Distributing Cuda Drivers with Software
I have a few question about distributing CUDA software.

Is in necessary to install the dev driver on computers that run the software? If so should I make users download and install from nvidia or is there a better way?

Does nVidia have any suggestions on checking/installing drivers when you write an installer for your own program?

If anyone is distributing their CUDA programs I would be grateful to hear how you deal with nvidia driver requirements.
I have a few question about distributing CUDA software.



Is in necessary to install the dev driver on computers that run the software? If so should I make users download and install from nvidia or is there a better way?



Does nVidia have any suggestions on checking/installing drivers when you write an installer for your own program?



If anyone is distributing their CUDA programs I would be grateful to hear how you deal with nvidia driver requirements.

#1
Posted 01/08/2012 01:45 AM   
Short Answer, you don't need any other driver than the nVidia graphic driver, that contains CUDA and OpenCL drivers. So for CUDA you just have to distribute your application, it will run out-of-the-box on any cuda-enabled PC (meaning a CUDA-capable nVidia GPU with recent nVidia drivers).
Short Answer, you don't need any other driver than the nVidia graphic driver, that contains CUDA and OpenCL drivers. So for CUDA you just have to distribute your application, it will run out-of-the-box on any cuda-enabled PC (meaning a CUDA-capable nVidia GPU with recent nVidia drivers).

Parallelis.com, Parallel-computing technologies and benchmarks. Current Projects: OpenCL Chess & OpenCL Benchmark

#2
Posted 03/09/2012 04:37 PM   
This is incorrect. If your application is using the CUDA runtime, you will need to package the runtime library.
This is incorrect. If your application is using the CUDA runtime, you will need to package the runtime library.

#3
Posted 03/09/2012 08:03 PM   
Scroll To Top