[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <f7627afa58294c4480d8f507ee792c2f@AcuMS.aculab.com>
Date: Fri, 4 Aug 2023 14:04:21 +0000
From: David Laight <David.Laight@...LAB.COM>
To: 'Shrikanth Hegde' <sshegde@...ux.vnet.ibm.com>,
Mathieu Desnoyers <mathieu.desnoyers@...icios.com>
CC: "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
Ingo Molnar <mingo@...hat.com>,
Valentin Schneider <vschneid@...hat.com>,
Steven Rostedt <rostedt@...dmis.org>,
Ben Segall <bsegall@...gle.com>, Mel Gorman <mgorman@...e.de>,
Daniel Bristot de Oliveira <bristot@...hat.com>,
"Vincent Guittot" <vincent.guittot@...aro.org>,
Juri Lelli <juri.lelli@...hat.com>,
Swapnil Sapkal <Swapnil.Sapkal@....com>,
Aaron Lu <aaron.lu@...el.com>,
"x86@...nel.org" <x86@...nel.org>,
Peter Zijlstra <peterz@...radead.org>,
Srikar Dronamraju <srikar@...ux.vnet.ibm.com>
Subject: RE: [RFC PATCH 1/1] sched: Extend cpu idle state for 1ms
From: Shrikanth Hegde
> Sent: 26 July 2023 09:05
...
> > + if (sched_clock() < READ_ONCE(rq->idle_end_time) + IDLE_CPU_DELAY_NS)
>
>
> Wouldn't this hurt the latency badly? Specially on a loaded system with
> a workload that does a lot of wakeup.
Having spotted this I'm also rather worried about systems
that are doing (eg) real time audio and need to wakeup a
lot of threads (less than the number of cpu) every (say) 10ms.
It is hard enough waking up a lot of threads quickly without
another 1ms delay being added.
(I'm only talking about 30 threads as well, not 300.)
David
-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)
Powered by blists - more mailing lists