Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Wed, 5 Jun 2019 20:00:35 +0200
From:   Peter Zijlstra <peterz@infradead.org>
To:     Matt Fleming <matt@codeblueprint.co.uk>
Cc:     linux-kernel@vger.kernel.org,
        "Suthikulpanit, Suravee" <Suravee.Suthikulpanit@amd.com>,
        Mel Gorman <mgorman@techsingularity.net>,
        "Lendacky, Thomas" <Thomas.Lendacky@amd.com>,
        Borislav Petkov <bp@alien8.de>
Subject: Re: [PATCH] sched/topology: Improve load balancing on AMD EPYC
Message-ID: <20190605180035.GA3402@hirez.programming.kicks-ass.net>
References: <20190605155922.17153-1-matt@codeblueprint.co.uk>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20190605155922.17153-1-matt@codeblueprint.co.uk>
User-Agent: Mutt/1.10.1 (2018-07-13)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Wed, Jun 05, 2019 at 04:59:22PM +0100, Matt Fleming wrote:
> SD_BALANCE_{FORK,EXEC} and SD_WAKE_AFFINE are stripped in sd_init()
> for any sched domains with a NUMA distance greater than 2 hops
> (RECLAIM_DISTANCE). The idea being that it's expensive to balance
> across domains that far apart.
> 
> However, as is rather unfortunately explained in
> 
>   commit 32e45ff43eaf ("mm: increase RECLAIM_DISTANCE to 30")
> 
> the value for RECLAIM_DISTANCE is based on node distance tables from
> 2011-era hardware.
> 
> Current AMD EPYC machines have the following NUMA node distances:
> 
> node distances:
> node   0   1   2   3   4   5   6   7
>   0:  10  16  16  16  32  32  32  32
>   1:  16  10  16  16  32  32  32  32
>   2:  16  16  10  16  32  32  32  32
>   3:  16  16  16  10  32  32  32  32
>   4:  32  32  32  32  10  16  16  16
>   5:  32  32  32  32  16  10  16  16
>   6:  32  32  32  32  16  16  10  16
>   7:  32  32  32  32  16  16  16  10
> 
> where 2 hops is 32.
> 
> The result is that the scheduler fails to load balance properly across
> NUMA nodes on different sockets -- 2 hops apart.
> 

> Update the code in sd_init() to account for modern node distances, and
> maintaining backward-compatible behaviour by respecting
> RECLAIM_DISTANCE for distances more than 2 hops.

And then we had two magic values :/

Should we not 'fix' RECLAIM_DISTANCE for EPYC or something? Because
surely, if we want to load-balance agressively over 30, then so too
should we do node_reclaim() I'm thikning.