It is so slow with gtx1080:
GPU #0: GeForce GTX 1080, 8192 MB available, 20 compute units, capability: 6.1
GPU #1: GeForce GTX 1070, 8192 MB available, 15 compute units, capability: 6.1
...
ETH - Total Speed: 43.967 Mh/s, Total Shares: 3, Rejected: 0, Time: 00:03 ETH: GPU0 20.961 Mh/s, GPU1 23.006 Mh/s ETH: 04/30/17-17:22:35 - New job from eth-eu1.nanopool.org:9999 ETH - Total Speed: 45.882 Mh/s, Total Shares: 3, Rejected: 0, Time: 00:03 ETH: GPU0 20.523 Mh/s, GPU1 25.359 Mh/s ETH: 04/30/17-17:22:39 - New job from eth-eu1.nanopool.org:9999 ETH - Total Speed: 45.595 Mh/s, Total Shares: 3, Rejected: 0, Time: 00:04 ETH: GPU0 20.689 Mh/s, GPU1 24.906 Mh/s GPU0 t=73C fan=54%, GPU1 t=61C fan=39% ETH: 04/30/17-17:22:41 - New job from eth-eu1.nanopool.org:9999 ETH - Total Speed: 46.282 Mh/s, Total Shares: 3, Rejected: 0, Time: 00:04 ETH: GPU0 20.994 Mh/s, GPU1 25.288 Mh/s ETH: 04/30/17-17:23:04 - New job from eth-eu1.nanopool.org:9999 ETH - Total Speed: 45.375 Mh/s, Total Shares: 3, Rejected: 0, Time: 00:04 ETH: GPU0 20.943 Mh/s, GPU1 24.432 Mh/s
This is very bad.. I'm using Windows 10 64 bit, Cuda 8 and 381.65 driver. Is there any solution for this? Or can I try to fix it? There is a source code somewhere? Is it some possibility to change thread per block size?
|