CUDA debuging modes

Would anyone please explain me the differences among

debug
emudebug
emurelease
release

Thanks in advance

“debug” and “release” hold the same semantics as any other VC project. THe compiler generates efficient object code for “RElease” than “Debug”. Now, I dont know if this affects “NVCC” as well.

emu* configurations relate to compilation in “Device Emulation” mode. I hope you know what is “Device emulation” mode.

If some1 can tell , if this choice affects “nvcc” as well – it would be GR8! Is the PTX code generation optimized in “Release” mode???

Debug and release builds run on the GPU whilst the emu version run on the CPU.

All GPU calls are translated to CPU calls.

In my opinion emudebug and emurelease don’t really help anyone because a program that works in these builds isn’t guaranteed to work in the debug and release builds.

Cheers