Date: Fri, 8 Apr 2016 09:22:45 -0700
From: Juri Lelli <juri.lelli@arm.com>
To: Peter Zijlstra <peterz@infradead.org>
Cc: Joe Perches <joe@perches.com>, Yuyang Du <yuyang.du@intel.com>,
        mingo@kernel.org, linux-kernel@vger.kernel.org, bsegall@google.com,
        pjt@google.com, morten.rasmussen@arm.com, vincent.guittot@linaro.org,
        dietmar.eggemann@arm.com
Subject: Re: [PATCH] sched/fair: Optimize sum computation with a lookup table
Message-ID: <20160408162245.GB11241@pablo>
References: <1460081240-8074-1-git-send-email-yuyang.du@intel.com>
 <1460111501.1800.28.camel@perches.com>
 <20160408105423.GN3448@twins.programming.kicks-ass.net>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20160408105423.GN3448@twins.programming.kicks-ass.net>
User-Agent: Mutt/1.5.23 (2014-03-12)
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 974
Lines: 28

Hi,

On 08/04/16 12:54, Peter Zijlstra wrote:
> On Fri, Apr 08, 2016 at 03:31:41AM -0700, Joe Perches wrote:
> > On Fri, 2016-04-08 at 10:07 +0800, Yuyang Du wrote:
> > > __compute_runnable_contrib() uses a loop to compute sum, whereas a
> > > table lookup can do it faster in a constant time.
> > 
> > Perhaps this becomes rather fragile code overly dependent on the
> > current #define values of LOAD_AVG_MAX_N and LOAD_AVG_PERIOD.
> 
> Too late for that, these tables already heavily depend on these values.
> 
> Commit 5b51f2f80b3b ("sched: Make __update_entity_runnable_avg() fast"),
> which introduced these tables, includes a program to re-compute the
> values if we ever need to change them.
> 
> It might maybe make sense for Yuyang to update that program to also
> compute this new table and include the whole new program in the
> Changelog of this patch.
> 

+1 on this. That program turns out to be really useful if one needs to
recompute tables.

Best,

- Juri