If you haven't read the "README" file from Con Kolivas himself, when discussing Litecoin he delineates an optimizing algorithm in which you first find a sweat spot with intensity, then with another value, then another. Usually, once you hit the point of diminishing returns with the intensity, thread-concurrency and work unit those are the next values to tweak. Something you do can do without going the esoteric route of OS tweaking, you could try modifying the GPU's mem clock as well as (slightly) overclocking.
Cgminer gpu software#
The best results I've seen are with Gentoo Linux - as you may know, with this distro you first stablish your specific hardware configuration and then compile everything from code, thus optimizing all our software for the hardware you have to extract the most value off of it. Optimum hash rate depends not just on the parameters sent to cgminer but also on how was this program compiled, if it's 32bits/64bits, and the OS configuration. How can I achieve workable 320 kh/s, as wiki told? I do not touch default Core and Memory Clocks: $ amdconfig -odgcĭefault Adapter - AMD Radeon HD 6800 SeriesĬonfigurable Peak Range : What is the magical border between 11 and 12 intensity? This told me cgminer, when I make +1 to -thread-concurrency 2049 However, there is another mistake in wiki: -thread-concurrency 8192, my card have no enough RAM: Maximum buffer memory device 0 supports says 134217728 This is around declared in Wiki 320ks/s, but it is not a workable state, there are only errors - HW. 18 - very huge desktop lag, no accepted shares, and HW growth very fast:.16 - speed growth, but no accepted shares, only HW :.
However, I can not use -I more than 11, because hardware errors HW The wiki recommend use this -I 19 -g 1 -w 256 -thread-concurrency 8192 I'm found that speed of my cgminer should be around 320kh/s:
0 kommentar(er)
