use hive, there will be a normal speed.
https://picua.org/images/2019/02/12/1a956830735a246cda54193988795313.png

Cool... When AMD and Grin?

This.
Upgraded to CUDA 10 and 418.81 driver, on windows 10 & Bminer 14.3.1.
Page file is 60GB for 4X1080ti's. Cannot get it to run on c31, and on c29 the graph rate is not as expected.
Could it be CPU related? Core duo something.
Any fix please?

On nicehash BEAM on AMD crads, works only with 1 card only?

even with intensity -12 my hashrate is less than 14.0 build 27.2g vs around 26.7-26.8.0 g/sec, with default intensity hashrate is only 26.6g with new 14.3.1 build
new 14.3.1 build

https://i.imgur.com/XsFZlyv.png

old 14.0 build

https://i.imgur.com/qglZGxH.png

that's on TDP 70% and memory +500 using pill as well and different core settings on different models

well GTX 1080Ti has little increase on newer build, but EVGA FTW 1080 models have higher hashrate on older build, msi gamer x GTX1080 is more or less same, fidelity increased as well on newer build but that's only when i set intensity -12

i think if you use same cpu settings in next build miner can achieve even higher hashrates than 14.0 build, on newer build there's some tweaks missing than older 14.0 build has, on both builds cpu usage is still very high so it's better to return old 14.0 build tweaks in next release, at least GTX 1080 need some more optimizations

Would it be possible to add an option to disable NVML hooking? I can't manage to run the miner on docker...

Failed to initialize the CUDA platform: CudaError: Could not initialize the NVML library

@UnclWish: Why don't you write it yourself?
You WISH to get something.... There is no real need. I am sure the devs do their best.

It's your opinion? If you can't say something useful, don't write at all.
I wouldn't write if it were not real need in AMD Grin support.

No. You have to use another miner.