offloading tasks from CPU and GPU

[Agnès]

Hi,

I have a pandaboard ES (omap 4460) running under Ubuntu 12.04.

I have an image processing projet to do and I would like to offload tasks from CPU to GPU to improve performance and reduce workload on the two cpu cortex A9.

How can I do it (communicate between this two cores)??

Which API I must use to successfully do that.

Thank you for any reply.

[Tom Mitchell]

The GPU is not an ARM processor so you need a lot more.

There might be CUDA support but if there is not it is a lot more difficult.

There is no public document on the GPU that I know of and TI no longer

has any ongoing support for the Pandaboard.

Do look for CUDA support ... I may have missed something.

--
You received this message because you are subscribed to the Google Groups "pandaboard" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pandaboard+...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

  T o m    M i t c h e l l

[Agnès]

Thank you for your reply

But what I know, is that CUDA support only Nvidia processor

What about opencl??

I have seen a demo in where the powervr has running an opencl image processing algorithm.

Also, if these two progrmming APIs (cuda and opencl) are not supported, there isn't another method to ensure the offloading tasks??

Thank you for your help

[Agnès]

Please I need help 

Can someone give me an idea on how offload certain functions from the CPU to the powervr GPU to speed them up??

thank you

Le lundi 16 février 2015 18:12:03 UTC+1, Agnès a écrit :

[menon.n...@gmail.com]

You probably want to talk about openCL support than CUDA.

[Tom Mitchell]

On Tue, Feb 17, 2015 at 12:36 PM, menon.n...@gmail.com <menon.n...@gmail.com> wrote:

You probably want to talk about openCL support than CUDA.

 

Yes... 

The library support is hardware specific...

I looked and I see nothing for the pandaboard with "sudo aptitude" and friends.

Other searches found a bit of interest about 2012 that hit the NDA documentation brick wall.

I looked for both CUDA and openCL.

There may be some daylight on the Raspberry Pi side of the low cost community.

The new quad core Pi might prove to the ticket.

http://petewarden.com/2014/08/07/how-to-optimize-raspberry-pi-code-using-its-gpu/