From a general .NET perspective, with the 4.0 release, I’d like to start an open discussion about .NET integration with CUDA and/or OpenCL where available. I haven’t seen much discussion about this and what I have seen introduces new classes and methods rather than embracing what is already available.
Ultimately, I have started this topic as I have seen the benefits of CUDA, but I have found it difficult to get my company to embrace its benefits in a company of .NET developers. I am interested in what exactly the disconnect is between .NET and CUDA --> In other words --> If I were to write code using asynchronous and parallel methods that are consistent with the CUDA model, what are the limitations to porting this code directly to CUDA via translation rather than requiring the developer to learn a separate set of methods and classes to do this?
Forgive my ignorance if the answers to this question should be obvious. Only through awareness can I understand – I have tried searching the net and forums for similar discussion topics, but I have not found this topic having been discussed.
“Simple Solutions, Sound Logic”