Welcome to MilkyWay@home

CPU Scheduling question!!


Advanced search

Message boards : Number crunching : CPU Scheduling question!!
Message board moderation

To post messages, you must log in.

AuthorMessage
Cliff Harding
Avatar

Send message
Joined: 2 Jul 09
Posts: 26
Credit: 5,577,615
RAC: 26,263
5 million credit badge10 year member badge
Message 53137 - Posted: 13 Feb 2012, 13:52:51 UTC

Currently running both SETI (100%) and Milkyway (50%) on my i7/950 machine, running on 6 of 8 cores. I have noticed that when Milkyway 1.00 tasks are running they are running on all available cores, placing the SETI tasks in 'waiting to run' state. Is this normal? I was expecting to see a mix of both SETI & Milkyway tasks executing at the same time. Deadline for the Milkyway tasks is far enough away to see this happening.

It should be noted that the machine is running BOINC 7.0.15 and there might be a scheduler problem, but this situation has been in existance even in the 6.10 & 6.12 versions.


I don't buy computers, I build them!
ID: 53137 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Len LE/GE

Send message
Joined: 8 Feb 08
Posts: 261
Credit: 104,050,322
RAC: 0
100 million credit badge10 year member badge
Message 53138 - Posted: 13 Feb 2012, 17:28:54 UTC - in response to Message 53137.  

Are you talking about 1 mw over all cores (multithreaded app like mw nbody) or 1 mw per core (makes 6 * mw)?
ID: 53138 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Cliff Harding
Avatar

Send message
Joined: 2 Jul 09
Posts: 26
Credit: 5,577,615
RAC: 26,263
5 million credit badge10 year member badge
Message 53140 - Posted: 13 Feb 2012, 18:24:14 UTC - in response to Message 53138.  

Are you talking about 1 mw over all cores (multithreaded app like mw nbody) or 1 mw per core (makes 6 * mw)?


Sorry 'bout that. I'm talking 6 tasks, 1 per core. It seems that it only applies to MW 1.00. Nbody is not affected in this manner, which appears to have a normal mix of between cores; meaning Nbody will not overtake more than its resource share.


I don't buy computers, I build them!
ID: 53140 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Len LE/GE

Send message
Joined: 8 Feb 08
Posts: 261
Credit: 104,050,322
RAC: 0
100 million credit badge10 year member badge
Message 53152 - Posted: 14 Feb 2012, 0:38:54 UTC
Last modified: 14 Feb 2012, 0:40:23 UTC

My first guess was the scheduler of BM 7 but since you said it happened in 6.10 and 6.12 too, I have no real idea what could cause the same fault in all 3 versions.

I just saw you are running a mix of separation v1.02 nvidia, separation v1.00 cpu and nbody v0.84.
I haven't see anyone reporting that BM got confused by nbody to run all mw cpu apps as multithreaded (That would be a bad BM bug).

Shooting into the dark: Maybe something in the code that could make BM think (under some conditions) mw separation is a multithread app and it needs to free all cores for it? And than it fills all free cores with separation apps since they are singlethreaded?
You should send Matt Arsenault a PM describing your problem as detailed as possible.
ID: 53152 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profilemikey
Avatar

Send message
Joined: 8 May 09
Posts: 2299
Credit: 393,034,908
RAC: 35,269
300 million credit badge10 year member badgeextraordinary contributions badge
Message 53158 - Posted: 14 Feb 2012, 12:08:56 UTC - in response to Message 53140.  

Are you talking about 1 mw over all cores (multithreaded app like mw nbody) or 1 mw per core (makes 6 * mw)?


Sorry 'bout that. I'm talking 6 tasks, 1 per core. It seems that it only applies to MW 1.00. Nbody is not affected in this manner, which appears to have a normal mix of between cores; meaning Nbody will not overtake more than its resource share.


Boinc is VERY complicated and can take a LONG time to balance itself out to the resource share settings we users set. Sometimes you will see nothing but Seti units running, sometimes nothing but MW units running. Normally you should see a combination of the two. To get an idea of some of the things Boinc is trying to balance go to the Boinc Manager, down by the clock, click on the Projects tab and then click each project in turn and then on the left click Properties and scroll down near the bottom. Lots of numbers that Boinc is trying to juggle to reach your particular resource share.
ID: 53158 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote

Message boards : Number crunching : CPU Scheduling question!!

©2020 Astroinformatics Group