From: keith.busch@intel.com (Keith Busch)
Subject: [PATCH rfc 0/6] convert nvme pci to use irq-poll service
Date: Wed, 5 Oct 2016 17:47:31 -0400 [thread overview]
Message-ID: <20161005214730.GA1053@localhost.localdomain> (raw)
In-Reply-To: <1475660534-16681-1-git-send-email-sagi@grimberg.me>
On Wed, Oct 05, 2016@12:42:08PM +0300, Sagi Grimberg wrote:
> I ran some tests with this and it seemed to work pretty well with
> my low-end nvme devices. One phenomenon I've encountered was that
> for single core long queue-depth'ed randread workload I saw around
> ~8-10% iops decrease. However when running multi-core IO I didn't
> see any noticeable performance degradation. non-polling Canonical
> randread latency doesn't seem to be affected as well. And also
> polling mode IO is not affected as expected.
>
> So in addition for review and feedback, this is a call for testing
> and benchmarking as this touches the critical data path.
Hi Sagi,
Just reconfirming your findings with another data point, I ran this on
controllers with 3D Xpoint media, and single depth 4k random read latency
increased almost 7%. I'll try see if there's anything else we can do to
bring that in.
next prev parent reply other threads:[~2016-10-05 21:47 UTC|newest]
Thread overview: 27+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-10-05 9:42 [PATCH rfc 0/6] convert nvme pci to use irq-poll service Sagi Grimberg
2016-10-05 9:42 ` [PATCH rfc 1/6] nvme-pci: Split __nvme_process_cq to poll and handle Sagi Grimberg
2016-10-05 13:21 ` Johannes Thumshirn
2016-10-05 16:52 ` Sagi Grimberg
2016-10-05 19:49 ` Jon Derrick
2016-10-10 7:25 ` Johannes Thumshirn
2016-10-05 9:42 ` [PATCH rfc 2/6] nvme-pci: Add budget to __nvme_process_cq Sagi Grimberg
2016-10-05 13:26 ` Johannes Thumshirn
2016-10-05 9:42 ` [PATCH rfc 3/6] nvme-pci: Use irq-poll for completion processing Sagi Grimberg
2016-10-05 13:40 ` Johannes Thumshirn
2016-10-05 16:57 ` Sagi Grimberg
2016-10-10 7:47 ` Johannes Thumshirn
2016-10-05 9:42 ` [PATCH rfc 4/6] nvme: don't consume cq in queue_rq Sagi Grimberg
2016-10-05 13:42 ` Johannes Thumshirn
2016-10-05 9:42 ` [PATCH rfc 5/6] nvme-pci: open-code polling logic in nvme_poll Sagi Grimberg
2016-10-05 13:52 ` Johannes Thumshirn
2016-10-05 17:02 ` Sagi Grimberg
2016-10-10 7:49 ` Johannes Thumshirn
2016-10-05 9:42 ` [PATCH rfc 6/6] nvme-pci: Get rid of threaded interrupts Sagi Grimberg
2016-10-05 13:47 ` Johannes Thumshirn
2016-10-05 21:47 ` Keith Busch [this message]
2016-10-05 21:55 ` [PATCH rfc 0/6] convert nvme pci to use irq-poll service Sagi Grimberg
2016-10-05 22:49 ` Keith Busch
2016-10-05 22:48 ` Sagi Grimberg
2016-10-27 10:09 ` Johannes Thumshirn
2016-10-27 10:31 ` Sagi Grimberg
2016-10-27 10:58 ` Johannes Thumshirn
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=20161005214730.GA1053@localhost.localdomain \
--to=keith.busch@intel.com \
/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