auto is turning off hardware AES when it should leave it on
08:49:43 <sech1> I expected more than 450 h/s
08:49:51 <sech1> It's probably limited by RAM latency
08:50:29 <hyc> https://mobile.twitter.com/stefanpenner/status/1331147710880120833
08:51:11 <hyc> having trouble getting it to use more than 1 mining thread
08:51:34 <hyc> dunno wbat's going onthere
08:53:17 <hyc> I see nothing in benchmark.cpp to cap the threadcount
08:54:26 <sech1> Even my Ryzen 4700U with 2400 MHz 17-17-17-39 RAM can do 690 h/s per thread
08:55:15 <sech1> maybe huge pages don't work on M1 (yet)
08:59:19 <hyc> don't think macos does huge pages
09:03:17 <selsta> I’m also doubtful at 450MH/s, even my Mac laptop from 2014 reaches 350MH/s on one thread.
09:04:27 <selsta> Also we don’t know if power or efficiency cores are used when testing one thread.
09:41:31 <Inge-> sech1: don't say that :/ my 3970x only does 806 H/s at 3900MHz 3600-14
09:50:50 <sech1> Mobile CPUs has memory controller on die, they have lower latencies overall
09:52:22 <Inge-> can't wait for my cooler to arrive. Maybe in jan.
09:58:50 <sech1> My 4700U boosts to 4.2 GHz btw and only achieves 690 h/s
14:58:25 <hyc> and the M1 has HBM on-chip, latency shouldn't be an issue