I’ve recently started using a Quadro 6000 to run GPU-accelerated hashing applications. I’m looking to purchase additional hardware and notice that the GTX 690 has many more CUDA cores available than the Quadro 6000 (3,072 on the GTX 690 vs 448 on the Quadro 6000).
My question is, will this alone provide more computing performance or is the Quadro better suited to GPU processing, seeing as it is a workstation product? Will using the GTX 690 drastically improve the speed?
Does your application scale to multiple GPUs? The GTX 690 is two CUDA devices on one board, so if your application cannot use multiple CUDA devices at once, you will definitely see no benefit to a GTX 690.
Keep in mind that the GeForce 690 is one generation newer than the Quadro 6000, AND is a dual GPU setup, so in almost all cases, the 690 will be significantly faster.
This is not true, first if the application is not developped to use two GPU, the GeForce GTX 690 will be used as a downclocked GTX 680.
Then, there is usage of Double-Preceision floating point arithmetic, where the Quadro 6000 deliver more than 500 Gflops DP, while each downclocked GK104 on GTX 690 delivers half the DP!
Third there is the lack of registers and fewer L1-cache per thread, that impair GPGPU performance of GK104 on complex algorithms.
So it all depend on the profile of the GPGPU application you plan to run, if it support multi-gpu setup, don’t use Double-Precision FP, and don’t need too much registers or cache per threrad, it might be faster on GTX 690 than on Quadro 6000.
does anyone know if it’s wise to buy a tesla c2070 or the nvidia quadro to run after effects and final cut pro or an you guys tell me which is the best card for my project i already am using the quadro 4800 for mac
