I am a developer and I wrote an application in Pascal language, which requires to calculate the value of 20 parameters to optimize a result.
Usually, the best combination for a sub-set of 3 to 4 parameters only is obtained after 1 and 10 million loops where I scan multiple alternatives.
Each loop realizes roughly 100 different computations to reach a result.
This is typically a very repetitive process for which I believe an NVIDIA card should be particularly well adapted.
My actual NVIDIA card does not offer the HIgh-Performance Computing option. (please see above my card type)
I have two questions:
- Could you recommend me a card adapted for the job above,
- One potential drawback I am facing to take advantage of the full potential of an NVIDIA card is the language I am using for the application, Pascal. Do you think that a transcoder solution (Pascal-> Python or Pascal -> CUDA) exists where the NVIDIA card will manipulate Python requests or CUDA requests instead of a Pascal executable?
I am prepared to invest in any hardware NVIDIA solution which will lead me to a significant computation speed.
Thanks for the help.