Laptop for college student

Accelerated Computing CUDA CUDA Programming and Performance
1

Hi, first post. I’m shopping for a cuda capable laptop for my son to get him introduced to cuda programming. Using the following link:

I can check the specs (especially # cuda cores) for any desktop card, but for some reason, the laptop GPU specs (e.g. GT 740M) don’t list the # cuda cores. Why is that? How is one suppose to adequately compare the laptop GPU’s?

2

The ultimate (and overwhelming) reference for NVIDIA GPUs is here:

Any new NVIDIA GPU made in the last 4 years will support CUDA, though I would aim for something running the Kepler architecture if possible. One thing you will want to make sure is that the GPU has at least 1 GB of memory so that there is enough memory to run both a GUI desktop and have room for CUDA programs at the same time. Beyond that, # of CUDA cores won’t be so important for learning purposes.

3

I am quite happy with a laptop equipped with an nVidia GT 750M. It has 4 gigs of VRAM and is plenty fast. The 750M has 2 SMX, meaning 2 x 192 CUDA Cores.

Newer GPU models like the GTX 860M will already incorporate nVidia’s Maxwell architecture, which has a couple of additional features for CUDA programming. This one has 640 CUDA cores.

4

Thanks for the comments! Looking at the chart, it seems the # of Unified Shaders (xx in the core config xx:yy:zz) corresponds to the number of cores in the Nvidia documentation. The GLOPS column is great. Just what I need.