[vox-tech] GPU calculations

[Richard Harke]

On Tue June 13 2006 10:28, Peter Jay Salzman wrote:
> On Tue 13 Jun 06, 10:20 AM, Richard Harke <rharke at earthlink.net> said:
> > On Tue June 13 2006 08:44, Peter Jay Salzman wrote:
> > > I recently read some papers where people performed an FFT on Nvidia
> > > hardware.  The idea is that a GPU is capable of performing certain
> > > types of operations very quickly, faster than a CPU.
> > >
> > > Has anyone looked into this?  I've seen one project port FFTW to be
> > > a GPU-enhanced FFTW.
> > >
> > > Any idea on what it would take to write a "hello world" type program
> > > where 1 + 1 is thrown onto a GPU and the result is returned to a local
> > > variable?
> >
> > Why don't you look at   www.gpgpu.org
> > GPGPU -> General Purpose computing on a GPU
> >
> > Richard Harke
>
> Thanks, Richard.  Good find!  I think you do numerical computing as well --
> have you done any of this?  I've seen GPU implementations for solving
> sparse and dense linear systems -- which is, essentially, solving partial
> differential equations using implicit discretization methods.  You do this
> in your own work, don't you?
>
> In my own work with finance, I'd be interested in the FFT, high performance
> sorting and database work (which I just found in that link you mentioned!)
> I've also been thinking of extending my dissertation work, solving the
> Schrodinger-Newton PDE in 2 and 3 dimensions.
>
> I'm not entirely sure why, but every paper I've read on the subject, so
> far, uses NVidia hardware.  Need to do more reading to find out why...
>
I did some work on gravitational collapse in spherical symmetry. Needing the
symmetry to keep the computational demands doable. I read about the
GPU stuff a couple of years ago but I have never pursued it.
Richard