Develop Tegra K1 Cuda application

Hi all,
I’ve read about the new Tegra K1 chip, supporting CUDA programming.
I think it’s very interesting technology and I want to invest some time learning it but I need some beginner’s information.
What is the best way to develop CUDA application for Tegra K1? What does I need in terms of
software, adapters (like the SECO one), etc? May someone suggest me books or tutorial on this topic?

Thank you in advance and have a nice day.

I think the best proxy for the K1 is to obtain an sm_35 (sm_32?) GPU like a GT630 or GT635. Or maybe the rumored GM108 in a week or two.

The next step is to choose a traditional CPU problem that interests you and see how efficient you can make it on your GPU.

The CUDA C Programming Guide is the best place to start learning about CUDA. After that books like Nick Wilt’s Cuda Handbook are good for understanding what is going on “under the hood” of the APIs.

Thank you for your answer, Allanmac.
My problem is that I’m able to develop CUDA software on ordinary GPU for PC but
I don’t know the difference between PC and mobile approach.
For example, I want to make my first app for a mobile with Android and a Tegra K1 chip.
How can I achieve that?

Thank you!

Presumably when an Android, WinRT or Linux ARM device ships then CUDA headers and libraries will be made available.

The SECO board is probably as close as you can get at this point to a pure ARM dev experience. I’m guessing that will change by mid-year.

My point about sticking with the PC is that you can be more productive compiling, debugging and measuring today on a small GPU that is similar to the K1. And when the K1 finally arrives you’ll be ready.