Message boards :
Number crunching :
computing on CUDA with Nvidia
Message board moderation
| Author | Message |
|---|---|
|
Send message Joined: 8 Jul 06 Posts: 3 Credit: 288,476 RAC: 0  
|
Hello, LHCatHome is capable to work with CUDA on GPU Nvidia ? Or the GPU are not complex ? (like many other projects) LHC is it a project paralelisable ? ps: i'm french |
|
Send message Joined: 8 Jul 06 Posts: 3 Credit: 288,476 RAC: 0
|
up nobody know if LHCathome is compatible with CUDA ? :-( |
 Send message Joined: 18 Sep 04 Posts: 163 Credit: 1,682,370 RAC: 0 |
Hi, AFAIK sixtrack is FORTRAN-code and the project-team was lucky to find a compiler, which produced identical results on windows && linux. So we have no PPC application and I doubt we will see a CUDA (or Cell) application (for sixtrack). But that is just the information I gathered from reading the message board. Michael Team Linux Users Everywhere 
|
|
Send message Joined: 31 Dec 05 Posts: 68 Credit: 8,691 RAC: 0 |
Hi, Also, LHC@Home has far more computer power attached than it needs, so the need for CUDA is less than at, say, Rosetta. |
|
Send message Joined: 14 Sep 05 Posts: 1 Credit: 3,928 RAC: 0 |
Hi, My first post on the boards... I belie its possible to rewrite the diffeential algebra functions or/and every/some math related functions of the SixTrack application , to use the Fortran to Cuda kernel calls and benefit from the gpu FLOP computing power. My knowledge of the boinc/sixtrack api and architecture is quite limited, but if anyone is kind enough to point me at the right direction, I am more than willing to look into the issue... I understand that the SixTrack source code (the one with Boinc bindigs etc) is NOT available to the public. Why is that? Is there some kind of license that forbids it? Why such a policy on the project? |
|
Send message Joined: 29 Dec 06 Posts: 100 Credit: 184,937 RAC: 0 |
Also, LHC@Home has far more computer power attached than it needs, so the need for CUDA is less than at, say, Rosetta. This is a good point. However, if we ever find that we have more work than our LHC@Home users can handle, then at that time, the CUDA question should definitely be revisited. This is a likely scenario if the GARFIELD application matures and is utilized... more likely if other apps are introduced to LHC@Home and still more likely if we can get our hands on some of the ATLAS data. JMHMO (Just My Higher-Mortal Opinion) ;} _____________________ _______ "Three quarks for Muster Mark!" . . . . . . . - James Joyce, Finnegans Wake . . . .  
|
©2024 CERN