given the broad support of *NIX platforms based on the gcc compilers it is saddening if not ridiculous that NVIDIA does not support gcc for Windows. Also looking around here and in other forums the non-support of gcc/MinGW on Windows keeps developers from their main objective which can’t be what NVIDIA wants. Given the existence of Mingw-w64 toolchains exist for both bitnesses, and from my experience including the compilation of Qt, boost, and Mono those compilers just work as their *NIX counterparts. Further, there is no other toolchain with such a broad support of languages - and sometimes another language than C/C++ is just what you need just like in one of NVIDIAs showcase 3d party tools MAGMA which has no windows port because of this compiler trouble.
If you want to have MinGW/gcc supported on windows … drop a comment, I might not change very much but at least we can try to make NVIDIA reconsider their decision.
I would love this. Having just spent a day trying to get a simple application to compile in cygwin with the Microsoft compiler, the thought of trying to port my large scale applications to Windows fills me with dread.
I would like to let the GPU’s of my NVIDIA graphic card work for me (to be concrete: I am planning to employ it for astronomical ephemeris calculations, which seem to be a good job for a GPU, involving many sin() and cos()), BUT I don’t like Microsoft Visual C++. I prefer gcc under MinGW on my Windows computer. I therefor strongly vote for CUDA’s MinGW gcc support!
I can only tell for a few math departments at universities that we didn’t pick CUDA due to the incompatibility on Windows with MinGW.
So yes, it would be nice to see, hence I believe it would already be too late. There are other similar powerful alternatives meanwhile that support MinGW.
Second the need for mingw support. Visual studio has its issue stopping it for production code, resorting to a Visual STudio C DLL interface to our mingw C++ program is a real pain!
I use Windows at home because of its office suite and the fact that I only have one computer. But at work, everything runs on *nix. If I can get my project running at home under gcc, then I know I can get it running at work. Ergo I avoid Visual Studio. It would be wonderful if NVIDIA extended support to MinGW.
I am a developer for blender 3D and native MinGW-w64 support would be helpful for us as well. Just for the record, it is possible to compile CUDA binaries for a program compiled with MinGW but you need to have MSVC installed, run the compilation from a MSVC environment and load the libraries dynamically, which sort of defeats the purpose.
I am a game developer who wants to make his own games for multiple platforms, in native c++, and cannot use your gpgpu optimizations without support for minGW. Please NVidia, please do the right thing. I can only use OpenCL for AMD as that is my only option. My games won’t have acceleration for NVidia graphics cards. Imagine all the possibilities for technology to advance.
We cannot compile on Visual Studios Express because it isn’t free for commercial use. Many of us are indie devs and software developers who are on a budget. Please NVidia, we BEG you!
YES - I have to work from home using cygwin with windows 7 laptop - use cl.exe as the windows compler. Really must avoid MS as the applications must port to linux - will never be ran on a widows OS, ever
Come on NVIDIA - remember not all of the world are game developers developing commercial products!