comp.lang.idl-pvwave archive: archive » Re: type conversion in GPULIB

Home » Public Forums » archive » Re: type conversion in GPULIB

Show: Today's Messages :: Show Polls :: Message Navigator
E-mail to friend

Re: type conversion in GPULIB [message #79586 is a reply to message #79584]

Fri, 09 March 2012 10:09

Michael Galloy
Messages: 1114
Registered: April 2006

Senior Member

On 3/9/12 3:39 AM, alx wrote:
> On some computer equipped with a CPU/GPU combination (8 cores intel
> and Tesla C1050), I have a fast ADC which repeatedly delivers large
> int16 arrays to CPU memory. I would like to process them "in the fly"
> by using the GPU and GPULIB software. Functions to convert float and
> complex of various sizes are available in GPULIB.
> Is it a way to convert int16 arrays to float arrays in GPU memory by
> using GPULIB (I would like to get faster conversion and smaller bus
> data transfer as well) ?
> Thanks for any insight.
> alx.

Unfortunately, GPULib currently only supports floating point types.
Right now, you would have to convert the ints to floats on the CPU
before transferring to the GPU.

There is no inherent need for this type restriction from CUDA, I would
like to extend the types to include the various IDL integer types at
some point.

Mike
--
Michael Galloy
www.michaelgalloy.com
Modern IDL, A Guide to Learning IDL: http://modernidl.idldev.com
Research Mathematician
Tech-X Corporation

Report message to a moderator

[Message index]

		Re: type conversion in GPULIB By: lecacheux.alain on Sat, 10 March 2012 03:22
		Re: type conversion in GPULIB By: Michael Galloy on Fri, 09 March 2012 10:09

Previous Topic:	Re: Question about storing arrays in pointer array.
Next Topic:	read a C written binary file with IDL

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Sat Oct 11 09:46:49 PDT 2025

Total time taken to generate the page: 0.72145 seconds