Open alphaonex86 opened 11 years ago
Or use ony X% of the core: If I have 200 Gpu core, Use only 50 core for 25% The target is: do a quiet mining to include as background progress into my software. Put 10% to not change temperature and GPU speed fan. And of course, little cpu for stratum+tcp is appreciated.
Does intensity not do this well enough? I suppose it's possible to set the --temp-target to whatever it is at startup, but I'm not sure I see a valid use case for this.
No, in best case I have 80% usage of the GPU. --temp-target will depands of the normal/native gpu temp... I have included bfgminer into my software (Ultracopier), but it put the GPU at 100%, bad on laptop, the user ask them why their GPU have the fan at max, ...
For example, --intensity 1 - is this still 80%?
Mining on others' computers without their express permission is likely a criminal action... I will not support a use case that depends on the owners' ignorance of the mining.
Yes, remain at 80%.
The mining is clearly displayed, at exchange of free Ultimate version. I have discute with my community, all it's agree (at condition remain free version and paid Ultimate version without mining) like you can see on the forum. But mostly the user wish only <10% of GPU, 80% disturb it. (I understand that's).
I don't currently have a working GPU to do new development with, but my 5850 that I recently RMA'd arrived today, so hopefully I will be able to do something soon. Please ping me here in a few days if I forget.
I'm writing my new software version, I will release with this features, I think give you more BTC when it's finish. Have you forgeten me?
Still trying to get my new GPU dev environment setup here.. Did you by any chance try with -I -1 (negative intensity)?
No, it block at 0 k/s! If you are under windows, setup the dev env is easy, under linux it's impossible (see the irc channel), I have tryied with multiple ati card. Under linux the unique way it's vt-d + qemu and bing the graphic card into the vm.
I have found nothing into the opencl to simply limit it.
Spent a few hours trying to get this working tonight... I was only able to limit GPU usage by refusing to give it sufficient work, which caused the CPU usage to be maxed out. And there was no way to auto-detect the limits either. So I'm afraid this enhancement isn't possible with today's drivers :(
Very strange no? What is the max idle/use time with this method, if it's < 1s, sood good for throttling, the target is throttling on GPU with power (because the low end GPU is fan less).
No news? Limit the MH/s is possible?
Can you implement the feature? I have fund the feature. Or MH/s limit, or GPU usage with device fision of OpenCL: http://software.intel.com/en-us/articles/opencl-device-fission-for-cpu-performance
Hello,
Can you add an options to use only 25% (or X%) of the GPU? Then have access to throttling.
Cheer,