Hello,
I am considering to buy a GPU for parallel computation research. I am interested in using dynamic Parallelism feature.
However, Tesla devices currently beyond my budget.
Some GeForce have Kepler architecture but not sure if they have this feature… Looking for the most cost-effective solution…
can you help me…?
thank you…
A GPU with compute capability 3.5 or higher will support dynamic parallelism.
I have a GeForce GT 640 which is cc 3.5 and can be purchased for less than $100.
GT 740 is another low cost option:
[url]http://www.amazon.com/gp/product/B00KJGYS4Y/ref=pd_lpo_sbs_dp_ss_3/178-3292229-6638457?pf_rd_m=ATVPDKIKX0DER&pf_rd_s=lpo-top-stripe-1&pf_rd_r=0QHS9BAAWM2RHNDWC20T&pf_rd_t=201&pf_rd_p=1944687642&pf_rd_i=B00IDG3NDY[/url]
GeForce GTX 750 (cc 5.0) is another low-cost option.
You can get an idea of available GPUs and their compute capabilities here:
[url]https://developer.nvidia.com/cuda-gpus[/url]