| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Help Conquer Cancer
Thread: HCC with GPU |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 486
|
|
| Author |
|
|
cristipurdel
Senior Cruncher Joined: Dec 13, 2008 Post Count: 158 Status: Offline Project Badges: 
|
Still waiting to see who is going to be first to release a MD project running on opencl: hcc, poem++, gpugrid or drugsdiscovery :)
---------------------------------------- |
||
|
|
Dataman
Ace Cruncher Joined: Nov 16, 2004 Post Count: 4865 Status: Offline Project Badges: 
|
Still waiting to see who is going to be first to release a MD project running on opencl: hcc, poem++, gpugrid or drugsdiscovery :) Yes, will be interesting who's next. Doubt if it will be HCC. POEM says they are close to release but have been saying that for a year. Jack seems to be chasing his tail running both DrugDiscovery and Hydrogen.  |
||
|
|
TimAndHedy
Senior Cruncher Joined: Jan 27, 2009 Post Count: 267 Status: Offline Project Badges: 
|
Heard it through the grapevine (CCR)... seemingly the code is progressing well... not CUDA, but OpenCL. Now we need a set and forget client and who knows is 6.12 going to deliver this half way ;>) Good news. Any rumors on the performance increase? |
||
|
|
rilian
Veteran Cruncher Ukraine - we rule! Joined: Jun 17, 2007 Post Count: 1452 Status: Offline Project Badges:
|
Still waiting to see who is going to be first to release a MD project running on opencl: hcc, poem++, gpugrid or drugsdiscovery :) HCC has no relations to MD (molecular dynamic) ... It is image analysis |
||
|
|
BladeD
Ace Cruncher USA Joined: Nov 17, 2004 Post Count: 28976 Status: Offline Project Badges:
|
Heard it through the grapevine (CCR)... seemingly the code is progressing well... not CUDA, but OpenCL. Now we need a set and forget client and who knows is 6.12 going to deliver this half way ;>) What is the grapevine saying now? |
||
|
|
fablefox
Senior Cruncher Joined: May 31, 2010 Post Count: 161 Status: Offline Project Badges: 
|
There is an article, but I forgot the link. There is another post by me in another sub-forum regarding that. But it's related to the Rice project. In the article it already test it on local GPU based grid, and now working on GPU client. It seems that (my opinion) they test it on older GPU card to make sure as many people can join. Since it was single presicion, I guess they can target a lot of older GPU card.
---------------------------------------- |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Heard it through the grapevine (CCR)... seemingly the code is progressing well... not CUDA, but OpenCL. Now we need a set and forget client and who knows is 6.12 going to deliver this half way ;>) Good news. Any rumors on the performance increase? Let me try this FTM, HITTG: The multiplication factor could be having a theoretical limit of the number of cores on a card [and the lowest common denominator in a target population]. Sounds demographic-ie, does it not ;>) Heard nothing on a 1:1 comparison... reading posts here [on the public side] that an OpenCL performance is barely matching a CPU, but then, it's still an extra "core" so to speak and WCG has tackled the NVIDIA/ATI differences without a need of a 6.12 client, smelling like platform optimization. On wattage, not at all informed extensively what a GPU versus an efficient power-managed CPU draws. Sandy Bridges and coming Bulldozers and their successors seem to be gaining fast. Should give a little to fill a few more pages of discussion, until go-live which is as per the official news, when it comes. ;D --//-- |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
so are you saying the gpu code is progressing but it might not be much faster than the cpu code?
|
||
|
|
BladeD
Ace Cruncher USA Joined: Nov 17, 2004 Post Count: 28976 Status: Offline Project Badges:
|
I think he was talking about OpenCL performance not a specific project.
---------------------------------------- |
||
|
|
sk..
Master Cruncher http://s17.rimg.info/ccb5d62bd3e856cc0d1df9b0ee2f7f6a.gif Joined: Mar 22, 2007 Post Count: 2324 Status: Offline Project Badges:
|
Folding have similar joke threads - a CPU can do more but at least your GPU is doing something! Wha, wha wha...
A mid range 3 generation old GPU can do more than a SB, if on the correct project. To think any CPU could do more work than a GTX590 or an HD 6970 is bonkers. On a suitable project a $700 GTX590 or HD 6970 would easily outperform an 80thread quad skt E7 server series system costing over $20,000. How many Boinc Credits does a SB get at WCG per day? How many Boinc Credits does a SB get at MW per day? How many Boinc Credits does an HD 6970 get at MW per day? This does not mean SB is not an excellent CPU, or that OpenCL is a viable cross-platform GPU solution (it's not; CUDA is faster and OpenCL on AMD/ATI is buggy). |
||
|
|
|