All About BTC, LTC, ETH mining as well as other alternative crypto currencies
The fork of ethminer for Nvidia CUDA GPUs has been updated to a new version – Ethminer 0.9.41-genoil-1.0.5 (source). The latest release of Genoil’s ethminer for CUDA should bring back support for older Compute 3.0 GPUs, so if you have one of these you should check to see if it will work as it should now (if you had issues with the previous 0.9.41-genoil-1.0.4b3 release). It also has some cleaned up code, but you should not expect to see a performance boost with it. Do note that when running with CUDA it will run only on the first GPU, so you need to manually set the option
--cuda-devices 0 1 2 3 4 5 to have for example the miner run on a 6-card GPU mining rig. The fork from Genoil, although targeted at Nvidia CUDA miners, will also work on AMD OpenCL GPUs, but there it should not be any different than from the default ethminer that is OpenCL only.
The Blake2s algorithm is a new entry in the crypto mining scene and while initially there was only CPU mining support available for it now there is also a ccMiner fork with support for Blake2s algorithm on Nvidia GPUs (source). The initial CPU mining support for Blake2s was first introduced in cpumuiner-multi from tpruvot, though there is now a new more-efficient CPU miner available here. With the release of a GPU miner however CPU mining might be not worth it anymore, even though for the moment there is only GPU mining available for Nvidia-based video cards and no for AMD, but soon we are probably going to see Blake2s supported on AMD as well. The miner below is a 64-bit binary compiled with VS2013 and CUDA 7.5 for Compute 3.5, 5.0 and 5.2 GPUs, a 32-bit binary compiled with CUDA 6.5 for Compute 2.1, 3.0. 3.5, 5.0 and 5.2 GPUs is also included. Do have in mind however that the 64-bit version of the miner seems to be offering better performance for mining Blake2s coins.
Back to the ccminer 1.7.5 fork with Blake2s support. Our tests have shown that you can get somewhere around 1080 MHS for a Geforce GTX 970 GPU with a slight increase of the intensity over the default one, the power usage of the GPU while mining is about 168 Watts. For the moment it seems that there are just two altcoins that can be mined with Blake2s, but we are not going to be surprised if we soon see some more getting released. The first one is called NevaCoin (NEVA) and the more recent one launched yesterday is called OXEN. You might however want to be extra careful with OXEN as it seems that there might be a hidden trojan/virus inside the client as some users are reporting suspicious activity after installing the wallet!
We have compiled ccMiner 1.7.4 tpruvot’s fork Windows binary in a version with the optimal settings for mining Vanillacoin (VNL) on more recent Nvidia-based GPUs as suggested by Alexis Provos and with his performance optimizations as implemented by tpruvot in the latest code. This Windows binary is a 32-bit one compiled with CUDa 6.5 and only with Compute 3.5 support as apparently this provides the optimal performance at least on Maxwell-based Nvidia GPUs (Faster than using Compute 5.2 on them). Below you can also see a comparison that we did with Nvidia GeForce GTX 970 from Gigabyte (WF3OC) and a reference design Gigabyte GTX 980 Ti to see the difference in performance with the regular version when using CUDA 7.5 and Compute 5.2 and with the optimized CUDA 6.5 and Compute 3.5 version available for download below. Do note that although this binary will work with other algorithms as well as the Blake 256 8-rounds used by VanillaCoin it may not provide the optimal performance for other algorithms as it is targeted to provide the best VNL mining hashrate.
GeForce GTX 970 – CUDA 7.5 Compute 5.2
Default – 2490 MHS @ 164 W
Intensity 25 – 2535 MHS
Intensity 26 – 2637 MHS
Intensity 27 – 2692 MHS
Intensity 28 – 2718 MHS
Intensity 29 – 2722 MHS
Intensity 30 – 2729 MHS
Intensity 31 – 2732 MHS
GTX 980 Ti – CUDA 7.5 Compute 5.2
Default – 3390 MHS @ 211 W
Intensity 25 – 3797 MHS
Intensity 26 – 3884 MHS
Intensity 27 – 3986 MHS
Intensity 28 – 4009 MHS
Intensity 29 – 4043 MHS
Intensity 30 – 4075 MHS
Intensity 31 – 4132 MHS
GeForce GTX 970 – CUDA 6.5 Compute 3.5
Intensity 31 – 2942 MHS @ 177 W
GeForce980 – CUDA 6.5 Compute 3.5
Intensity 31 – 4418 MHS @ 238 W
Do note that increasing the intensity over the default setting (no manual -i intensity option used) even on a regular release will provide some increase in performance as you can see in our comparison going up to the maximum supported intensity of 31. Have in mind that increasing intensity also makes the system a bit less responsive the higher you go and at 30/31 it will be lagging more, but the extra performance gain you will get with the optimal binary and higher Intensity setting is significant, so for a dedicated mining rig for VanillaCoin you should definitely go for it.
Update: This particular performance optimization is for VanillaCoin’s Blake256 8-rounds implementation, the CUDA 6.5 and Compute 3.5 compilation of the particular CUDA code for that algorithm give better performance than Compute 5.0 or 5.2. The CUDA code is different for other Blake 256 algorithms, including the one for Decred’s Blake-256 14-rounds implementation, so do not expect performance increase in others. In fact specifically for Decred, a 64-bit compilation with CUDA 7.5 and for Compute 5.x does seem to provide best performance with the publicly available ccMiner 1.7.4 fork from tpruvot.