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 |
|
fablefox
Senior Cruncher Joined: May 31, 2010 Post Count: 161 Status: Offline Project Badges: |
Please understand that they have a lot of thing on their hand, current research and implementing future research that currently being applied.
----------------------------------------I'm sure WCG has think through and through. Fermi does support double precision point, I just read the whitepaper. Ingleside, are you saying that of all project that __uses the GPU__, milky way is the most popular? Still, I admit that despite MW only support DP GPU, and has much less user compared to WCG, they generate much more point (GPU obviously help, no question about that). I don't know how MW works, but at CG, we have different sciences all the time with all kind of different process and calculation. It's not like we only have one app to convert into GPU-enabled and let it run for years. So we have to look into this aspect too. Anyway, the time will come for GPU I'm sure, and I rather crunch to save life. Once WCG moves to GPU, I'm going to get myself one of these: http://www.workstationspecialist.com/hpc/personal_super_computer/ So I can start asking, can you support 8 GPU, NOW!!?? :-) |
||
|
Falconet
Master Cruncher Portugal Joined: Mar 9, 2009 Post Count: 3295 Status: Offline Project Badges: |
Fablefox,
----------------------------------------you can use your GPU to crunch on GPUgrid. It's also medical related. AMD Ryzen 5 1600AF 6C/12T 3.2 GHz - 85W AMD Ryzen 5 2500U 4C/8T 2.0 GHz - 28W AMD Ryzen 7 7730U 8C/16T 3.0 GHz |
||
|
petnek
Advanced Cruncher Czech Republic Joined: Mar 17, 2008 Post Count: 89 Status: Offline Project Badges: |
Projects with GPUs app:
----------------------------------------Aqua: nVidia (Win, Linux) Collatz nVidia (Win, Linux), ATI (with/without double-precision) DistrRTgen: nVidia (Win) DNETC: nVidia (Win, Linux), ATI (with/without DP) DrugDiscovery: nVidia (Win, Linux), ATI (with/without DP) - without work GPUGRID: nVidia (Win, Linux) Hydrogen: nVidia (Win) - without work Lattice: nVidia (Win) - without work MilkyWay: nVidia (Win, Linux), ATI (only GPUs with DP) PrimeGrid (Prooth Prime Search - Sieve): nVidia (Win, Linux) Seti: nVidia (Win) - app is not good POEM - in development Everything is here (Czech) - If you preferred one of app, you can select by clicking on the image at the beginning of the table. |
||
|
Ingleside
Veteran Cruncher Norway Joined: Nov 19, 2005 Post Count: 974 Status: Offline Project Badges: |
Ingleside, are you saying that of all project that __uses the GPU__, milky way is the most popular? Yes, a quick look on BoincStats shows that Milkyway has 34211 active hosts, and while a couple of these is cpu-only, since the GPU's is so much faster than a cpu would guess atleast 90% of these is using their GPU. Also, some of the computers has multiple GPU's (either multiple cards or dual-GPU-cards), so would guess 30k - 40k active GPU's runs Milkyway. The 2nd. most popular GPU-project is Folding@Home, with roughly 21k active GPU's (this is GPU's, not computers with GPU's like is counted by BOINC-projects) Collatz has 12k, DNET 4541, and GPUGRID has 3425. The difficult part is where projects like SETI@home is in all of this, since while SETI@home does have GPU-application for CUDA that gives a big boost, it's still very many that runs SETI@home on cpu-only. Projects like Lattice and Docking@home has little or no GPU-work (or work at all), so I haven't counted these projects at all... and in any case they're below 10k active computers. Most of the GPU-projects has one or more disadvantages, some of the most important are: As mentioned, Milkyway is double-precision only. GPUGRID, SETI@home and Einstein@home is Nvidia-only. Docking@home and Lattice have little or no GPU-work, and is mainly on testing-state. Einstein@home has "hybrid"-application, GPU only runs part of the time, and uses 1 cpu like without using GPU. Also, the speedup from GPU is small. Folding@Home's Ati-client has mediocre speed compared to Nvidia, and due to their boinus-points-system running SMP + Ati often gives little or no difference in points/day compared to running SMP-only, in some cases you'll actually get less points/day if you also crunches on the ATi... "I make so many mistakes. But then just think of all the mistakes I don't make, although I might." |
||
|
BladeD
Ace Cruncher USA Joined: Nov 17, 2004 Post Count: 28976 Status: Offline Project Badges: |
Yep, that's about what I saw when I was looking for something for my GPUs to do.
---------------------------------------- |
||
|
BladeD
Ace Cruncher USA Joined: Nov 17, 2004 Post Count: 28976 Status: Offline Project Badges: |
Yeah, only Tesla capable of doing double precision point at the moment (AFAIK). FYI, Milkyway uses double-precision, and even some GPU-cards therefore can't run Milkyway, they're still the #1 GPU-project (by #active GPU's). Maybe because they have a very good ATI app. |
||
|
rilian
Veteran Cruncher Ukraine - we rule! Joined: Jun 17, 2007 Post Count: 1452 Status: Offline Project Badges: |
Please stay on topic, which is HCC via GPU. There are really tons of threads to discuss which GPU project is best.
----------------------------------------Thanks |
||
|
cristipurdel
Senior Cruncher Joined: Dec 13, 2008 Post Count: 158 Status: Offline Project Badges: |
AMD SDK 2.3 is out, is this useful?
----------------------------------------Any e.t.a. on a beta testing? |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
If HCC start using GPU, I will support it surely, if it's possible. ;)
|
||
|
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 ;>)
|
||
|
|