Re: [PATCH 1/2] sched/uclamp: Add a new sysctl to control RT default boost value

From: Dietmar Eggemann <dietmar.eggemann@arm.com>
To: Peter Zijlstra <peterz@infradead.org>, Qais Yousef <qais.yousef@arm.com>
Cc: Ingo Molnar <mingo@redhat.com>,
	Randy Dunlap <rdunlap@infradead.org>,
	Jonathan Corbet <corbet@lwn.net>,
	Juri Lelli <juri.lelli@redhat.com>,
	Vincent Guittot <vincent.guittot@linaro.org>,
	Steven Rostedt <rostedt@goodmis.org>,
	Ben Segall <bsegall@google.com>, Mel Gorman <mgorman@suse.de>,
	Luis Chamberlain <mcgrof@kernel.org>,
	Kees Cook <keescook@chromium.org>,
	Iurii Zaikin <yzaikin@google.com>,
	Quentin Perret <qperret@google.com>,
	Valentin Schneider <valentin.schneider@arm.com>,
	Patrick Bellasi <patrick.bellasi@matbug.net>,
	Pavan Kondeti <pkondeti@codeaurora.org>,
	linux-doc@vger.kernel.org, linux-kernel@vger.kernel.org,
	linux-fsdevel@vger.kernel.org
Subject: Re: [PATCH 1/2] sched/uclamp: Add a new sysctl to control RT default boost value
Date: Thu, 28 May 2020 21:20:34 +0200	[thread overview]
Message-ID: <f3081cf2-24f1-d0f9-e76b-d868538f3245@arm.com> (raw)
In-Reply-To: <20200528182913.GQ2483@worktop.programming.kicks-ass.net>

On 28/05/2020 20:29, Peter Zijlstra wrote:
> On Thu, May 28, 2020 at 05:51:31PM +0100, Qais Yousef wrote:
> 
>> In my head, the simpler version of
>>
>> 	if (rt_task(p) && !uc->user_defined)
>> 		// update_uclamp_min
>>
>> Is a single branch and write to cache, so should be fast. I'm failing to see
>> how this could generate an overhead tbh, but will not argue about it :-)
> 
> Mostly true; but you also had a load of that sysctl in there, which is
> likely to be a miss, and those are expensive.
> 
> Also; if we're going to have to optimize this, less logic is in there,
> the less we need to take out. Esp. for stuff that 'never' changes, like
> this.
> 
>>> It's more code, but it is all outside of the normal paths where we care
>>> about performance.
>>
>> I am happy to take that direction if you think it's worth it. I'm thinking
>> task_woken_rt() is good. But again, maybe I am missing something.
> 
> Basic rule, if the state 'never' changes, don't touch fast paths.
> 
> Such little things can be very difficult to measure, but at some point
> they cause death-by-a-thousnd-cuts.
> 
>>> Indeed, that one. The fact that regular distros cannot enable this
>>> feature due to performance overhead is unfortunate. It means there is a
>>> lot less potential for this stuff.
>>
>> I had a humble try to catch the overhead but wasn't successful. The observation
>> wasn't missed by us too then.
> 
> Right, I remember us doing benchmarks when we introduced all this and
> clearly we missed something. I would be good if Mel can share which
> benchmark hurt most so we can go have a look.

IIRC, it was a local mmtests netperf-udp with various buffer sizes?

At least that's what we're trying to run right now on a '2 Sockets Xeon
E5 2x10-Cores (40 CPUs)' w/ 3 different kernel ((1) wo_clamp (2)
tsk_uclamp (3) tskgrp_uclamp).

We have currently Ubuntu Desktop on it. I think that systemd uses
cgroups (especially cpu controller) differently on a (Ubuntu) Server.
Maybe this has an influence here as well?