Compute Capability 2.1 DEFINITION in cuda-faq, PLS! Compute Capability 2.1

Compute Capability 2.1 exact DEFINITION in cuda-faq, PLS!

Don´t you find it a bit embarrassing that I stumbled over the Compute Capability 2.1 specs in - and not in the appropriate FAQ-section, in the docs (There? But hidden in how many pages?) or in this forum?

Architecture specifications - just the differences 2.0 / 2.1

Number of cores for integer and floating-point arithmetic functions operations 32 / 48

Number of instructions issued at once by scheduler 1 / 2 where the first scheduler is in charge of the warps with an odd ID and the second scheduler is in charge of the warps with an even ID.

That seem nice incremental improvements and I´m glad my brand new Tosh with a GeForce 525M has got it!

Interestingly the referenced http://www.geeks3d.c…parative-table/ article, apparently from 6/6/2010 is outdated too, though it says “Updated on June 9, 2011” and only tells about 2.x.



Just FYI: Compute capability 2.1 is defined in Appendix F.4 (Titled “Compute Capability 2.x”) in the CUDA Programming Guide.

Thanks, Seibert,

I guessed ´t was hidden in the pile of docs somewhere, but somehow Windows´ latest über-drüber local search engine doesn´t cut it.

Sorrily the NVIDIA online version (FAQ) I did find is outdated and that should be corrected.