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: 11
|
![]() |
Author |
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Finally I build a new shiny quad machine.
Is it possible for example to dedicate one core to ‘seti’ and the others to ‘wcg’? I am plying with ‘resource share” property but not to my satisfaction… Advice pls? Thanks, DB ![]() |
||
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
No is the short answer. Set the weight (resource share) to 300 for WCG and leave Seti at default 100 and BOINC takes care of allocating the resources over time. You could trick the client into running one very long project like CPDN continuously and WCG using the other cores, but not with the Seti/WCG combination.
----------------------------------------
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
retsof
Former Community Advisor USA Joined: Jul 31, 2005 Post Count: 6824 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
There are 4 profiles available for BOINC, default, home, school, work.
----------------------------------------Each core (device) could point to a different profile. It would be possible to separate out the WCG internal projects...one core running African Climate Change, another with FAAH, another with DDD-T and another with Cancer Curing. The resource share is still the mechanism for percentages of mix with non-WCG projects.
SUPPORT ADVISOR
----------------------------------------Work+GPU i7 8700 12threads School i7 4770 8threads Default+GPU Ryzen 7 3700X 16threads Ryzen 7 3800X 16 threads Ryzen 9 3900X 24threads Home i7 3540M 4threads50% [Edit 2 times, last edit by retsof at Dec 5, 2007 7:43:31 PM] |
||
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Sometimes have my creative moments. If you set any profile to 1 core / project, BOINC will just run 1 core overall. Setting 1 core on Seti and 3 cores on WCG, will worst case lead to conflicts and at best using 3 cores.
----------------------------------------Anyway, plug it into Didactylos' thread on the future of BOINC. It's been proposed multiple times on the BOINC dev forum, which the BOINC dev's dont read at all.... create a Trac Ticket there with motivation as it seems to me leading to absolute minute efficiency improvement and complicate the software controls and logic. Added: Hey restsof, changing your post while a reply is already underway does not help conversation. The above is in response to your unedited post. And No, all 4 cores are 1 device, attached to 1 profile, be it one of the 4 available. Think a device will only consider one 'general' crunching profile of the 4 available and always the last one revised.
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 2 times, last edit by Sekerob at Dec 5, 2007 7:52:44 PM] |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Sometimes have my creative moments. If you set any profile to 1 core / project, BOINC will just run 1 core overall. Setting 1 core on Seti and 3 cores on WCG, will worst case lead to conflicts and at best using 3 cores. Creative it is. Any change to a boinc profile will take precedence no matter on which project you are. (seti / wcg/ cpdn etc) I’ve tried this and it does not work. It seems to me that this should be an enhancement request to the boinc client it self (on the ‘preferences’ tab) Anyway I will play with the project weight thing until I get some ‘satisfaction’… ![]() Thanks! DB |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I believe the last time the issue of CPU affinity (the fancy name for tying a process to a core) was raised, Dr. Anderson decided it wasn't part of the roadmap.
My memory may be faulty on this. |
||
|
retsof
Former Community Advisor USA Joined: Jul 31, 2005 Post Count: 6824 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Added: Hey restsof, changing your post while a reply is already underway does not help conversation. I had something in there that was wrong after checking, and took it out of there. Checking takes longer on dial-up, but that may change soon. I also thought that the CPDN Seasonal Attribution Project was winding down. I found that they added a few workunits, but I don't know how long it will last. Restsof must be somebody else.
SUPPORT ADVISOR
----------------------------------------Work+GPU i7 8700 12threads School i7 4770 8threads Default+GPU Ryzen 7 3700X 16threads Ryzen 7 3800X 16 threads Ryzen 9 3900X 24threads Home i7 3540M 4threads50% [Edit 2 times, last edit by retsof at Dec 5, 2007 9:53:26 PM] |
||
|
tekennelly
Cruncher Joined: Oct 10, 2005 Post Count: 45 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
I believe the last time the issue of CPU affinity (the fancy name for tying a process to a core) was raised, Dr. Anderson decided it wasn't part of the roadmap. My memory may be faulty on this. I believe that allowing the work units to float among available processors causes cache disruption thereby increasing the processing time. A simple test would be to take a dual core and benchmark two work units on the dual core and see how long it takes. Next, start the two processes on the same machine and use the windows set affinity capability to set each work unit to a different core and see if there is a decrease in run time. Now, can we get someone to run this test and provide the results. The test should have all the characteristics of a proper benchmark. The bottom line is that if it is found that processing times decrease with cpu affinity then it could be decided whether the coding effort to set affinity dynamically would be worth it. |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Oddly, that's not true (or so I understood from the discussion).
It is forcing processes to specific cores that causes cache problems. Normally, the OS is very good at managing processes, and normally core switching will be minimal or non-existent anyway. This makes the phenomenal amount of work needed to implement CPU affinity on the various BOINC platforms (it is distinctly non-trivial) completely moot. It would be good to have benchmark data to back this. |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Restsof must be somebody else. ![]() |
||
|
|
![]() |