10:06:03 <hyc> I wonder if all GPUs have gone, or some still trying
10:06:12 <sech1> Some ARE trying
10:06:32 <sech1> xmrig gets endless github issues from GPU owners trying to run it
10:09:43 <sech1> last one was "Adrenaline 2020 19.2.2 driver not working with RandomX"
10:09:50 <sech1> It turned out to be AMD's fault (again)
10:10:08 <sech1> When I tried to run it with no optimizations, OpenCL compiler crashed inside amdocl64.dll
10:10:25 <sech1> it doesn't crash with optimizations, but produces incorrect code
10:10:32 <tevador> there have always been endless issues with GPU mining
10:10:38 <hyc> sheesh
10:10:40 <tevador> mostly driver issues
10:10:57 <sech1> this time it's buggy OpenCL compiler in new Adrenaline drivers
10:11:16 <tevador> this is one of the biggest weaknesses of ProgPow
10:11:37 <hyc> so, does previous version still work? driver version
10:11:38 <tevador> it relies on the correctness of closed source compilers
10:11:52 <sech1> 19.2.1 works perfectly, 19.2.2 doesn't work
10:12:40 <hyc> and does .1 opencl compiler work withb.2 driver?
10:12:59 <sech1> I can't imagine how they test their compiler. Is it hard to grab all OpenCL source code from github and just compile it in a loop with different flags?
10:13:23 <sech1> If I put 19.2.2 compiler to compile RandomX in a loop, it crashes within a second or two
10:15:33 <sech1> I guess it's possible to compile binary kernels with 19.2.1 then update driver and use old compiled kernels, but that's not a proper solution
10:16:41 <sech1> and the worst thing is AMD doesn't have any official place to report these bugs
10:18:13 <hyc> it's been standard practice on linux to extract opencl compiler from driver package and use with mesa drivers
10:51:26 <p3rL> something wrong with xmrig 5.3.0 > Segmentation fault (core dumped)
10:51:41 <p3rL> 5.0 works fine on same machine
10:51:50 <Inge-> hyc: until I find something better, or sell off my Vega64 cards - when it is cold I still mine XMR with them ...
10:52:36 <Inge-> But I need different settings. RandomX doesn't generate enough heat :(
10:58:32 <SR71-miner> Inge-, do Ethash ~50MHs and heat you home ;)
11:08:21 <Inge-> that's an idea
11:08:32 <Inge-> not too hot on cryptos where I can't run a node tho :P
11:20:07 <tevador> if someone has an approved account to comment here: https://lukminer.org/2019/11/30/experimental-randomx-support/#comments can you please ask for librandomx attribution?
11:29:30 <vegawasserheizer> SR71-miner, freezin to death since fork, X-P  https://images2.imgbox.com/cd/01/ogrhxXoN_o.png
11:44:39 <SR71-miner> .tell Tevador you can contact Lukas the Lukminer Guy directly via  lukas.l.alexander⊙gc
22:16:48 <tevador> Can anyone test this? https://github.com/tevador/randomx-sniffer/issues/2
22:17:03 <tevador> I'm pretty sure it will work