I have a windows XP 64 bit system with a pair of Tesla boards. I’m developing an application in C++ with a few threads that process image bitmaps in for loops. I would like to have the processing passed on to the Tesla boards if present in the target system. If not the system CPU. The C code will be complete soon. I’m very new to this and have enough time to learn to understand what is written but not originate the implementation. I would like to know if there are any developers out there that would like to take this on a small project like this on contract basis. As I mentioned its not a big task but its important that a knowledgeable professional tackles it.
If anyone is interested please give me an idea of your expertise level and how you’d like to operate. If this is an inappropriate post I apologize. Please let me know where would be a good place to put the word out.
First of you have to complete a C code of your project which should work in the CPU. Next is to convert that code to cuda kernels that should capable of running in Single/Multiple(if the problem can be split and process into multiple cards). After completing the the works you can integrate into a single application as you said. You can query for the CUDA device in your machine and give the task into it. Sample program for Device query and Setting it is in the SDK. Also if there is more than one CUDA capable card, you have to Set the devices from different threads from your application.