I can't run Cudafy without Emulator on CUDA SUPPORT NVIDIA GPU !

Feb 18, 2015 at 8:24 PM
Hi every one !
I have Cudafy V1.28 + CUDA Toolkit 6.5 + VS Ultimate 2013 + Windows 8.1 x64
I am trying to run this program :
https://cudafytsp.codeplex.com/SourceControl/latest#CudafyTsp/CudafyTsp/Program.cs

But I get this runtime error :
An unhandled exception of type 'System.Runtuime.InteropServices.SEHException' occured in Cudafy.NET.dll
on line 262 :
GPGPU gpu = CudafyHost.GetDevice();
I tried changing that line into this line :
GPGPU gpu = CudafyHost.GetDevice(CudafyModes.Target, CudafyModes.DeviceId);
and adding these three lines in the Main () function :
CudafyModes.Target = eGPUType.Emulator;
CudafyModes.DeviceId = 0;
CudafyTranslator.Language = CudafyModes.Target == eGPUType.Emulator ? eLanguage.OpenCL : eLanguage.Cuda;
it makes it so slow that takes 136 seconds to complete GPU tsp . (while CPU tsp takes 2 seconds and MPU tsp takes 0.5 seconds .)

If I change those emulators to Cuda like tihs :
CudafyModes.Target = eGPUType.Cuda;
CudafyModes.DeviceId = 0;
CudafyTranslator.Language = CudafyModes.Target == eGPUType.Cuda ? eLanguage.OpenCL : eLanguage.Cuda;
I get the same runtime error but whith a message on console in no debugging mode :
Unhandled Exception: GASS.CUDA.CUDAException : ErrorNotInitialized
at GASS.CUDA.CUDA.set_LastError(CUResult value)
at GASS.CUDA.CUDA.get_Devices ()
at GASS.CUDA.CreateContext(Int32 ordinal,CUCtxFlags flags)
at Cudafy.Host.Cuda..ctor(Int32 deviceId)
at Cudafy.Host.CudafyHost.DoCreateDevice(eGPUType target, Int32 deviceId)
at Cudafy.Host.CudafyHost.CreateDevice(eGPUType typeInt32 deviceId)
at Cudafy.Host.CudafyHost.GetDevice(eGPUType type, Int32 deviceId)
at CudafyTsp.Tsp..ctor in Program.cs:line 266
at CudafyTsp.Program.Main() in Program.cs:line 44
Also the native example projects of Cudafy which is included in CudafyByExample folder needs Emulator to run . Without Emulator I get this message on console :
No suitable CUDA devices found
What can I do to get Cudafy run on mu GPU ?
I can run CUDA without problem .
Thanks ! :)
Feb 22, 2015 at 1:31 PM
HI Everyone ! I'm the guy who asked the question . The problem was not including the bin folder of VC in PATH . Sorry :)