From: Avi Kivity <avi@redhat.com>
To: Takuya Yoshikawa <takuya.yoshikawa@gmail.com>
Cc: linux-arch@vger.kernel.org, x86@kernel.org, arnd@arndb.de,
kvm@vger.kernel.org, kvm-ia64@vger.kernel.org,
fernando@oss.ntt.co.jp, mtosatti@redhat.com, agraf@suse.de,
kvm-ppc@vger.kernel.org, linux-kernel@vger.kernel.org,
yoshikawa.takuya@oss.ntt.co.jp, linuxppc-dev@ozlabs.org,
mingo@redhat.com, paulus@samba.org, hpa@zytor.com,
tglx@linutronix.de
Subject: Re: [RFC][PATCH 0/12] KVM, x86, ppc, asm-generic: moving dirty bitmaps to user space
Date: Mon, 10 May 2010 15:06:47 +0300 [thread overview]
Message-ID: <4BE7F6D7.3060005@redhat.com> (raw)
In-Reply-To: <20100504215645.6448af8f.takuya.yoshikawa@gmail.com>
On 05/04/2010 03:56 PM, Takuya Yoshikawa wrote:
> [Performance test]
>
> We measured the tsc needed to the ioctl()s for getting dirty logs in
> kernel.
>
> Test environment
>
> AMD Phenom(tm) 9850 Quad-Core Processor with 8GB memory
>
>
> 1. GUI test (running Ubuntu guest in graphical mode)
>
> sudo qemu-system-x86_64 -hda dirtylog_test.img -boot c -m 4192 -net ...
>
> We show a relatively stable part to compare how much time is needed
> for the basic parts of dirty log ioctl.
>
> get.org get.opt switch.opt
>
> slots[7].len=32768 278379 66398 64024
> slots[8].len=32768 181246 270 160
> slots[7].len=32768 263961 64673 64494
> slots[8].len=32768 181655 265 160
> slots[7].len=32768 263736 64701 64610
> slots[8].len=32768 182785 267 160
> slots[7].len=32768 260925 65360 65042
> slots[8].len=32768 182579 264 160
> slots[7].len=32768 267823 65915 65682
> slots[8].len=32768 186350 271 160
>
> At a glance, we know our optimization improved significantly compared
> to the original get dirty log ioctl. This is true for both get.opt and
> switch.opt. This has a really big impact for the personal KVM users who
> drive KVM in GUI mode on their usual PCs.
>
> Next, we notice that switch.opt improved a hundred nano seconds or so for
> these slots. Although this may sound a bit tiny improvement, we can feel
> this as a difference of GUI's responses like mouse reactions.
>
100 ns... this is a bit on the low side (and if you can measure it
interactively you have much better reflexes than I).
> To feel the difference, please try GUI on your PC with our patch series!
>
No doubt get.org -> get.opt is measurable, but get.opt->switch.opt is
problematic. Have you tried profiling to see where the time is spent
(well I can guess, clearing the write access from the sptes).
>
> 2. Live-migration test (4GB guest, write loop with 1GB buf)
>
> We also did a live-migration test.
>
> get.org get.opt switch.opt
>
> slots[0].len=655360 797383 261144 222181
> slots[1].len=3757047808 2186721 1965244 1842824
> slots[2].len=637534208 1433562 1012723 1031213
> slots[3].len=131072 216858 331 331
> slots[4].len=131072 121635 225 164
> slots[5].len=131072 120863 356 164
> slots[6].len=16777216 121746 1133 156
> slots[7].len=32768 120415 230 278
> slots[8].len=32768 120368 216 149
> slots[0].len=655360 806497 194710 223582
> slots[1].len=3757047808 2142922 1878025 1895369
> slots[2].len=637534208 1386512 1021309 1000345
> slots[3].len=131072 221118 459 296
> slots[4].len=131072 121516 272 166
> slots[5].len=131072 122652 244 173
> slots[6].len=16777216 123226 99185 149
> slots[7].len=32768 121803 457 505
> slots[8].len=32768 121586 216 155
> slots[0].len=655360 766113 211317 213179
> slots[1].len=3757047808 2155662 1974790 1842361
> slots[2].len=637534208 1481411 1020004 1031352
> slots[3].len=131072 223100 351 295
> slots[4].len=131072 122982 436 164
> slots[5].len=131072 122100 300 503
> slots[6].len=16777216 123653 779 151
> slots[7].len=32768 122617 284 157
> slots[8].len=32768 122737 253 149
>
> For slots other than 0,1,2 we can see the similar improvement.
>
> Considering the fact that switch.opt does not depend on the bitmap length
> except for kvm_mmu_slot_remove_write_access(), this is the cause of some
> usec to msec time consumption: there might be some context switches.
>
> But note that this was done with the workload which dirtied the memory
> endlessly during the live-migration.
>
> In usual workload, the number of dirty pages varies a lot for each iteration
> and we should gain really a lot for relatively clean cases.
>
Can you post such a test, for an idle large guest?
--
error compiling committee.c: too many arguments to function
next prev parent reply other threads:[~2010-05-10 12:07 UTC|newest]
Thread overview: 36+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-05-04 12:56 [RFC][PATCH 0/12] KVM, x86, ppc, asm-generic: moving dirty bitmaps to user space Takuya Yoshikawa
2010-05-04 12:58 ` [RFC][PATCH 1/12 applied today] KVM: x86: avoid unnecessary bitmap allocation when memslot is clean Takuya Yoshikawa
2010-05-04 13:00 ` [RFC][PATCH 2/12] KVM: introduce slot level dirty state management Takuya Yoshikawa
2010-05-04 13:01 ` [RFC][PATCH 3/12] KVM: introduce wrapper functions to create and destroy dirty bitmaps Takuya Yoshikawa
2010-05-04 13:02 ` [RFC][PATCH 4/12] x86: introduce copy_in_user() for 32-bit Takuya Yoshikawa
2010-05-04 13:02 ` [RFC][PATCH 5/12] x86: introduce __set_bit() like function for bitmaps in user space Takuya Yoshikawa
2010-05-04 13:03 ` [RFC][PATCH 6/12 not tested yet] PPC: introduce copy_in_user() for 32-bit Takuya Yoshikawa
2010-05-04 13:04 ` [RFC][PATCH 7/12 not tested yet] PPC: introduce __set_bit() like function for bitmaps in user space Takuya Yoshikawa
2010-05-11 16:00 ` Alexander Graf
2010-05-12 9:25 ` Takuya Yoshikawa
2010-05-04 13:05 ` [RFC][PATCH resend 8/12] asm-generic: bitops: introduce le bit offset macro Takuya Yoshikawa
2010-05-04 15:03 ` Arnd Bergmann
2010-05-04 16:08 ` Avi Kivity
2010-05-05 2:59 ` Takuya Yoshikawa
2010-05-06 13:38 ` Arnd Bergmann
2010-05-10 11:46 ` Takuya Yoshikawa
2010-05-10 12:01 ` Avi Kivity
2010-05-10 12:01 ` Arnd Bergmann
2010-05-10 12:09 ` Takuya Yoshikawa
2010-05-04 13:06 ` [RFC][PATCH 9/12] KVM: introduce a wrapper function of set_bit_user_non_atomic() Takuya Yoshikawa
2010-05-04 13:07 ` [RFC][PATCH RFC 10/12] KVM: move dirty bitmaps to user space Takuya Yoshikawa
2010-05-11 3:28 ` Marcelo Tosatti
2010-05-12 6:27 ` Takuya Yoshikawa
2010-05-04 13:08 ` [RFC][PATCH 11/12] KVM: introduce new API for getting/switching dirty bitmaps Takuya Yoshikawa
2010-05-11 3:43 ` Marcelo Tosatti
2010-05-11 5:53 ` Takuya Yoshikawa
2010-05-11 14:07 ` Marcelo Tosatti
2010-05-12 6:03 ` Takuya Yoshikawa
2010-05-04 13:11 ` [RFC][PATCH 12/12 sample] qemu-kvm: use " Takuya Yoshikawa
2010-05-10 12:06 ` Avi Kivity [this message]
2010-05-10 12:26 ` [RFC][PATCH 0/12] KVM, x86, ppc, asm-generic: moving dirty bitmaps to user space Takuya Yoshikawa
2010-05-11 10:11 ` Takuya Yoshikawa
2010-05-13 11:47 ` Avi Kivity
2010-05-17 9:06 ` Takuya Yoshikawa
2010-05-11 15:55 ` Alexander Graf
2010-05-12 9:19 ` Takuya Yoshikawa
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=4BE7F6D7.3060005@redhat.com \
--to=avi@redhat.com \
--cc=agraf@suse.de \
--cc=arnd@arndb.de \
--cc=fernando@oss.ntt.co.jp \
--cc=hpa@zytor.com \
--cc=kvm-ia64@vger.kernel.org \
--cc=kvm-ppc@vger.kernel.org \
--cc=kvm@vger.kernel.org \
--cc=linux-arch@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linuxppc-dev@ozlabs.org \
--cc=mingo@redhat.com \
--cc=mtosatti@redhat.com \
--cc=paulus@samba.org \
--cc=takuya.yoshikawa@gmail.com \
--cc=tglx@linutronix.de \
--cc=x86@kernel.org \
--cc=yoshikawa.takuya@oss.ntt.co.jp \
/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).