[*] The second has to do with reading constant memory. For some reason in some of my kernels this fails, read out zero or other weird values, unless I prefix the kernel with __synchronizeThreads();
This one is kind of arcane and I’m not sure exactly when and why they occur. All I did was create a minimal program that still demonstrated the bug to NVidia.
[*] Under Linux, when you use CUDA in a shared library, and the program terminates you sometimes get a nasty segmentation fault.
A great idea! Considering that we are somewhat beta-testers here I think the nVidia bug database should be publicly available. This would only be fair.
It took me 2 days to figure this out but I’m experiencing a similar thing when I copy a struct to constant memory and then try to access it from within a kernel. And as my plan was to pass parameters to the kernel with this struct half of the time it crashed my workstation because some values are really off… really really frustrating :(