From: Joerg Roedel <joro-zLv9SwRftAIdnm+yROfE0A@public.gmane.org>
To: Ray Jui <ray.jui-dY08KVG/lbpWk0Htik3J/w@public.gmane.org>
Cc: sunil.goutham-YGCgFSpz5w/QT0dZR+AlfA@public.gmane.org,
will.deacon-5wv7dgnIgG8@public.gmane.org,
iommu-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org,
linu.cherian-YGCgFSpz5w/QT0dZR+AlfA@public.gmane.org,
linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r@public.gmane.org
Subject: Re: [PATCH 0/8] io-pgtable lock removal
Date: Wed, 21 Jun 2017 17:47:26 +0200 [thread overview]
Message-ID: <20170621154726.GQ30388@8bytes.org> (raw)
In-Reply-To: <b7830be1-9a78-e29f-a29c-4798aaa28c0a-dY08KVG/lbpWk0Htik3J/w@public.gmane.org>
On Wed, Jun 14, 2017 at 05:40:30PM -0700, Ray Jui wrote:
> With the NVMf target test with 4 SSDs, fio based test, random read, 4k,
> 8 jobs:
>
> Without IOMMU:
>
> IOPS = 1080K
>
> With IOMMU, but without your latest patch:
>
> IOPS = 520K
>
> With IOMMU and your latest patch:
>
> IOPS = 500K ~ 850K (a lot of variation observed during the same test run)
That variation might come from the RB-Tree used in the IOVA allocator.
For block-device workloads the allocation size of iova ranges might be
bigger than what is cached in the magazines, so that the fall-back to
the old (locked) allocator is used.
Joerg
WARNING: multiple messages have this Message-ID (diff)
From: joro@8bytes.org (Joerg Roedel)
To: linux-arm-kernel@lists.infradead.org
Subject: [PATCH 0/8] io-pgtable lock removal
Date: Wed, 21 Jun 2017 17:47:26 +0200 [thread overview]
Message-ID: <20170621154726.GQ30388@8bytes.org> (raw)
In-Reply-To: <b7830be1-9a78-e29f-a29c-4798aaa28c0a@broadcom.com>
On Wed, Jun 14, 2017 at 05:40:30PM -0700, Ray Jui wrote:
> With the NVMf target test with 4 SSDs, fio based test, random read, 4k,
> 8 jobs:
>
> Without IOMMU:
>
> IOPS = 1080K
>
> With IOMMU, but without your latest patch:
>
> IOPS = 520K
>
> With IOMMU and your latest patch:
>
> IOPS = 500K ~ 850K (a lot of variation observed during the same test run)
That variation might come from the RB-Tree used in the IOVA allocator.
For block-device workloads the allocation size of iova ranges might be
bigger than what is cached in the magazines, so that the fall-back to
the old (locked) allocator is used.
Joerg
next prev parent reply other threads:[~2017-06-21 15:47 UTC|newest]
Thread overview: 50+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-06-08 11:51 [PATCH 0/8] io-pgtable lock removal Robin Murphy
2017-06-08 11:51 ` Robin Murphy
2017-06-08 11:52 ` [PATCH 1/8] iommu/io-pgtable-arm-v7s: Check table PTEs more precisely Robin Murphy
2017-06-08 11:52 ` Robin Murphy
2017-06-08 11:52 ` [PATCH 2/8] iommu/io-pgtable-arm: Improve split_blk_unmap Robin Murphy
2017-06-08 11:52 ` Robin Murphy
2017-06-08 11:52 ` [PATCH 3/8] iommu/io-pgtable-arm-v7s: Refactor split_blk_unmap Robin Murphy
2017-06-08 11:52 ` Robin Murphy
2017-06-08 11:52 ` [PATCH 4/8] iommu/io-pgtable: Introduce explicit coherency Robin Murphy
2017-06-08 11:52 ` Robin Murphy
2017-06-08 11:52 ` [PATCH 5/8] iommu/io-pgtable-arm: Support lockless operation Robin Murphy
2017-06-08 11:52 ` Robin Murphy
2017-06-08 11:52 ` [PATCH 6/8] iommu/io-pgtable-arm-v7s: " Robin Murphy
2017-06-08 11:52 ` Robin Murphy
2017-06-08 11:52 ` [PATCH 7/8] iommu/arm-smmu: Remove io-pgtable spinlock Robin Murphy
2017-06-08 11:52 ` Robin Murphy
2017-06-08 11:52 ` [PATCH 8/8] iommu/arm-smmu-v3: " Robin Murphy
2017-06-08 11:52 ` Robin Murphy
[not found] ` <cover.1496921366.git.robin.murphy-5wv7dgnIgG8@public.gmane.org>
2017-06-09 19:28 ` [PATCH 0/8] io-pgtable lock removal Nate Watterson
2017-06-09 19:28 ` Nate Watterson
[not found] ` <458ad41d-6679-eeca-3c0f-13ccb6c933b6-sgV2jX0FEOL9JmXXK+q4OQ@public.gmane.org>
2017-06-15 0:40 ` Ray Jui via iommu
2017-06-15 0:40 ` Ray Jui
[not found] ` <b7830be1-9a78-e29f-a29c-4798aaa28c0a-dY08KVG/lbpWk0Htik3J/w@public.gmane.org>
2017-06-15 12:25 ` John Garry
2017-06-15 12:25 ` John Garry
2017-06-20 13:37 ` Robin Murphy
2017-06-20 13:37 ` Robin Murphy
[not found] ` <cdc1799b-f142-09ed-a7e5-d7fd2e70268f-5wv7dgnIgG8@public.gmane.org>
2017-06-27 16:43 ` Ray Jui via iommu
2017-06-27 16:43 ` Ray Jui
[not found] ` <e43ba1fe-696e-fabb-a800-52fadaa2fa93-dY08KVG/lbpWk0Htik3J/w@public.gmane.org>
2017-06-28 11:46 ` Will Deacon
2017-06-28 11:46 ` Will Deacon
[not found] ` <20170628114609.GD11053-5wv7dgnIgG8@public.gmane.org>
2017-06-28 17:02 ` Ray Jui via iommu
2017-06-28 17:02 ` Ray Jui
[not found] ` <87d53115-3d80-5a3d-6632-c31986cb7018-dY08KVG/lbpWk0Htik3J/w@public.gmane.org>
2017-07-04 17:31 ` Will Deacon
2017-07-04 17:31 ` Will Deacon
[not found] ` <20170704173155.GN22175-5wv7dgnIgG8@public.gmane.org>
2017-07-04 17:39 ` Ray Jui via iommu
2017-07-04 17:39 ` Ray Jui
[not found] ` <6814b246-22f0-bfaa-5002-a269b2735116-dY08KVG/lbpWk0Htik3J/w@public.gmane.org>
2017-07-05 1:45 ` Ray Jui via iommu
2017-07-05 1:45 ` Ray Jui
[not found] ` <2d5f5ef3-32b1-76c6-6869-ff980557f8e8-dY08KVG/lbpWk0Htik3J/w@public.gmane.org>
2017-07-05 8:41 ` Will Deacon
2017-07-05 8:41 ` Will Deacon
[not found] ` <20170705084143.GA9378-5wv7dgnIgG8@public.gmane.org>
2017-07-05 23:24 ` Ray Jui via iommu
2017-07-05 23:24 ` Ray Jui
[not found] ` <5149280b-a214-249c-c5e2-3712b1f941d2-dY08KVG/lbpWk0Htik3J/w@public.gmane.org>
2017-07-06 15:08 ` Will Deacon
2017-07-06 15:08 ` Will Deacon
[not found] ` <20170706150838.GB15574-5wv7dgnIgG8@public.gmane.org>
2017-07-06 18:14 ` Ray Jui via iommu
2017-07-06 18:14 ` Ray Jui
[not found] ` <94ba5d4a-0dae-9394-79ef-90da86e49c86-dY08KVG/lbpWk0Htik3J/w@public.gmane.org>
2017-07-07 12:46 ` Will Deacon
2017-07-07 12:46 ` Will Deacon
2017-06-21 15:47 ` Joerg Roedel [this message]
2017-06-21 15:47 ` Joerg Roedel
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=20170621154726.GQ30388@8bytes.org \
--to=joro-zlv9swrftaidnm+yrofe0a@public.gmane.org \
--cc=iommu-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org \
--cc=linu.cherian-YGCgFSpz5w/QT0dZR+AlfA@public.gmane.org \
--cc=linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r@public.gmane.org \
--cc=ray.jui-dY08KVG/lbpWk0Htik3J/w@public.gmane.org \
--cc=sunil.goutham-YGCgFSpz5w/QT0dZR+AlfA@public.gmane.org \
--cc=will.deacon-5wv7dgnIgG8@public.gmane.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.