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;
Date:   Thu, 18 Feb 2021 15:16:44 -0800
From:   Andrew Morton <akpm@linux-foundation.org>
To:     Jens Axboe <axboe@kernel.dk>
Cc:     "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH] percpu_counter: increase batch count
Message-Id: <20210218151644.df430e4f77f763b7db2a004f@linux-foundation.org>
In-Reply-To: <0bf90e07-8758-b238-b3f3-a330725a1134@kernel.dk>
References: <0bf90e07-8758-b238-b3f3-a330725a1134@kernel.dk>
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Precedence: bulk

On Thu, 18 Feb 2021 14:36:31 -0700 Jens Axboe <axboe@kernel.dk> wrote:

> Currently we cap the batch count at max(32, 2*nr_online_cpus), which these
> days is kind of silly as systems have gotten much bigger than in 2009 when
> this heuristic was introduced.
> 
> Bump it to capping it at 256 instead. This has a noticeable improvement
> for certain io_uring workloads, as io_uring tracks per-task inflight count
> using percpu counters.
> 

It will also make percpu_counter_read() and
percpu_counter_read_positive() more inaccurate than at present.  Any
effects from this will take a while to discover.

But yes, worth trying - I'll add it to the post-rc1 pile.