| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Help Conquer Cancer
Thread: Help Conquer Cancer September 2011 Update |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 73
|
|
| Author |
|
|
rilian
Veteran Cruncher Ukraine - we rule! Joined: Jun 17, 2007 Post Count: 1454 Status: Offline Project Badges: 
|
i'd like to know if that would be separate phase of project, or we will crunch till the end current tasks and that's all
----------------------------------------would be results of CPU and GPU calculations compatible ? |
||
|
|
BladeD
Ace Cruncher USA Joined: Nov 17, 2004 Post Count: 28976 Status: Offline Project Badges: 
|
The task still has to be programmed to run in the particular gpu. It's a lot easier to program the tasks for a cpu core then a gpu core. Didn't you read the OP? |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Yes. I was explaining to a prior poster as to why the tasks aren't made first to run on a gpu instead of a cpu or for that matter on both. It's a whole different matter of programming techniques since gpus and cpus process tasks differently.
|
||
|
|
sk..
Master Cruncher http://s17.rimg.info/ccb5d62bd3e856cc0d1df9b0ee2f7f6a.gif Joined: Mar 22, 2007 Post Count: 2324 Status: Offline Project Badges:
|
There is almost no information about any future plans. Would disagree with that. There has been past reference to this pending project, and others. Even in this thread you can see clear intent develop a HCC2 project, at least from the scientists.... I don't think we need to know what new projects are coming and when... Knowing it was coming months before wouldn't have made any difference... Perhaps for you, but some people will see this as an opportunity to plan to replace existing hardware in a way that can contribute to this project. I think the announcement has been well received. |
||
|
|
Somervillejudson@netscape.net
Veteran Cruncher USA Joined: May 16, 2008 Post Count: 1065 Status: Offline Project Badges:
|
Lots of high powered GPU's sitting around idle so great news and as building new machine also was going to put minimal GPU but I'd now able to produce points may go wild!
|
||
|
|
cubes
World Community Grid Tech, Mapping Cancer Markers and Help Conquer Scientist Canada Joined: Mar 3, 2007 Post Count: 58 Status: Offline Project Badges: 
|
rilian, HCC on GPU will run as an additional compute platform for the main HCC project, drawing from the same pool of work units.
The CPU and GPU results are compatible. Our chief GPU wizard, Yulia, worked hard to achieve that result. We tested the numeric accuracy of our OpenCL (GPU) implementation against the CPU version on an extensive dataset. The results show a few small-magnitude differences between CPU and GPU, but no real-world effects on crystal detection. Christian |
||
|
|
rilian
Veteran Cruncher Ukraine - we rule! Joined: Jun 17, 2007 Post Count: 1454 Status: Offline Project Badges:
|
Thanks, that's really cool they are compatible!
----------------------------------------Could you also shed a light on the total number of work units on this phase ? |
||
|
|
ness1602
Cruncher Serbia Joined: Apr 30, 2010 Post Count: 10 Status: Offline Project Badges: 
|
So, GPU computing will be similar to CPU , performance wise?
----------------------------------------I have read that 5770 ATI is same as Core I7 single core, so i guess GPU will be one more core to crunch.
Crunch for BENCHforum team 26290 :)
|
||
|
|
krakatuk
Advanced Cruncher Germany Joined: Oct 3, 2008 Post Count: 141 Status: Offline Project Badges: 
|
ness1602, it's mentioned in the paper that it's 6 to 60 times faster in comparison to a cpu core. And they had some older GPUs there for the tests.
----------------------------------------So I would expect it to be even more faster with some modern high-end GPUs.  |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
In the paper there's concern over precision in GPU computing, will this mean the implementation will need a double-precision capable GPU?
I recently upgraded my 4850 to a 6870, and found that it's not supported by Milkyway@Home because it lacks double precision. However it is supported by Folding@Home, though I'd like to bring the computing power back into BOINC, while contributing to a net-good application, not wasting coal computing primes or solving chess. |
||
|
|
|