From: Rusty Russell <rusty@rustcorp.com.au>
To: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>,
linux-kernel@vger.kernel.org,
virtualization@lists.linux-foundation.org
Subject: Re: [PATCH 02/16] virtio_ring: virtqueue_add_sgs, to add multiple sgs.
Date: Wed, 27 Feb 2013 21:51:09 +1030 [thread overview]
Message-ID: <87fw0idmu2.fsf@rustcorp.com.au> (raw)
In-Reply-To: <20130227074920.GA11775@redhat.com>
"Michael S. Tsirkin" <mst@redhat.com> writes:
> On Wed, Feb 27, 2013 at 05:58:37PM +1030, Rusty Russell wrote:
>> Paolo Bonzini <pbonzini@redhat.com> writes:
>> > Il 24/02/2013 23:12, Michael S. Tsirkin ha scritto:
>> >> On Tue, Feb 19, 2013 at 06:26:20PM +1030, Rusty Russell wrote:
>> >>> virtio_scsi can really use this, to avoid the current hack of copying
>> >>> the whole sg array. Some other things get slightly neater, too.
>> >>>
>> >>> Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
>> >>
>> >> Hmm, this makes add_buf a bit slower. virtio_test results
>> >> (I'll send a patch to update the test shortly):
>> >>
>> >> Before:
>> >> 0.09user 0.01system 0:00.12elapsed 91%CPU (0avgtext+0avgdata 480maxresident)k
>> >> 0inputs+0outputs (0major+145minor)pagefaults 0swaps
>> >>
>> >> After:
>> >> 0.11user 0.01system 0:00.13elapsed 90%CPU (0avgtext+0avgdata 480maxresident)k
>> >> 0inputs+0outputs (0major+145minor)pagefaults 0swaps
>> >
>> > Not unexpected at all... :(
>> >
>> > Some of it can be recovered, but if it's 20% I doubt all of it. So my
>> > patches were not premature optimization; you really can take just two
>> > among speed, flexibility, and having a nice API.
>>
>> The error bars on this are far too large to say "20%".
>>
>> Here are my numbers, using 50 runs of:
>> time tools/virtio/vringh_test --indirect --eventidx --parallel and
>> stats --trim-outliers:
>>
>> Baseline (before add_sgs):
>> 2.840000-3.040000(2.927292)user
>>
>> After add_sgs:
>> 2.970000-3.150000(3.053750)user
>>
>> After simplifying add_buf a little:
>> 2.950000-3.210000(3.081458)user
>>
>> After inlining virtqueue_add/vring_add_indirect:
>> 2.920000-3.150000(3.026875)user
>>
>> After passing in iteration functions (chained vs unchained):
>> 2.760000-2.970000(2.883542)user
Oops. This result (and the next) is bogus. I was playing with -O3, and
accidentally left that in :(
The final result was 3.005208, ie. 3% slowdown. Which almost makes it
worth duplicating the whole set of code :(
>> After removing the now-unnecessary chain-cleaning in add_buf:
>> 2.660000-2.830000(2.753542)user
>>
>> Any questions?
>> Rusty.
>
> Sorry, so which patches are included in the last stage?
> Something I didn't make clear: I tested 2/16 (the patch I replied to).
I wanted to tidy them up, add commentry, and integrate your tool cleanup
patches first. That's when I noticed my screwup.
I'll push them now, but I want to revisit to see if there's something
cleverer I can do...
Cheers,
Rusty.
next prev parent reply other threads:[~2013-02-27 23:59 UTC|newest]
Thread overview: 60+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-02-19 7:56 [PATCH 00/16] virtio ring rework Rusty Russell
2013-02-19 7:56 ` [PATCH 01/16] scatterlist: introduce sg_unmark_end Rusty Russell
2013-02-19 7:56 ` [PATCH 02/16] virtio_ring: virtqueue_add_sgs, to add multiple sgs Rusty Russell
2013-02-19 9:15 ` Wanlong Gao
2013-02-20 9:18 ` Asias He
2013-02-24 22:12 ` Michael S. Tsirkin
2013-02-26 5:14 ` Rusty Russell
2013-02-26 9:30 ` Michael S. Tsirkin
2013-02-26 7:02 ` Paolo Bonzini
2013-02-27 7:28 ` Rusty Russell
2013-02-27 7:49 ` Michael S. Tsirkin
2013-02-27 11:21 ` Rusty Russell [this message]
2013-02-28 9:24 ` Paolo Bonzini
2013-03-01 1:01 ` Rusty Russell
2013-02-19 7:56 ` [PATCH 03/16] virtio-blk: reorganize virtblk_add_req Rusty Russell
2013-02-19 7:56 ` [PATCH 04/16] virtio-blk: use virtqueue_start_buf on bio path Rusty Russell
2013-02-20 9:19 ` Asias He
2013-02-21 6:23 ` Rusty Russell
2013-02-19 7:56 ` [PATCH 05/16] virtio-blk: use virtqueue_add_sgs on req path Rusty Russell
2013-02-20 9:20 ` Asias He
2013-02-19 7:56 ` [PATCH 06/16] virtio_blk: remove nents member Rusty Russell
2013-02-20 9:20 ` Asias He
2013-02-19 7:56 ` [PATCH 07/16] virtio_ring: don't count elements twice for add_buf path Rusty Russell
2013-02-20 10:09 ` Wanlong Gao
2013-02-19 7:56 ` [PATCH 08/16] virtio_ring: virtqueue_add_outbuf / virtqueue_add_inbuf Rusty Russell
2013-02-20 10:09 ` Wanlong Gao
2013-02-21 17:09 ` Michael S. Tsirkin
2013-02-22 0:02 ` Rusty Russell
2013-02-25 21:35 ` Michael S. Tsirkin
2013-02-28 5:08 ` Rusty Russell
2013-02-28 7:01 ` Michael S. Tsirkin
2013-03-06 6:03 ` Rusty Russell
2013-02-19 7:56 ` [PATCH 09/16] virtio_net: use simplified virtqueue accessors Rusty Russell
2013-02-20 10:09 ` Wanlong Gao
2013-02-19 7:56 ` [PATCH 10/16] virtio_net: use virtqueue_add_sgs[] for command buffers Rusty Russell
2013-02-20 10:11 ` Wanlong Gao
2013-02-21 6:27 ` Rusty Russell
2013-02-21 8:30 ` Wanlong Gao
2013-02-21 9:41 ` Jason Wang
2013-02-21 9:43 ` Wanlong Gao
2013-02-19 7:56 ` [PATCH 11/16] virtio_rng: use simplified virtqueue accessors Rusty Russell
2013-02-20 10:12 ` Wanlong Gao
2013-02-19 7:56 ` [PATCH 12/16] virtio_console: " Rusty Russell
2013-02-20 10:12 ` Wanlong Gao
2013-02-19 7:56 ` [PATCH 13/16] caif_virtio: " Rusty Russell
2013-02-20 10:13 ` Wanlong Gao
2013-02-19 7:56 ` [PATCH 14/16] virtio_rpmsg_bus: " Rusty Russell
2013-02-20 10:14 ` Wanlong Gao
2013-02-19 7:56 ` [PATCH 15/16] virtio_balloon: " Rusty Russell
2013-02-20 10:15 ` Wanlong Gao
2013-02-19 7:56 ` [PATCH 16/16] 9p/trans_virtio.c: use virtio_add_sgs[] Rusty Russell
2013-02-19 9:15 ` [PATCH 00/16] virtio ring rework Paolo Bonzini
2013-02-21 6:30 ` Rusty Russell
2013-02-20 8:37 ` [PATCH 17/16] virtio-scsi: use virtqueue_add_sgs for command buffers Wanlong Gao
2013-02-20 9:38 ` Asias He
2013-02-20 9:41 ` Wanlong Gao
2013-02-20 9:47 ` [PATCH 17/16 V2] " Wanlong Gao
2013-02-20 10:54 ` Paolo Bonzini
2013-02-20 12:17 ` Asias He
2013-02-21 6:34 ` Rusty Russell
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=87fw0idmu2.fsf@rustcorp.com.au \
--to=rusty@rustcorp.com.au \
--cc=linux-kernel@vger.kernel.org \
--cc=mst@redhat.com \
--cc=pbonzini@redhat.com \
--cc=virtualization@lists.linux-foundation.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).