Message boards :
Graphics cards (GPUs) :
BOINC 6.3.18 for LINUX ist out
Message board moderation
| Author | Message |
|---|---|
|
Send message Joined: 24 Aug 08 Posts: 45 Credit: 3,431,862 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]()
|
6.3.18 for LINUX is available. http://boinc.berkeley.edu/dl/?C=M;O=D |
Bender10Send message Joined: 3 Dec 07 Posts: 167 Credit: 8,368,897 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]()
|
|
koschiSend message Joined: 14 Aug 08 Posts: 127 Credit: 913,858,161 RAC: 18 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
You are right, its the i686 build, but it works with PS3GRID :-D The scheduling problem isn't fixed yet. When I start BOINC on the Dualcore, it runs 2CPU and one GPU task, so far so good, but after short time it stops one of the CPU tasks... With 6.3.14 it was the GPU task that was stopped. This way its easier to handle, ncpus to 3 and it starts with 3 CPU tasks but stops one, so I'm left with 2+1 tasks which is ok for me :) |
koschiSend message Joined: 14 Aug 08 Posts: 127 Credit: 913,858,161 RAC: 18 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
Hmm, can't edit, so I'll make a new post... I said it runs, and half way thats right, it runs even 64bit tasks (PS3GRID and SETI v8), but doesn't fetch new work for them, so this morning the GPU was idle... No problems for 32bit Projekts like Spinhenge or QMC, but thats kind of logical, isn't it? So to get more work, I switched back to 64bit 6.3.14, having again the same problems, that it puts my GPU task to "waiting to run" after some short time, leaving me with 2 or 3 CPU tasks. So I'm back to crunching projects with big work units, suspending all and then resuming two at a time to run. That way I can keep 2 CPUs and the GPU busy... The strange thing is that on my Quadcore everything works fine with 6.3.14, its always running 4+1 tasks without any scheduler hickups :-/ boincmgr GUI 6.3.18 was using 100% of one CPU while runnning, I replaced it by 6.3.14 and the problem diappeared... I wonder whats holding Berkeley back to compile 6.3.18 as 64bit? A 32bit client is useless in this projekt, and other projects don't have any advantage from these dev. builds either... |
©2025 Universitat Pompeu Fabra