How people are devloping applications in C and CUDA? integrating java and CUDA-C


I have a very basic question. I understand that developing application in C is really time consuimng and a tough job. Thats why people develop in C++, Java , .NET etc. Then how are CUDA developers developing application in C using CUDA. Are they integrating different modules written in different language. Like they would write a cuda for computation using C-CUDA and the rest of the code is writtent in java, and they will combine it? By the way how do we integrate diffrent modules written in differenet language. such as a module written in CUDA and another module in Java, how do we combine it?


Well, I don’t know about Python, Java, or FORTRAN, but I use CUDA from .NET (C#) and write my kernels in C, compile them to cubin or PTX and use p/invoke with the driver API to run my kernels.

JNI should help, I thought…

thanks profquail!! Can u tell me how do u compile cuda code in cubin or Ptx ? and what do mean by “p/invoke with the driver APLI”