GPU Compute

'DirectCompute and OpenCL Benchmark' demonstrates the significance of what is already known, that Radeon is simply the choice when one needs GPU computing power especially in OpenCL, delivering up to 2x performance in OpenCL for this benchmark.

With Luxmark, having only selected the GPU under test and not the CPU, we see almost 3x between NVIDIA and AMD in OpenCL.

For Folding at Home, we were not able to run the CUDA test so we only compare using OpenCL and not CUDA optimisations.

Coin Mining

For AMD we used standard GUIMiner with OpenCL

For reference, the HD 4600 Integrated GPU in Intel's current flagship consumer CPU, the i7-4770K only acheives 5.9 mhash/s in Guiminer/OpenCL.

For NVIDIA, Since CUDA generation GeForce GPUs are poor at standard mining algorithms, we tried the 'new' Cudaminer for Scrypt based mining by Christian Buchner and got better results from this software

C:\guiminer-scrypt-64-004\cudaminer>cudaminer --benchmark -C 2
           *** CudaMiner for nVidia GPUs by Christian Buchner ***
                     This is version 2013-11-20 (alpha)
        based on pooler-cpuminer 2.3.2 (c) 2010 Jeff Garzik, 2012 pooler
               Cuda additions Copyright 2013 Christian Buchner
           My donation address: LKS1WDKGED647msBQfLBHV3Ls8sveGncnm

[2013-12-02 05:04:29] 1 miner threads started, using 'scrypt' algorithm.
[2013-12-02 05:04:29] Binding thread 0 to cpu 0
[2013-12-02 05:04:49] GPU #0: GeForce GTX 760 with compute capability 3.0
[2013-12-02 05:04:49] GPU #0: interactive: 0, tex-cache: 2D, single-alloc: 1
[2013-12-02 05:04:50] GPU #0: Performing auto-tuning (Patience...)
[2013-12-02 05:06:05] GPU #0:  204.80 khash/s with configuration K48x4
[2013-12-02 05:06:05] GPU #0: using launch configuration K48x4
[2013-12-02 05:06:05] GPU #0: GeForce GTX 760, 6144 hashes, 0.06 khash/s
[2013-12-02 05:06:05] Total: 0.06 khash/s
----------------------------------
[2013-12-02 05:07:08] GPU #0: GeForce GTX 760, 903168 hashes, 179.52 khash/s
[2013-12-02 05:07:08] Total: 179.52 khash/s