[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-Id: <20210920142614.4891-1-mgorman@techsingularity.net>
Date: Mon, 20 Sep 2021 15:26:12 +0100
From: Mel Gorman <mgorman@...hsingularity.net>
To: Peter Zijlstra <peterz@...radead.org>
Cc: Ingo Molnar <mingo@...nel.org>,
Vincent Guittot <vincent.guittot@...aro.org>,
Valentin Schneider <valentin.schneider@....com>,
Aubrey Li <aubrey.li@...ux.intel.com>,
Barry Song <song.bao.hua@...ilicon.com>,
Srikar Dronamraju <srikar@...ux.vnet.ibm.com>,
LKML <linux-kernel@...r.kernel.org>,
Mel Gorman <mgorman@...hsingularity.net>
Subject: [PATCH 0/2] Scale wakeup granularity relative to nr_running
It's unfortunate this is colliding with Plumbers but part of the
discussions are on select_idle_sibling and hackbench is a common
reference workload for evaluating select_idle_sibling. The
downside is that hackbench is sensitive to other factors much
more than the SIS cost. A major component is the value of
kernel.sched_wakeup_granularity_ns and this series tackles that
bit.
Patch 1 is minor, it was spotted while developing patch 2.
Patch 2 scales kernel.sched_wakeup_granularity_ns so allow
tasks to avoid preemption longer when the machine
is heavily overloaded.
--
2.31.1
Powered by blists - more mailing lists