4
$\begingroup$

This question asked how to target an Nvidia GTX 1080 with MMA 11.0.1 (which is also my current version) and from the comments it is clear that the user was then able to use this card successfully even though it is not [22/04/2017] on Wolfram's list of supported hardware.

How can one determine reliably whether a particular GPU is in fact compatible with (specifc versions) of Mathematica and/or CUDAlink?

I intend shortly to build a new Windows (10) based PC with a GeForce GTX 1080 Ti and would like certainty of usability first, but the currency of the supported hardware lists and actual usability of specific cards is surely a more widespread interest as HPC becomes (relatively) more affordable.

Improve this question
$\endgroup$
0

1 Answer 1

Reset to default
3
$\begingroup$

ALL Nvidia GPUs with Compute Capability >=3.0 are supported for neural networks. However, you're out of luck with great hardware and a buggy version of Mathematica. 11.0 had an issue with all 10xx cards, which was fixed in 11.1. Besides, SO MUCH has been added to neural networks in this version that is really worth the upgrade. If you want to use CUDALink for development, then ALL CUDA and OpenCL GPUs are supported. Just make sure you have the absolutely latest drivers.

Improve this answer
$\endgroup$
1
  • $\begingroup$ Thanks! But it leaves me wondering how user in the referenced question succeeded (though that's unanswerable and OT here) It also completes my justification for upgrading, but the general question of what hardware works with which MMA etc. versions remains - how is one to know such things without relying on knowledgeable suers such as you? $\endgroup$
    – Julian Moore
    Apr 24, 2017 at 6:32

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.