All About BTC, LTC, ETH mining as well as other alternative crypto currencies
Version 1.7.2 of tpruvot’s ccMiner fork source) for GPU mining on Nvidia-based video cards has been released and it comes with official support for Vanillacoin (VNL) as well as Decred (DCR) support. Below you can find compiled Windows binaries to try out the mining them, both are Blake256-based, but with some different requirements such as the need to use longpoll and getwork, and no support for stratum for Decred for example. There are two versions available, the 32-bit one is with support for older Compute 2.1 and 3.0 GPUs while the 64-bit version is for the newer Compute 3.5, 5 and 5.2 cards. We have tested the new release on GTX 970 and the results are slightly better at about 1.35 GHS (with the 64-bit miner) than what we are getting with the cgminer fork for Decred that relies on OpenCL that is also compatible with Nvidia cards. The ccMiner fork might be helpful to people that are having very high CPU usage with cgminer that results in lower mining performance, though the CPU load is still not that low even with ccMiner. Also the use of ccMiner for mining Decred could result in less rejected shares from the pool and no weird errors as some people are experiencing with cgminer, so if you are mining DCR on Nvidia with cgminer, then you should try with ccMiner as well.
The ccMiner SPMOD fork for Nvidia Maxwell GPUs (source) has recently been migrated to use the newer CUDA 7.5 from the older CUDA 6.5. The latest version with CUDA 6.5 support was version ccMiner 1.5.74 and the latest version is ccMiner 1.5.78 that already uses CUDA 7.5. The first release with CUDA 7.5 had not so good performance and was slower in some algorithms in terms of performance as compared to the latest CUDA 6.5 release, but the latest code has been optimized to offer similar or even better hashrate in most supported algorithms. We took an Nvidia GeForce GTX 970 (Gigabyte GV-N970WF3OC-4GD) running at stock frequencies for a spin with some of the currently popular algorithms to compare the performance of the latest CUDA 6.5 to the latest CUDA 7.5 release of ccMiner spmod. The results in the table clearly show that aside from NeoScrypt the latest CUDA 7.5 release offers very similar or even slightly better hashrate compared to the latest CUDA 6.5 version. The hashrate of X13, X15 and Qubit is still slightly slower, but not much. So unless you are mining NeoScrypt you should have no reasons not to upgrade to the newer CUDA 7.5 releases and soon we are probably going to see even better performance thanks to further optimizations of the code. Do note however that some other Maxwell GPUs might have some more variance in terms of performance difference between the two compared versions.
New year and a new update in the form of a freshly compiled Windows binary from the latest ccminer 1.7.1-dev (source). The latest updates include some small bug fixes as well as the addition of the GPU mining support for the sib algorithm (X11 + Russian Streebog-512/GOST). The tpruvot fork of ccMiner is compatible with older Nvidia GPUs, so this is the version to go for if you have older video card that you want to try mining cryptocurrencies with. This Windows binary is compiled with VS2013 and CUDA 6.5 and supports Compute 2.1, 3.0, 5.0 and 5.2 GPUs, however have in mind that not all of the supported algorithms may work on older Compute 2.1/3.0 GPUs, but you can at least try with it.