... flattening group > weights down to task-level so that the scheduler ... task > weights. That'd make group fairness as good as task ... the weight for a task group-weight - the weight for a ... iterates over the tree levels (not the cpus). h_load represents the ...

linux-kernel - Peter Zijlstra - May 22 2008 - 17:13

... I thought you were flattening group weights down to task-level ... @sgi.com >Subject: Re: fair group scheduler not so fair? > >On ... t really dug into the scheduler yet (although that's next ... the load balancer is >> properly group-aware. > >I'm trying, but ...

linux-kernel - Li, Tong N - May 22 2008 - 16:18

... started playing with the fair > group scheduler. (This is on a ... with two hogs in one group and three hogs in the > other group. Unfortunately, the cpu shares were not equally distributed ... What you're testing is SMP fairness of group scheduling and ...

linux-kernel - Peter Zijlstra - May 22 2008 - 02:56

... head and started playing with the fair > group scheduler. (This is on a dual cpu ... would attain perfect balance. This does not happen because: load_balance_fair() iterates thr' the ... in the order they were created. So it first examines what tasks it ...

linux-kernel - Srivatsa Vaddagiri - May 27 2008 - 13:15

... test env as below: Done. Overall the group scheduler results look better, but I'm ... allow for a tradeoff between performance and fairness? I have people that are looking for ... hogs all in the default "sys" group. Over multiple retries using 10-sec ...

linux-kernel - Chris Friesen - May 29 2008 - 17:30

... -0600, Chris Friesen wrote: > Overall the group scheduler results look better, but I'm ... allow for a tradeoff between > performance and fairness? The tuneables I can think of are: ... 5%): PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ #C ...

linux-kernel - Srivatsa Vaddagiri - May 30 2008 - 07:36

... patches, but I thought you were flattening group > weights down to task-level so that the scheduler only looks at ... upon every fork/exit? > That'd make group fairness as good as task fairness gets. Is > this still the case? -- Regards, ...

linux-kernel - Srivatsa Vaddagiri - May 23 2008 - 05:42

... much better, but still some fairness issues with more complex > setups. > ... 19.9% > 2490 19.9% > > a re-test gave errors of up ... , moved_load; + unsigned long busiest_cfs_rq_load; /* * empty group @@ -1452,25 +1454,31 @@ load_balance_fair( ...

linux-kernel - Srivatsa Vaddagiri - May 29 2008 - 12:46

... that would allow for a tradeoff between > > performance and fairness? I have people that are looking for within 4 ... . Anyway I dont think !group case is better at achieving fairness over shorter intervals. -- Regards, vatsa -- To unsubscribe ...

linux-kernel - Srivatsa Vaddagiri - May 30 2008 - 06:21

... /sched-smp-group-fixes/ > > make any difference for you? Not much ... case pid 2438 is in group "a" and pids 2439/2440 ... t really dug into the scheduler yet (although that's next ... the load balancer is properly group-aware. Chris -- To unsubscribe from ...

linux-kernel - Chris Friesen - May 22 2008 - 16:02

... your patches, but I thought you were flattening group > > > weights down to task-level so that the scheduler only looks at ... will change each task's weight lazily? If so how? -- Regards, vatsa -- To unsubscribe from this list: ...

linux-kernel - Srivatsa Vaddagiri - May 23 2008 - 06:19

... at your patches, but I thought you were flattening group > > > > weights down to task-level so that the scheduler only looks at per- ... . > Basically here's what I had in mind: > > Group A shares = 1024 > # of tasks in group A = ...

linux-kernel - Peter Zijlstra - May 23 2008 - 06:16

... in the > other group. Unfortunately, the cpu shares were not equally distributed > ... :03.62 0 hogb Note that fairness (using load balance approach we have ... longer it takes to converge to fairness. -- Regards, vatsa -- To unsubscribe from this ...

linux-kernel - Srivatsa Vaddagiri - May 27 2008 - 13:28

... -ass.net/kernel-patches/sched-smp-group-fixes/) >> below. Pls let me know ... much better, but still some fairness issues with more complex > setups. > ... of 6%: So it is converging to a fair state. How ... .9% > 2490 19.9% > > a re-test gave errors of up ...

linux-kernel - Dhaval Giani - May 28 2008 - 14:47

... results are good, the shorter-term fairness > isn't great. Is there a ... for a tradeoff between > performance and fairness? I have people that are looking ... fair does smp fairness look for a !group scenario? I don't expect group schould be able to do much ...

linux-kernel - Dhaval Giani - May 30 2008 - 02:43

... Skip over entities that are not tasks */ do { se = list_entry(next ... case, the lone task in Group a/b is always skipped ... FAIR_GROUP_SCHED || RT_GROUP_SCHED) prompt "Basis for grouping tasks" default USER_SCHED Index: current ...

linux-kernel - Srivatsa Vaddagiri - May 28 2008 - 12:33

... ass.net/kernel-patches/sched-smp-group-fixes/) > below. Pls let me know ... Looking much better, but still some fairness issues with more complex setups. ... 49.9% 2478 49.9% So far so good. I then created ... .9% 2490 19.9% a re-test gave errors of up ...

linux-kernel - Chris Friesen - May 28 2008 - 14:35

... case, the lone task in Group a/b is always skipped ... .net/kernel-patches/sched-smp-group-fixes/) >> below. Pls let me ... much better, but still some fairness issues with more complex > setups ... that the delta from ideal fairness is much more than desirable ...

linux-kernel - Srivatsa Vaddagiri - May 28 2008 - 22:50

... patches, but I thought you were flattening group > > weights down to task-level so that the scheduler only looks at ... it every time I need the weight. So every time I use a weight, I ... \Prod_{l=1} w_l/rw_{l-1} Not doing that will get you O(n) ...

linux-kernel - Peter Zijlstra - May 23 2008 - 05:39

... wrote: > But first, note that Groups "a" and "b" share bandwidth ... I've switched over to your group setup for testing. > The following ... .net/kernel-patches/sched-smp-group-fixes/) seems > to fix the ... . I put pid 2498 in group A, and pids 2499 and ...

linux-kernel - Chris Friesen - May 27 2008 - 14:13