[SOLVED] GPU for parallel computing

[AndrzejBo]

I need GPU for parallel computing neural networks.
I wonder about GTX and RX, both have CUDA and OpenCL?
For example between Nvidia-GTX-1060 and AMD-RX-590
https://gpu.userbenchmark.com/Compare/Nvidia-GTX-1060-3GB-vs-AMD-RX-590/3646vs4033
RX is cheaper, is good for computing?
Maybe better : GTX 2060 instead of GTX 1060 and ??? instead of RX 590.
What instead 590 a bit faster?

 

[Newtonius]

Depends on how much computing you will be doing. I would go for an RTX 2060 Super, they are much easier to overclock to get more performance and come with much faster VRAM which helps data encoding/computing.

Though these are gaming GPUs. NVidia has compute GPU's like the quadro series. But you seem to be on a budget. So the 2060 Super would be ideal, but an RX 590 would be just as fine.

 

[AndrzejBo]

After months, what is best (and price accessible) hardware for deep learning?
I search RTX 2080 and even RTX 2080 Ti, are better? not necessary GPU and maybe specialized AI hardware?