I found a block today on an overclocked Asus GTX 970 (4GB memory). Running the newer miner with the -s 0.2 and -c 1 set it is reaching 450 MH/s.
When running this I noticed it only uses 70% GPU. Of course I am not using a "custom coded" miner so for kicks launched the same instance 3 times and the collective hashrate looks higher than if I was running individually 250/244/72 MH for each. 566 MH/s total and GPU use goes to 90% I am not sure if this is more effective or not but will run overnight and see what happens.
HI guy
New bat file
Set up
Sia_GPU-Miner.exe -s 0.2 -c 1
Is that what?
Thank you
Yes, that is what I am using. Nice work Wolf on the optimization. If it does not work you may be using the wrong exe, the one I have is named gpu-miner.exe I think you may have a different version which does not support those options. Not sure which page of this thread I downloaded it on.
If you mean the actual rewrite I did and posted about earlier, thanks. If you mean what I have on GitHub, that's not an optimization, it's how you properly get the list of OpenCL platforms on a system, and the list of GPU devices in a selected platform. Besides that, I added options to give the user direct control over intensity, plus the number of kernel enqueues before stopping to fetch new work and print hashrate.
With those options available, it should be simple to mine on all devices on a system, and probably get a little higher hash out of them, simply because you can tune the intensity.
That GitHub is here:
https://github.com/wolf9466/Sia-GPU-Miner