how to add "emudebug" configurate choice

I am a new guy to CUDA, now I encounter some questions that puzzles me a lot.

It’s how I can add “emudebug” configurate choice when I built a new project in VS2005.

More general, how to configurate a programming enviroment for CUDA under VS2005.

Hi and welcome,

In the CUDA SDK there is an template example. I think its best to start from the template. If you use this template you can select what build configuration you want to build. And I think there is also an option for emudebug, I can’t check it now because I only have a Linux workstation here and @home my Windows.

Thank you for your answer.

But I think it’s better to configurate mu own CUDA programming enviroment. The method you say I have tried.

That is also a very strange phenomenon: I just change the sample prject to another directory. It can’t work and throw the error “can’t find cutil.h”. That really bothers me.

If you or somebody can give me the answer, I’ll very appreciate.

That is a normal error. the cutil.h is an specific file of the CUDA SDK so not from the toolkit. I don’t know how they set the build rules in VS 2k5 but it has to be there.

I checked the CUDA WINDOWS SDK on my linux I opened the .vcproj in a texteditor and saw something like

AdditionalIncludeDirectories="$(CUDA_INC_PATH);./;../../common/inc"

AdditionalLibraryDirectories="$(CUDA_LIB_PATH);../../common/lib"

I think its one of those.

This is in the project/template directory.

And if you never did anything with CUDA I think it’s best to begin from template because its kinda hard to make your own CUDA project while not knowing what you are doing.

Yeah, but it is going to be very time-consuming and error-prone to try and setup all the custom build commands for every project you create… If you really want to do this, you need to look at what the SDK projects do in the custom build commands section of the .cu files and recreate that.

Or, you could take the easier route and use the CUDA build rule: http://forums.nvidia.com/index.php?showtopic=30273 (disclaimer, I’ve never used it so I don’t know how well it works).

Another option is to use CMake and FindCUDA.cmake http://forums.nvidia.com/index.php?showtop…482&hl=findcuda

It’s probably a little different than what you are used to, but the plus side is that then your app can easily be compiled on linux/windows/mac without any fuss.

Thank all of you. I will try your suggested method.

Now that you mention it, I think I will also try the last method on my windows Machine at home. Because my Make abilities are far more developed than the VS2k5 build abilities. And that way I can easy try it on other computers without have to think of what I need to do to change from VS to Makefile