Is this typo on Compute Capability page?

Is it correct that 910m’s Compute Capabilit is 5.2 and 920m is 3.5? Should it be 3.5?

Hello,

Thanks for posting in the NVIDIA forums. I have reached out to the owner of the page to verify if this is a typo or not.

I will update here when I have new information.

Cheers,
Tom

There are several indications online ( 1, 2, and others) that at least one variant of the GeForce 910M is a Fermi device based on the GF117 chip. It’s possible that there were other variants as well, from here:

Actual implementation may vary by OEM model. Please refer to OEM website for actual shipping specifications.

If it is a Fermi class GPU/GF117, the compute capability is 2.1 and support was dropped in CUDA at the CUDA 9.0 generation.

It’s entirely possible that other variants were shipped, possibly with different compute capabilities. You could load CUDA 8.0 on the system, and then run deviceQuery sample app, to confirm the compute capability of a specific 910m example.

I cannot identify conclusively whether there is a typo, because I don’t know if there were other variants shipped.

Some pages list a GK208/GK208B Kepler chip instead or as second possible model:

Then it would be compute capability 3.5.

5.2 (Maxwell) definitely seems to be wrong. Its strange compute capability is shortly mentioned here:

A GK208 variant would be compute capability 3.5, for OP’s reference.