| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
| World Community Grid Forums
|
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 17
|
|
| Author |
|
|
dorik
Cruncher Joined: Jul 28, 2008 Post Count: 3 Status: Offline |
Hello !
I want to update my ATI X1950 graphic card for a new and most powerfull card (ATI HD4850 ? ) . I have read that graphic cards can help grid computing so wich manufacturer (and or model) do you advise (i plan a midrange card ...) . (Does the BOINC application use the graphic card currently ? ) Regards, |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
WCG currently does not have any projects that use graphics cards for crunching. I am sure they will eventually but we just don't know when. Please keep in mind that you will need to use a newer version of BOINC than the one recommended by WCG which makes it difficult to get support through the WCG forums if you do have any problems.
----------------------------------------There are other projects that do use graphics cards and Nvidia is by far the most popular because their programming api (CUDA) has been out a bit longer that ATI's. Crunching with your graphics card has less options that with your CPU and also tends to run hot which is usually OK but you do not have an option to slow the processing down. I am currently crunching both WCG (with cpu) and GPUGrid (with video card)using BOINC 6.6.20 without any problems returning valid WUs. If you are really interested in GPU crunching I would suggest that you check out GPUGrid, but certainly choose any project that appeals to you. They all have charts that explain which video cards they support which may help you in making your decision about what you want to buy, likely they will all handle your everyday tasks just fine. {edited to fix typos} [Edit 2 times, last edit by Former Member at Apr 19, 2009 10:35:13 AM] |
||
|
|
mikaok
Senior Cruncher Finland Joined: Aug 8, 2006 Post Count: 489 Status: Offline Project Badges:
|
There are other projects that do use graphics cards and Nvidia is by far the most popular because their programming api (CUDA) has been out a bit longer that ATI's. Crunching with your graphics card has less options that with your CPU and also tends to run hot which is usually OK but you do not have an option to slow the processing down. Let me add one thing. Because of heavy screen slowdown, it is usually needed to suspend GPU computation when computer is used to other activities. Some high end card with a good third party cooler would be best way to start I think... What I'm trying to say is, that choose a card other than my 8800gts 512MB with reference cooling.
to infinity and beyond
![]() |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello, I just wanted to point you to where you might find additional information on the stream technologies and the current direction it is moving in...
The industries are pushing for a more unified open platform called OpenCL... this evolution will eventually target both multi-core cpu's and the latest GPU's and other accelerators. Hopefully this will also enable projects like these to easily adopt and move to a code base that supports cross-vendor technologies. Here are a couple of links for you to check out : http://ati.amd.com/technology/streamcomputing/opencl.html http://ati.amd.com/technology/streamcomputing/ |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Just wanted to put in my two cents worth and say hurry up with the gpu numbers crunching client. I have a gaming rig using 2x ati 4870 x2's. That is 4 gpu cores i could be crunching with for this project. From what i have read it looks like ATI cards have an advantage over Nvidia and cuda when it comes to this stuff. Other distributed computer projects are reporting speed increases in the 20x-30x range over the standard cpu clients.
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
If speed was the only thing the scientists cared about, they would have simplified things, lowered the resolution, lowered the precision - and they would have finished already.
But that wouldn't be very useful. |
||
|
|
depriens
Senior Cruncher The Netherlands Joined: Jul 29, 2005 Post Count: 350 Status: Offline Project Badges:
|
If speed was the only thing the scientists cared about, they would have simplified things, lowered the resolution, lowered the precision - and they would have finished already. But that wouldn't be very useful. Hi Didactylos, I thought I've seen something here on the forums that a CUDA application is currently being developed for use on WCG. Is that true? If so, can you give any indication of when we can expect such application at earliest? Thanks and cheers! ![]() ![]() |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
No date yet.
But once things get moving, a new science application build will be needed, plus a new recommended BOINC version - and currently BOINC 6.6 has major problems, so I don't see that being recommended (certainly not in its present form). |
||
|
|
depriens
Senior Cruncher The Netherlands Joined: Jul 29, 2005 Post Count: 350 Status: Offline Project Badges:
|
Thanks!
---------------------------------------- ![]() |
||
|
|
Eric-Montreal
Cruncher Canada Joined: Nov 16, 2004 Post Count: 34 Status: Offline Project Badges:
|
If speed was the only thing the scientists cared about, they would have simplified things, lowered the resolution, lowered the precision - and they would have finished already. But that wouldn't be very useful. All new mid & high end video cards from Nvidia and ATI are capable of the very same IEEE754 double precision arithmetic calculations a normal CPU can perform. They have been shipping for a year now and have reached an affordable price point. List of double precision capable GPUs : Nvidia : GTX260 GTX275 GTX280 GTX285 GTX295 ATI/AMD : HD4850 GD4870 HD4890 These boards now start around 100US$ (HD4850) and 180$ (GTX260) http://www.newegg.com/Product/Product.aspx?Item=N82E16814161259 When buying a new video card, unless the budget is very tight, these cards are worth considering, even if WCG does not use them yet. Other projects that can run along with WCG do use them with little overhead from the main CPU. |
||
|
|
|