Advanced search

Message boards : Wish list : GPUgrid for new Jetson TK1

Author Message
Send message
Joined: 21 Mar 10
Posts: 23
Credit: 861,667,631
RAC: 0
Scientific publications
Message 36626 - Posted: 24 Apr 2014 | 22:10:56 UTC

Apr 23's Nvidia Newsletter included an opportunity to buy a small System based on a TEGRA cpu (quad-core ARMĀ® Cortex-A15 CPU) with 192 core nVidia GPU. I went ahead and ordered one to go with a small ARM-A9 system that I already have.

Does the project either have an interest in trying to obtain one (thinking it could be a small development platform for larger systems in the near future), or could I offer to obtain/download source code and test data to attempt to compile for the team? Based on what I've seen, it will require the new cuda 60 version.

GPUGRID Role account
Send message
Joined: 15 Feb 07
Posts: 134
Credit: 1,349,535,983
RAC: 0
Scientific publications
Message 36627 - Posted: 24 Apr 2014 | 22:42:16 UTC - in response to Message 36626.

We've got the older Kayla, for which we had working application.

Might be able to put it out on GPUGRID, but not imminently..


ExtraTerrestrial Apes
Volunteer moderator
Volunteer tester
Send message
Joined: 17 Aug 08
Posts: 2701
Credit: 1,311,122,549
RAC: 8,302
Scientific publications
Message 36634 - Posted: 25 Apr 2014 | 10:40:59 UTC

It's got a single SMX at probably moderate clock speed. I'd assume performance in the range of 1/3 - 1/2 of a GT640. This will make it struggle to return short run jobs within a day. So judging by previous GPU-Grid history there's not much use for such hardware here, especially if it requires extra coding effort.

However, if you do have a good use case for such systems... might I suggest you look into using lower-end PC GPUs first? The hardware is widely available, your tool to use them are established and we regularly tell people "sorry, too slow to be used here".

Scanning for our furry friends since Jan 2002

Post to thread

Message boards : Wish list : GPUgrid for new Jetson TK1