HN Explorer

Ask HN:GPU to Do Numerical Computations as a Hobby

2 points by newprint ↗ HN
My hobby is Numerical Computations, however, I am doing everything on Intel CPU using C++. I am thinking of buying a separate GPU to run numerical computations on it and learning OpenCL. What would be reasonable choice with the budget of $1000. Unfortunately, Tesla is out of my budget. Thank you.

2 comments

[ 3.1 ms ] story [ 13.9 ms ] thread
dekhn
The GTX 1080 is currently your best bet. It has the best price:performance ratio. The 1070 might also work, although it has slower cores and memory clock. I use a GTX 1080 for TensorFlow, it's ~100X faster than the CPU for training. I started with a 1060, which was way cheaper, but upgraded to the 1080 for more RAM and faster/more CUDA cores.

If you can increase your budget and your motherboard supports it, you can get two 1080s, many things will parallelize across multiple cards. I know people who put 8 cards in a single machine and barely use the CPUs.

BillBohan
When I looked at price and performance, a pair of 1070s cost only slightly more than a 1080 and gave me more RAM and cores. Even though the clock on a 1080 is 15% faster you can get more done with 50% more cores.