Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
IronPort-SDR: eJHLkP9lExaywZucswweuThWNZWESlDy92wjQG8UuwGgTkVu/16yRj+KlqyptAwsxv9AquyVw+
 lpytWFnmDgoQ==
IronPort-SDR: IFxsBVE6eMXGs+O+Ka4TEF45X9NKcB0cm+3+OfxpQy73jlEnTwolO/ucKXXv+fV62Xs0Myonlw
 sZY3FtFFncSA==
Subject: Re: [RFC][PATCH 1/5] sched/fair: Fix select_idle_cpu()s cost
 accounting
To:     Peter Zijlstra <peterz@infradead.org>
Cc:     mgorman@techsingularity.net, vincent.guittot@linaro.org,
        linux-kernel@vger.kernel.org, mingo@redhat.com,
        juri.lelli@redhat.com, valentin.schneider@arm.com,
        qais.yousef@arm.com, dietmar.eggemann@arm.com, rostedt@goodmis.org,
        bsegall@google.com, tim.c.chen@linux.intel.com, benbjiang@gmail.com
References: <20201214164822.402812729@infradead.org>
 <20201214170017.877557652@infradead.org>
 <c4e31235-e1fb-52ac-99a8-ae943ee0de54@linux.intel.com>
 <20201215075911.GA3040@hirez.programming.kicks-ass.net>
From:   "Li, Aubrey" <aubrey.li@linux.intel.com>
Message-ID: <ec2e0453-9fab-f7ac-effe-f64601806595@linux.intel.com>
Date:   Tue, 15 Dec 2020 20:13:07 +0800
User-Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:68.0) Gecko/20100101
 Thunderbird/68.9.0
MIME-Version: 1.0
In-Reply-To: <20201215075911.GA3040@hirez.programming.kicks-ass.net>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 2020/12/15 15:59, Peter Zijlstra wrote:
> On Tue, Dec 15, 2020 at 11:36:35AM +0800, Li, Aubrey wrote:
>> On 2020/12/15 0:48, Peter Zijlstra wrote:
>>> We compute the average cost of the total scan, but then use it as a
>>> per-cpu scan cost when computing the scan proportion. Fix this by
>>> properly computing a per-cpu scan cost.
>>>
>>> This also fixes a bug where we would terminate early (!--nr, case) and
>>> not account that cost at all.
>>
>> I'm a bit worried this may introduce a regression under heavy load.
>> The overhead of adding another cpu_clock() and calculation becomes 
>> significant when sis_scan is throttled by nr.
> 
> The thing is, the code as it exists today makes no sense what so ever.
> It's plain broken batshit.
> 
> We calculate the total scanning time (irrespective of how many CPUs we
> touched), and then use that calculate the number of cpus to scan. That's
> just daft.
> 
> After this patch we calculate the avg cost of scanning 1 cpu and use
> that to calculate how many cpus to scan. Which is coherent and sane.

I see and all of these make sense to me.

> 
> Maybe it can be improved, but that's a completely different thing.
> 

OK, I'll go through the workloads in hand and paste the data here.

Thanks,
-Aubrey