Re: [RFC][PATCH 2/2] cpufreq: schedutil: Force max frequency on busy CPUs

2017-03-27 Thread Vincent Guittot
On 25 March 2017 at 04:48, Joel Fernandes wrote: > Hi Vincent, > > On Thu, Mar 23, 2017 at 3:08 PM, Vincent Guittot > wrote: > [..] > So I'm not really aligned with the description of your problem: PELT > metric underestimates the load of the CPU. The PELT is just about > tracki

Re: [RFC][PATCH 2/2] cpufreq: schedutil: Force max frequency on busy CPUs

2017-03-24 Thread Joel Fernandes
Hi Vincent, On Thu, Mar 23, 2017 at 3:08 PM, Vincent Guittot wrote: [..] >>> So I'm not really aligned with the description of your problem: PELT metric underestimates the load of the CPU. The PELT is just about tracking CFS task utilization but not whole CPU utilization and

Re: [RFC][PATCH 2/2] cpufreq: schedutil: Force max frequency on busy CPUs

2017-03-23 Thread Vincent Guittot
On 23 March 2017 at 00:56, Joel Fernandes wrote: > On Mon, Mar 20, 2017 at 5:34 AM, Patrick Bellasi > wrote: >> On 20-Mar 09:26, Vincent Guittot wrote: >>> On 20 March 2017 at 04:57, Viresh Kumar wrote: >>> > On 19-03-17, 14:34, Rafael J. Wysocki wrote: >>> >> From: Rafael J. Wysocki >>> >> >>>

Re: [RFC][PATCH 2/2] cpufreq: schedutil: Force max frequency on busy CPUs

2017-03-22 Thread Joel Fernandes
On Mon, Mar 20, 2017 at 5:34 AM, Patrick Bellasi wrote: > On 20-Mar 09:26, Vincent Guittot wrote: >> On 20 March 2017 at 04:57, Viresh Kumar wrote: >> > On 19-03-17, 14:34, Rafael J. Wysocki wrote: >> >> From: Rafael J. Wysocki >> >> >> >> The PELT metric used by the schedutil governor underesti

Re: [RFC][PATCH 2/2] cpufreq: schedutil: Force max frequency on busy CPUs

2017-03-20 Thread Patrick Bellasi
On 20-Mar 14:05, Rafael J. Wysocki wrote: > On Monday, March 20, 2017 01:06:15 PM Patrick Bellasi wrote: > > On 20-Mar 13:50, Peter Zijlstra wrote: > > > On Mon, Mar 20, 2017 at 01:35:12PM +0100, Rafael J. Wysocki wrote: > > > > On Monday, March 20, 2017 11:36:45 AM Peter Zijlstra wrote: > > > > >

Re: [RFC][PATCH 2/2] cpufreq: schedutil: Force max frequency on busy CPUs

2017-03-20 Thread Rafael J. Wysocki
On Monday, March 20, 2017 01:06:15 PM Patrick Bellasi wrote: > On 20-Mar 13:50, Peter Zijlstra wrote: > > On Mon, Mar 20, 2017 at 01:35:12PM +0100, Rafael J. Wysocki wrote: > > > On Monday, March 20, 2017 11:36:45 AM Peter Zijlstra wrote: > > > > On Sun, Mar 19, 2017 at 02:34:32PM +0100, Rafael J.

Re: [RFC][PATCH 2/2] cpufreq: schedutil: Force max frequency on busy CPUs

2017-03-20 Thread Vincent Guittot
On 20 March 2017 at 13:59, Rafael J. Wysocki wrote: > On Monday, March 20, 2017 09:26:34 AM Vincent Guittot wrote: >> On 20 March 2017 at 04:57, Viresh Kumar wrote: >> > On 19-03-17, 14:34, Rafael J. Wysocki wrote: >> >> From: Rafael J. Wysocki >> >> >> >> The PELT metric used by the schedutil g

Re: [RFC][PATCH 2/2] cpufreq: schedutil: Force max frequency on busy CPUs

2017-03-20 Thread Rafael J. Wysocki
On Monday, March 20, 2017 09:27:45 AM Viresh Kumar wrote: > On 19-03-17, 14:34, Rafael J. Wysocki wrote: > > From: Rafael J. Wysocki > > > > The PELT metric used by the schedutil governor underestimates the > > CPU utilization in some cases. The reason for that may be time spent > > in interrupt

Re: [RFC][PATCH 2/2] cpufreq: schedutil: Force max frequency on busy CPUs

2017-03-20 Thread Rafael J. Wysocki
On Monday, March 20, 2017 01:50:09 PM Peter Zijlstra wrote: > On Mon, Mar 20, 2017 at 01:35:12PM +0100, Rafael J. Wysocki wrote: > > On Monday, March 20, 2017 11:36:45 AM Peter Zijlstra wrote: > > > On Sun, Mar 19, 2017 at 02:34:32PM +0100, Rafael J. Wysocki wrote: > > > > From: Rafael J. Wysocki

Re: [RFC][PATCH 2/2] cpufreq: schedutil: Force max frequency on busy CPUs

2017-03-20 Thread Rafael J. Wysocki
On Monday, March 20, 2017 09:26:34 AM Vincent Guittot wrote: > On 20 March 2017 at 04:57, Viresh Kumar wrote: > > On 19-03-17, 14:34, Rafael J. Wysocki wrote: > >> From: Rafael J. Wysocki > >> > >> The PELT metric used by the schedutil governor underestimates the > >> CPU utilization in some case

Re: [RFC][PATCH 2/2] cpufreq: schedutil: Force max frequency on busy CPUs

2017-03-20 Thread Patrick Bellasi
On 20-Mar 13:50, Peter Zijlstra wrote: > On Mon, Mar 20, 2017 at 01:35:12PM +0100, Rafael J. Wysocki wrote: > > On Monday, March 20, 2017 11:36:45 AM Peter Zijlstra wrote: > > > On Sun, Mar 19, 2017 at 02:34:32PM +0100, Rafael J. Wysocki wrote: > > > > From: Rafael J. Wysocki > > > > > > > > The

Re: [RFC][PATCH 2/2] cpufreq: schedutil: Force max frequency on busy CPUs

2017-03-20 Thread Peter Zijlstra
On Mon, Mar 20, 2017 at 01:35:12PM +0100, Rafael J. Wysocki wrote: > On Monday, March 20, 2017 11:36:45 AM Peter Zijlstra wrote: > > On Sun, Mar 19, 2017 at 02:34:32PM +0100, Rafael J. Wysocki wrote: > > > From: Rafael J. Wysocki > > > > > > The PELT metric used by the schedutil governor underest

Re: [RFC][PATCH 2/2] cpufreq: schedutil: Force max frequency on busy CPUs

2017-03-20 Thread Rafael J. Wysocki
On Monday, March 20, 2017 11:38:15 AM Peter Zijlstra wrote: > On Sun, Mar 19, 2017 at 10:24:24PM +0100, Rafael J. Wysocki wrote: > > > Honestly, if the processor had been capable of doing per-core P-states, that > > would have been a disaster and there are customers who wouldn't look at > > schedu

Re: [RFC][PATCH 2/2] cpufreq: schedutil: Force max frequency on busy CPUs

2017-03-20 Thread Rafael J. Wysocki
On Monday, March 20, 2017 11:36:45 AM Peter Zijlstra wrote: > On Sun, Mar 19, 2017 at 02:34:32PM +0100, Rafael J. Wysocki wrote: > > From: Rafael J. Wysocki > > > > The PELT metric used by the schedutil governor underestimates the > > CPU utilization in some cases. The reason for that may be tim

Re: [RFC][PATCH 2/2] cpufreq: schedutil: Force max frequency on busy CPUs

2017-03-20 Thread Patrick Bellasi
On 20-Mar 09:26, Vincent Guittot wrote: > On 20 March 2017 at 04:57, Viresh Kumar wrote: > > On 19-03-17, 14:34, Rafael J. Wysocki wrote: > >> From: Rafael J. Wysocki > >> > >> The PELT metric used by the schedutil governor underestimates the > >> CPU utilization in some cases. The reason for th

Re: [RFC][PATCH 2/2] cpufreq: schedutil: Force max frequency on busy CPUs

2017-03-20 Thread Peter Zijlstra
On Sun, Mar 19, 2017 at 02:34:32PM +0100, Rafael J. Wysocki wrote: > From: Rafael J. Wysocki > > The PELT metric used by the schedutil governor underestimates the > CPU utilization in some cases. The reason for that may be time spent > in interrupt handlers and similar which is not accounted for

Re: [RFC][PATCH 2/2] cpufreq: schedutil: Force max frequency on busy CPUs

2017-03-20 Thread Peter Zijlstra
On Sun, Mar 19, 2017 at 10:24:24PM +0100, Rafael J. Wysocki wrote: > Honestly, if the processor had been capable of doing per-core P-states, that > would have been a disaster and there are customers who wouldn't look at > schedutil again after being confronted with these numbers. This, I feel, is

Re: [RFC][PATCH 2/2] cpufreq: schedutil: Force max frequency on busy CPUs

2017-03-20 Thread Vincent Guittot
On 20 March 2017 at 04:57, Viresh Kumar wrote: > On 19-03-17, 14:34, Rafael J. Wysocki wrote: >> From: Rafael J. Wysocki >> >> The PELT metric used by the schedutil governor underestimates the >> CPU utilization in some cases. The reason for that may be time spent >> in interrupt handlers and si

Re: [RFC][PATCH 2/2] cpufreq: schedutil: Force max frequency on busy CPUs

2017-03-19 Thread Viresh Kumar
On 19-03-17, 14:34, Rafael J. Wysocki wrote: > From: Rafael J. Wysocki > > The PELT metric used by the schedutil governor underestimates the > CPU utilization in some cases. The reason for that may be time spent > in interrupt handlers and similar which is not accounted for by PELT. > > That ca

Re: [RFC][PATCH 2/2] cpufreq: schedutil: Force max frequency on busy CPUs

2017-03-19 Thread Rafael J. Wysocki
On Sunday, March 19, 2017 10:24:24 PM Rafael J. Wysocki wrote: > On Sunday, March 19, 2017 02:34:32 PM Rafael J. Wysocki wrote: > > From: Rafael J. Wysocki > > > > The PELT metric used by the schedutil governor underestimates the > > CPU utilization in some cases. The reason for that may be time

Re: [RFC][PATCH 2/2] cpufreq: schedutil: Force max frequency on busy CPUs

2017-03-19 Thread Rafael J. Wysocki
On Sunday, March 19, 2017 02:34:32 PM Rafael J. Wysocki wrote: > From: Rafael J. Wysocki > > The PELT metric used by the schedutil governor underestimates the > CPU utilization in some cases. The reason for that may be time spent > in interrupt handlers and similar which is not accounted for by

[RFC][PATCH 2/2] cpufreq: schedutil: Force max frequency on busy CPUs

2017-03-19 Thread Rafael J. Wysocki
From: Rafael J. Wysocki The PELT metric used by the schedutil governor underestimates the CPU utilization in some cases. The reason for that may be time spent in interrupt handlers and similar which is not accounted for by PELT. That can be easily demonstrated by running kernel compilation on a