Double Precision code on Single Precision GPU

Accelerated Computing CUDA CUDA Programming and Performance
1

This is most likely a ridiculous n00b question, but I’ll ask it anyway.

What happens when I run a double precision CUDA kernel on a single precision-only GPU?

Will the initialization fail and/or get an error?? [because I believe my code is correct… and this is what’s happening]

2

If you compile with -arch sm_13 and then try to run on a Compute 1.0 or 1.1 GPU, then yes, I believe it will just fail to launch the kernel or maybe even earlier.

3

I’m using the latest production release of CUDA 2.0 [just released a few weeks ago]. I recompiled with the architecture specification, same result. The code actually runs, but simply fails to launch the kernel.

Thanks for the help.