[PlanetCCRMA] sound processing in GPU w/ Nvidia CUDA ? (was Re: fm synthesis software?)

[Niels Mayer]

On a slightly related note, are there any C//C++ based sound generation
tools that have been re-hosted to work using a spare graphics processor
unit, for example using Nvidia's CUDA:
http://blogs.nvidia.com/ntersect/2010/03/gpgpu-developers-get-boost-from-new-cuda-toolkit-30.html
.
Although most people think of CUDA for only high-end processing, the CUDA
compatible GPU's are rapidly coming down in price as they benefit from
economies of scale. This means in the future we'll have the equivalent
processing power of Native Instruments Core available in a $50.00 graphics
card.

GPU processing for sound via CUDA has already been done a little bit in the
windows/mac world:
http://www.acusticaudio.net/modules.php?name=Products&file=nebula3
http://www.kvraudio.com/forum/viewtopic.php?t=222978
http://www.kvraudio.com/forum/viewtopic.php?t=240824

What about Linux? Certainly, Linux HTPC's have already benefited from
GPU-processing as relatively low-end CPU's can now decode H.264 video using
http://en.wikipedia.org/wiki/VDPAU . Despite low-power CPUs, such decoding
can appear "real-time" and without glitches, due to offloading the
real-time-intensive "DSP" computation to the graphics card. Such GPU
decoding of video is now available in $50-range GPU's like the 9500GT.

 Wouldn't it be nice to have jconvolver or softsynths doing their "grunt
work" in a GPU and not loading up the CPU? Also, with real-timey-DSP
processing offloaded to the GPU, the CPU can better keep up with the
interrupts and GUI processing.

-- Niels
http://nielsmayer.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://ccrma-mail.stanford.edu/pipermail/planetccrma/attachments/20100405/fc0b9c61/attachment.html