@hyc can you make a fixed version of tevador/RandomX #195 ?

if it's still needed

<\x> hyc: intel seems to be gonna actually doing this

<\x> browser.geekbench.com/v5/cpu/8011325

<\x> >14 Cores, 20 Threads

<\x> 8 core alder with 6 core gracemont

<\x> 8c/8t + 6c/12t

\x maybe 6 cores alder lake (6c/12t) + 8 cores gracemont???

<\x> that works out too

tevador how is #211 working in randomx-benchmark and xmrig benchmark (correct hashes) but doesn't work in tests, I don't get it

are you also getting a fail on test 84?

sech1: it looks like the full dataset mode works, but JIT+cache gives wrong hashes

try ./randomx-benchmark --verify --jit

yes, it fails on 84 for me too

ahhh I get it

light mode?

then it will certainly fail :D

will submit a fix soon

submitted

xmrig probably has the same problem now :D

people run xmrig in light mode?

hyc they do sometimes: xmrig/xmrig #2377

I'll take a look at #195 now

ok done as #213

both merged

since there are asm code changes, it would be nice to test it on linux too (I don't have any at hand now)

I tested it on linux

then it's all fine

I tested the arm code on my phone

tevador: I would like to update our submodule. could you tag v1.1.9 or should I just set it to latest master commit?

OK, I will tag a release

thanks