Thursday, July 22, 2010

CUDA 3.1 out

Just read that cuda 3.1 toolkit is available (has been for some time in fact)

http://www.theregister.co.uk/2010/07/22/nvidia_cuda_parallel_nsight/

As you can see from the release notes, CUDA 3.1 also gives 16-way kernel concurrency, allowing for up to 16 different kernels to run at the same time on Fermi GPUs. Banks said a bunch of needed C++ features were added, such as support for function pointers and recursion to allow for more C++ apps to run on GPUs as well as a unified Visual Profiler that supports CUDA C/C++ as well as OpenCL. The math libraries in the CUDA 3.1 SDK were also goosed, with some having up to 25 per cent performance improvements, according to Banks.


The support for recursion and concurrent kernels should be great for CUDA path tracers running on Fermi and I'm curious to see the performance gains. Maybe the initial claims that Fermi will have 4x the path tracing performance of GT200 class GPUs could become true after all.

No comments: