Do not input private or sensitive data. View Qlik Privacy & Cookie Policy.
Skip to main content

Announcements
See why IDC MarketScape names Qlik a 2025 Leader! Read more
cancel
Showing results for 
Search instead for 
Did you mean: 
Anonymous
Not applicable

Using GPUs?

Has anyone done any work on modifying components or building components to work on GPUs? Possibly using something like JCuda?
http://www.jcuda.org/
I'm wondering on what use cases you had for it? It seems that GPUs can provide solutions to heavily cpu dependent processes, and indeed you can order a GPU server on amazon aws. We do have massively heavy CPU processes so it may well work nicely for us without having to have loads and loads of servers!
Dan
Labels (3)
3 Replies
Anonymous
Not applicable
Author

I guess a typical ETL job does not need such kind of heavily CPU power, but they need much more IO.
I would think, reporting tools like Jasperreports which renders reports need much more CPU power the the data transfer programs and they could take advantage from GPUs.
Anonymous
Not applicable
Author

Agreed; Typically ETL isnt massively CPU bound; However I have a process which is, and is doing all sorts of advanced analytics/statistics. So it could be these would help us - seems it could be a lot of work to prove the point though.
Anonymous
Not applicable
Author

probably the easy way to do it would be to write a java routine that uses jcuda-- then call it in a tJavaRow or tMap as needed.
bottom line-- you want jcuda, you'll probably have to write it yourself 0683p000009MACn.png