Yes, OpenCL is about parallel local processing, but any job that can be distributed can be parallelly processed. Main advantage that I see here is enabling an Ignite GPU node or nodes, where vast amount of data that can be processed distributatively can be routed to those nodes. For example: http://www.networkworld.com/article/2167576/tech-primers/hadoop---gpu--boost-performance-of-your-big-data-project-by-50x-200x-.html <http://www.networkworld.com/article/2167576/tech-primers/hadoop---gpu--boost-performance-of-your-big-data-project-by-50x-200x-.html>
> On May 5, 2015, at 4:37 AM, Dmitriy Setrakyan <[email protected]> wrote: > > On Mon, May 4, 2015 at 7:27 PM, Marko Jevtic <[email protected]> wrote: > >> Are there any current ideas about adding OpenCL to Ignite framework ? Like >> enhancing current computations with ability to execute on GPU if applicable? >> >> > I could be wrong, but I thought that OpenCL was all about parallel local > processing across multiple cores and CPUs. Do you think there is a use case > for distributing computations there? If yes, wouldn't users just use JOCL ( > http://www.jocl.org/) to interact with OpenCL jobs from Ignite jobs? Or do > you think additional integration is possible here?
