Quote from: Vcxz on July 11, 2019, 01:51:04 PM

Quote from: doktor83 on July 11, 2019, 12:45:23 PM

V1.9.1
- Added support for Navi10 (gfx1010)
- Arto, conceal, normal, normalv7, marketcash, b2n, fast, hycon, hospital algos have lower power consumption on Vegas with auto settings
- Hopefully fixed bug with invalid job id / same job id
- If more than 5 'invalid share' messages received from pool within same connection, auto miner restart
- Fixed cn-gpu hashrate (it was lower from v.1.8.8 -> my mistake)
- CN/R program precompilation process is now only logged, not displayed on screen

+ Added support for gfx1010 card. Well at least i hope, because i don't own a rx 5700 card, so i couldn't test Grin

It would be good if someone could test it and report back.

+ On Vega, intensity 0 (auto setting) now uses a little less power on next algorithms :
Arto, conceal, normal, normalv7, marketcash, b2n, fast, hycon, hospital

+ I tried to fix the 'invalid job', 'invalid share' issue, unfortunately i couldn't reproduce it neither earlier neither now, that's why i say i tried ..

Anyways if the problem still persists, i also added a 'protection' against this, if within the same pool connection more than 5 invalid share messages received from the pool, miner will auto restart.

+ I accidentally messed up CN/GPU hashrate from version 1.8.8, now it is fixed and hashing again at full speed

tested on 5700xt
Algo: normalv4
1 min avg: 231 H/s
5 min avg: 206 H/s
Max Speed: 866 H/s

Thanks for trying it out. You probably tried the auto setup, guess didnt play with manual config yet..

i hope more can be squeezed out from those cards and the right intensity/worksize/fragments.