How to Make GPU Hardware Decisions

Scott Computing Trends, CUDA, Hardware & Infrastructure, OpenCL Leave a Comment

We get questions all the time about how to make GPU hardware decisions. We’ve seen just about every scenario you can imagine, and so we always jump at the chance to help others through this decision process.

Here’s a recent question from a customer.

“I’ve just found your post on Analytic Bridge and have taken a look at your website … I’m replacing my two Tesla M1060 cards (computing capability too low) and I’m considering used Tesla M2070s or the new GTX 760 cards. Could you offer any insight? I believe the GTX 760 cards may well outperform the older 2070s and are much cheaper.”

And here’s our response.

“The GTX 760 will probably outperform the M2070 for single precision FLOPS but not for double precision, if that matters to you. Not sure what your budget is, but there are other GeForce GPUs that are a lot cheaper than Tesla that have a control panel option to sacrifice clocks for higher double precision throughput (see GeForce GTX Titan).”

We’ve found that there’s a lot of confusion out there regarding what’s the best compute hardware for people who are restricted within a certain price range. A few key takeaways should be considered when trying to make the best GPU hardware decision. Here are some best practices you should consider.

1) Do you want a reliable GPU that does fast double precision calculations? Choose the Tesla K20, or the K40.

2) Do you want a reliable GPU that does fast single precision calculations? Tesla K10 is for you.

3) Are you doing a lot of visualization or video rendering along with compute? Choose the Quadro line.

4) Are you an enthusiast or not worried about double precision and ECC memory? Buy a GTX GPU within your budget.

If you are working with OpenCL, then AMD, Intel, and other devices come into the fray as well. We will talk about that another day!

Do you have any questions about GPUs? Contact us!

Leave a Reply

Your email address will not be published.