Skip to contentSkip to conversation

PlugSched-5.2.1 for 2.6.11 and 2.6.12

RFC Peter Williams 2 participants 7 messages Jun 20, 2005 – Jul 5, 2005

lore
Share & tools
lore
#0 lore raw

PlugSched-5.2.1 is available for 2.6.11 and 2.6.12 kernels. This version applies Con Kolivas's latest modifications to his "nice" aware SMP load balancing patches.

A patch to bring PlugSched-5.2 for 2.6.11 to PlugSched-5.2.1 is available at:

<http://prdownloads.sourceforge.net/cpuse/plugsched-5.2-to-5.2.1-for-2.6.11.patch?download>

A patch for 2.6.12 is available at:

<http://prdownloads.sourceforge.net/cpuse/plugsched-5.2.1-for-2.6.12.patch?download>

Very Brief Documentation:

You can select a default scheduler at kernel build time. If you wish to boot with a scheduler other than the default it can be selected at boot time by adding:

cpusched=<scheduler>

to the boot command line where <scheduler> is one of: ingosched, nicksched, staircase, spa_no_frills or zaphod. If you don't change the default when you build the kernel the default scheduler will be ingosched (which is the normal scheduler).

The scheduler in force on a running system can be determined by the contents of:

/proc/scheduler

Control parameters for the scheduler can be read/set via files in:

/sys/cpusched/<scheduler>/

Peter

Signature · 4 lines
Peter Williams                                   [email protected]

"Learning, n. The kind of ignorance distinguishing the studious."
 -- Ambrose Bierce
#1 lore raw
Quoted 3 lines · Peter Williams

PlugSched-5.2.1 is available for 2.6.11 and 2.6.12 kernels. This version applies Con Kolivas's latest modifications to his "nice" aware SMP load balancing patches.

Thanks Peter.

Any word from your own testing/testers on how well smp nice balancing is working for them now?

Cheers,
Con

#2 lore raw
Quoted 10 lines · Con Kolivas
Quoted 3 lines · Peter Williams

PlugSched-5.2.1 is available for 2.6.11 and 2.6.12 kernels. This version applies Con Kolivas's latest modifications to his "nice" aware SMP load balancing patches.

Thanks Peter.

Any word from your own testing/testers on how well smp nice balancing is working for them now?

No, they got side tracked onto something else but should start working on it again soon. I'll give them a prod :-)

Peter

Signature · 4 lines
Peter Williams                                   [email protected]

"Learning, n. The kind of ignorance distinguishing the studious."
 -- Ambrose Bierce
#3 lore raw
Quoted 10 lines · Peter Williams
Quoted 8 lines · Con Kolivas
Quoted 3 lines · Peter Williams

PlugSched-5.2.1 is available for 2.6.11 and 2.6.12 kernels. This version applies Con Kolivas's latest modifications to his "nice" aware SMP load balancing patches.

Quoted 5 lines
Quoted 2 lines

Thanks Peter.
Any word from your own testing/testers on how well smp nice balancing is working for them now?

No, they got side tracked onto something else but should start working on it again soon. I'll give them a prod :-)

Con,
We've done some more testing with this with results that are still
disappointing. We think that one reason for this is that move_tasks() doesn't take "nice" into account. I'm going to look at modifying it so that it moves a certain amount of niceness rather than a specified number of tasks.
A second (more difficult to solve) issue is that we think it would work
better if both queue lengths and "nice" loads were kept approximately equal between the queues. It's hard to see how this can be managed without considerable overhead.

Peter

Signature · 4 lines
Peter Williams                                   [email protected]

"Learning, n. The kind of ignorance distinguishing the studious."
 -- Ambrose Bierce
#4 lore raw
Quoted 11 lines · Peter Williams
Quoted 10 lines · Peter Williams
Quoted 8 lines · Con Kolivas
Quoted 3 lines · Peter Williams

PlugSched-5.2.1 is available for 2.6.11 and 2.6.12 kernels. This version applies Con Kolivas's latest modifications to his "nice" aware SMP load balancing patches.

Thanks Peter.
Any word from your own testing/testers on how well smp nice balancing is working for them now?

Quoted 6 lines
Quoted 2 lines

No, they got side tracked onto something else but should start working on it again soon. I'll give them a prod :-)

Con,
We've done some more testing with this with results that are still
disappointing.

Is this with the migration thread accounted patch as well?

Con

#5 lore raw
Quoted 19 lines · Con Kolivas
Quoted 17 lines · Peter Williams
Quoted 14 lines · Peter Williams
Quoted 9 lines · Con Kolivas
Quoted 3 lines · Peter Williams

PlugSched-5.2.1 is available for 2.6.11 and 2.6.12 kernels. This version applies Con Kolivas's latest modifications to his "nice" aware SMP load balancing patches.

Thanks Peter.
Any word from your own testing/testers on how well smp nice balancing is working for them now?

No, they got side tracked onto something else but should start working on it again soon. I'll give them a prod :-)

Quoted 6 lines
Quoted 3 lines

Con,
We've done some more testing with this with results that are still
disappointing.

Is this with the migration thread accounted patch as well?

Yes.

Signature · 4 lines
Peter Williams                                   [email protected]

"Learning, n. The kind of ignorance distinguishing the studious."
 -- Ambrose Bierce
#6 lore raw
Quoted 13 lines · Peter Williams
Quoted 12 lines · Con Kolivas
Quoted 11 lines · Peter Williams
Quoted 10 lines · Peter Williams
Quoted 8 lines · Con Kolivas
Quoted 3 lines · Peter Williams

PlugSched-5.2.1 is available for 2.6.11 and 2.6.12 kernels. This version applies Con Kolivas's latest modifications to his "nice" aware SMP load balancing patches.

Thanks Peter.
Any word from your own testing/testers on how well smp nice balancing is working for them now?

Quoted 8 lines
Quoted 8 lines
Quoted 6 lines
Quoted 2 lines

No, they got side tracked onto something else but should start working on it again soon. I'll give them a prod :-)

Con,
We've done some more testing with this with results that are still
disappointing.

Is this with the migration thread accounted patch as well?

The results from smp nice I've received so far show that a 30% slowdown (instead of 5% on uniprocessor) occurs to high priority tasks in the presence of low priority tasks worst case scenario when the tasks sleep frequently. However excellent smp nice support with negligible slowdown occurs (ie nice is well respected) when the tasks are fully cpu bound. This suggests that what is happening on task wakeup is the limiting factor with this smp nice implementation. This makes sense given that most of the balancing occurs during busy rebalance when the queues are busy. I tried ignoring the length of queue (ie not having the single task running check for idle rebalance) and while this improved the nice effect substantially, it also cost us heavily in throughput making it unwarranted.

Cheers,
Con