From: Yunsheng Lin <linyunsheng@huawei.com>
To: "Michael S. Tsirkin" <mst@redhat.com>
Cc: <davem@davemloft.net>, <kuba@kernel.org>, <jasowang@redhat.com>,
<brouer@redhat.com>, <paulmck@kernel.org>, <peterz@infradead.org>,
<will@kernel.org>, <shuah@kernel.org>,
<linux-kernel@vger.kernel.org>, <netdev@vger.kernel.org>,
<linux-kselftest@vger.kernel.org>, <linuxarm@openeuler.org>
Subject: Re: [Linuxarm] Re: [PATCH net-next v2 2/2] ptr_ring: make __ptr_ring_empty() checking more reliable
Date: Mon, 28 Jun 2021 10:11:23 +0800 [thread overview]
Message-ID: <0372e117-71fc-5696-783d-43a58a013c8a@huawei.com> (raw)
In-Reply-To: <20210627020440-mutt-send-email-mst@kernel.org>
On 2021/6/27 14:07, Michael S. Tsirkin wrote:
> On Fri, Jun 25, 2021 at 05:20:10PM +0800, Yunsheng Lin wrote:
>> On 2021/6/25 14:39, Michael S. Tsirkin wrote:
>>> On Fri, Jun 25, 2021 at 11:18:56AM +0800, Yunsheng Lin wrote:
>>>> Currently r->queue[] is cleared after r->consumer_head is moved
>>>> forward, which makes the __ptr_ring_empty() checking called in
>>>> page_pool_refill_alloc_cache() unreliable if the checking is done
>>>> after the r->queue clearing and before the consumer_head moving
>>>> forward.
>>>>
>>>> Move the r->queue[] clearing after consumer_head moving forward
>>>> to make __ptr_ring_empty() checking more reliable.
>>>>
>>>> As a side effect of above change, a consumer_head checking is
>>>> avoided for the likely case, and it has noticeable performance
>>>> improvement when it is tested using the ptr_ring_test selftest
>>>> added in the previous patch.
>>>>
>>>> Using "taskset -c 1 ./ptr_ring_test -s 1000 -m 0 -N 100000000"
>>>> to test the case of single thread doing both the enqueuing and
>>>> dequeuing:
>>>>
>>>> arch unpatched patched delta
>>>> arm64 4648 ms 4464 ms +3.9%
>>>> X86 2562 ms 2401 ms +6.2%
>>>>
>>>> Using "taskset -c 1-2 ./ptr_ring_test -s 1000 -m 1 -N 100000000"
>>>> to test the case of one thread doing enqueuing and another thread
>>>> doing dequeuing concurrently, also known as single-producer/single-
>>>> consumer:
>>>>
>>>> arch unpatched patched delta
>>>> arm64 3624 ms + 3624 ms 3462 ms + 3462 ms +4.4%
>>>> x86 2758 ms + 2758 ms 2547 ms + 2547 ms +7.6%
>>>
>>> Nice but it's small - could be a fluke.
>>> How many tests did you run? What is the variance?
>>> Did you try pinning to different CPUs to observe numa effects?
>>> Please use perf or some other modern tool for this kind
>>> of benchmark. Thanks!
>>
>> The result is quite stable, and retest using perf stat:
>
> How stable exactly? Try with -r so we can find out.
Retest with "perf stat -r":
For unpatched one:
Performance counter stats for './ptr_ring_test -s 1000 -m 1 -N 100000000' (100 runs):
6780.97 msec task-clock # 2.000 CPUs utilized ( +- 5.36% )
73 context-switches # 0.011 K/sec ( +- 5.07% )
0 cpu-migrations # 0.000 K/sec ( +-100.00% )
81 page-faults # 0.012 K/sec ( +- 0.76% )
17629544748 cycles # 2.600 GHz ( +- 5.36% )
25496488950 instructions # 1.45 insn per cycle ( +- 0.26% )
<not supported> branches
11489031 branch-misses ( +- 1.69% )
3.391 +- 0.182 seconds time elapsed ( +- 5.35% )
For patched one:
Performance counter stats for './ptr_ring_test_opt -s 1000 -m 1 -N 100000000' (100 runs):
6567.83 msec task-clock # 2.000 CPUs utilized ( +- 5.53% )
71 context-switches # 0.011 K/sec ( +- 5.26% )
0 cpu-migrations # 0.000 K/sec
82 page-faults # 0.012 K/sec ( +- 0.85% )
17075489298 cycles # 2.600 GHz ( +- 5.53% )
23861051578 instructions # 1.40 insn per cycle ( +- 0.07% )
<not supported> branches
10473776 branch-misses ( +- 0.60% )
3.284 +- 0.182 seconds time elapsed ( +- 5.53% )
The result is more stable when using taskset to limit the running cpu, but I suppose
the above data is stable enough to justify the performance improvement?
next prev parent reply other threads:[~2021-06-28 2:12 UTC|newest]
Thread overview: 20+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-06-25 3:18 [PATCH net-next v2 0/2] add benchmark selftest and optimization for ptr_ring Yunsheng Lin
2021-06-25 3:18 ` [PATCH net-next v2 1/2] selftests/ptr_ring: add benchmark application " Yunsheng Lin
2021-06-25 3:36 ` Jason Wang
2021-06-25 3:52 ` Yunsheng Lin
2021-06-27 6:09 ` Michael S. Tsirkin
2021-06-28 1:42 ` Yunsheng Lin
2021-06-25 6:37 ` Michael S. Tsirkin
2021-06-25 7:40 ` Yunsheng Lin
2021-06-25 3:18 ` [PATCH net-next v2 2/2] ptr_ring: make __ptr_ring_empty() checking more reliable Yunsheng Lin
2021-06-25 6:32 ` Michael S. Tsirkin
2021-06-25 7:21 ` Yunsheng Lin
2021-06-25 7:30 ` Michael S. Tsirkin
2021-06-25 8:33 ` Yunsheng Lin
2021-06-27 6:03 ` Michael S. Tsirkin
2021-06-28 2:17 ` Yunsheng Lin
2021-06-25 6:39 ` Michael S. Tsirkin
2021-06-25 9:20 ` Yunsheng Lin
2021-06-27 6:07 ` Michael S. Tsirkin
2021-06-28 2:11 ` Yunsheng Lin [this message]
2021-06-25 6:42 ` [PATCH net-next v2 0/2] add benchmark selftest and optimization for ptr_ring Michael S. Tsirkin
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=0372e117-71fc-5696-783d-43a58a013c8a@huawei.com \
--to=linyunsheng@huawei.com \
--cc=brouer@redhat.com \
--cc=davem@davemloft.net \
--cc=jasowang@redhat.com \
--cc=kuba@kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-kselftest@vger.kernel.org \
--cc=linuxarm@openeuler.org \
--cc=mst@redhat.com \
--cc=netdev@vger.kernel.org \
--cc=paulmck@kernel.org \
--cc=peterz@infradead.org \
--cc=shuah@kernel.org \
--cc=will@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).