This is very strange.
I have 2 versions of some code, one slightly older than the other. On the newest code, I have it so that the number of threads per block is initialized by some deviceProp.maxThreadsPerBlock variable. Everything works fine.
Turns out, my older code performs better than my newer code, so I’m making some changes on it to make it more usable to any GPU card, not just my own. I added in these 2 lines:
[codebox]cudaDeviceProp deviceProp;
getDeviceProperties(deviceProp);[/codebox]
…and I get the error: “error: identifier “getDeviceProperties” is undefined”
Anybody know why?