From: Andrea Arcangeli <aarcange@redhat.com>
To: Nai Xia <nai.xia@gmail.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>,
dlaor@redhat.com, Ingo Molnar <mingo@kernel.org>,
Hillf Danton <dhillf@gmail.com>,
linux-kernel@vger.kernel.org, linux-mm@kvack.org,
Dan Smith <danms@us.ibm.com>,
Linus Torvalds <torvalds@linux-foundation.org>,
Andrew Morton <akpm@linux-foundation.org>,
Thomas Gleixner <tglx@linutronix.de>, Ingo Molnar <mingo@elte.hu>,
Paul Turner <pjt@google.com>,
Suresh Siddha <suresh.b.siddha@intel.com>,
Mike Galbraith <efault@gmx.de>,
"Paul E. McKenney" <paulmck@linux.vnet.ibm.com>,
Lai Jiangshan <laijs@cn.fujitsu.com>,
Bharata B Rao <bharata.rao@gmail.com>,
Lee Schermerhorn <Lee.Schermerhorn@hp.com>,
Rik van Riel <riel@redhat.com>,
Johannes Weiner <hannes@cmpxchg.org>,
Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>,
Christoph Lameter <cl@linux.com>, Alex Shi <alex.shi@intel.com>,
Mauricio Faria de Oliveira <mauricfo@linux.vnet.ibm.com>,
Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>,
Don Morris <don.morris@hp.com>,
Benjamin Herrenschmidt <benh@kernel.crashing.org>
Subject: Re: [PATCH 13/40] autonuma: CPU follow memory algorithm
Date: Sat, 30 Jun 2012 14:48:16 +0200 [thread overview]
Message-ID: <20120630124816.GZ6676@redhat.com> (raw)
In-Reply-To: <CAPQyPG7Nx1Jdq7WBBDC41iRGOMx8CdQjcWTNOWyj1fzVeuRcgw@mail.gmail.com>
On Sat, Jun 30, 2012 at 10:43:41AM +0800, Nai Xia wrote:
> Well, I think I am not convinced by your this many words. And surely
> I will NOT follow your reasoning of "Having information is always
> good than nothing". We all know that an illy biased balancing is worse
> than randomness: at least randomness means "average, fair play, ...".
The only way to get good performance like the hard bindings is to
fully converge the load into one node (or as fewer nodes as possible),
randomness won't get you very far in this case.
> With all uncertain things, I think only a comprehensive survey
> of real world workloads can tell if my concern is significant or not.
I welcome more real world tests.
I'm just not particularly concerned about your concern. The young bit
clearing during swapping would also be susceptible to your concern
just to make another example. If that would be a problem swapping
wouldn't possibly work ok either because pte_numa or pte_young works
the same way. In fact pte_young is even less reliable because the scan
frequency will be more variable so the phase effects will be even more
visible.
The VM is an heuristic, it obviously doesn't need to be perfect at all
times, what matters is the probability that it does the right thing.
> So I think my suggestion to you is: Show world some solid and sound
> real world proof that your approximation is > 90% accurate, just like
> the pioneers already did to LRU(This problem is surely different from
> LRU. ). Tons of words, will not do this.
http://www.kernel.org/pub/linux/kernel/people/andrea/autonuma/autonuma_bench-20120530.pdf
http://dl.dropbox.com/u/82832537/kvm-numa-comparison-0.png
There's more but I haven't updated them yet.
Thanks,
Andrea
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
next prev parent reply other threads:[~2012-06-30 12:49 UTC|newest]
Thread overview: 177+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-06-28 12:55 [PATCH 00/40] AutoNUMA19 Andrea Arcangeli
2012-06-28 12:55 ` [PATCH 01/40] mm: add unlikely to the mm allocation failure check Andrea Arcangeli
2012-06-29 14:10 ` Rik van Riel
2012-06-28 12:55 ` [PATCH 02/40] autonuma: make set_pmd_at always available Andrea Arcangeli
2012-06-29 14:10 ` Rik van Riel
2012-06-28 12:55 ` [PATCH 03/40] autonuma: export is_vma_temporary_stack() even if CONFIG_TRANSPARENT_HUGEPAGE=n Andrea Arcangeli
2012-06-29 14:11 ` Rik van Riel
2012-06-28 12:55 ` [PATCH 04/40] xen: document Xen is using an unused bit for the pagetables Andrea Arcangeli
2012-06-29 14:16 ` Rik van Riel
2012-07-04 23:05 ` Andrea Arcangeli
2012-06-30 4:47 ` Konrad Rzeszutek Wilk
2012-07-03 10:45 ` Andrea Arcangeli
2012-06-28 12:55 ` [PATCH 05/40] autonuma: define _PAGE_NUMA_PTE and _PAGE_NUMA_PMD Andrea Arcangeli
2012-06-28 15:13 ` Don Morris
2012-06-28 15:00 ` Andrea Arcangeli
2012-06-29 14:26 ` Rik van Riel
2012-07-03 20:30 ` Andrea Arcangeli
2012-06-28 12:55 ` [PATCH 06/40] autonuma: x86 pte_numa() and pmd_numa() Andrea Arcangeli
2012-06-29 15:02 ` Rik van Riel
2012-07-04 23:03 ` Andrea Arcangeli
2012-06-28 12:55 ` [PATCH 07/40] autonuma: generic " Andrea Arcangeli
2012-06-29 15:13 ` Rik van Riel
2012-06-28 12:55 ` [PATCH 08/40] autonuma: teach gup_fast about pte_numa Andrea Arcangeli
2012-06-29 15:27 ` Rik van Riel
2012-06-28 12:55 ` [PATCH 09/40] autonuma: introduce kthread_bind_node() Andrea Arcangeli
2012-06-29 15:36 ` Rik van Riel
2012-06-29 16:04 ` Peter Zijlstra
2012-06-29 16:11 ` Rik van Riel
2012-06-29 16:38 ` Andrea Arcangeli
2012-06-29 16:58 ` Rik van Riel
2012-07-05 13:09 ` Johannes Weiner
2012-07-05 18:33 ` Glauber Costa
2012-07-05 20:07 ` Andrea Arcangeli
2012-06-30 4:50 ` Konrad Rzeszutek Wilk
2012-07-04 23:14 ` Andrea Arcangeli
2012-07-05 12:04 ` Konrad Rzeszutek Wilk
2012-07-05 12:28 ` Andrea Arcangeli
2012-07-05 12:18 ` Peter Zijlstra
2012-07-05 12:21 ` Andrea Arcangeli
2012-06-28 12:55 ` [PATCH 10/40] autonuma: mm_autonuma and sched_autonuma data structures Andrea Arcangeli
2012-06-29 15:47 ` Rik van Riel
2012-06-29 17:45 ` Rik van Riel
2012-07-04 23:16 ` Andrea Arcangeli
2012-06-28 12:55 ` [PATCH 11/40] autonuma: define the autonuma flags Andrea Arcangeli
2012-06-29 16:10 ` Rik van Riel
2012-06-30 4:58 ` Konrad Rzeszutek Wilk
2012-07-02 15:42 ` Konrad Rzeszutek Wilk
2012-06-30 5:01 ` Konrad Rzeszutek Wilk
2012-07-04 23:45 ` Andrea Arcangeli
2012-06-28 12:55 ` [PATCH 12/40] autonuma: core autonuma.h header Andrea Arcangeli
2012-06-28 12:55 ` [PATCH 13/40] autonuma: CPU follow memory algorithm Andrea Arcangeli
2012-06-28 14:46 ` Peter Zijlstra
2012-06-29 14:11 ` Nai Xia
2012-06-29 16:30 ` Andrea Arcangeli
2012-06-29 18:09 ` Nai Xia
2012-06-29 21:02 ` Nai Xia
2012-07-03 11:53 ` Peter Zijlstra
2012-06-28 14:53 ` Peter Zijlstra
2012-06-29 12:16 ` Hillf Danton
2012-06-29 12:55 ` Ingo Molnar
2012-06-29 16:51 ` Dor Laor
2012-06-29 18:41 ` Peter Zijlstra
2012-06-29 18:46 ` Rik van Riel
2012-06-29 18:51 ` Peter Zijlstra
2012-06-29 18:57 ` Peter Zijlstra
2012-06-29 19:03 ` Peter Zijlstra
2012-06-29 19:19 ` Rik van Riel
2012-07-02 16:57 ` Vaidyanathan Srinivasan
2012-07-05 16:56 ` Vaidyanathan Srinivasan
2012-07-06 13:04 ` Hillf Danton
2012-07-06 18:38 ` Vaidyanathan Srinivasan
2012-07-12 13:12 ` Andrea Arcangeli
2012-06-29 18:49 ` Peter Zijlstra
2012-06-29 18:53 ` Peter Zijlstra
2012-06-29 20:01 ` Nai Xia
2012-06-29 20:44 ` Nai Xia
2012-06-30 1:23 ` Andrea Arcangeli
2012-06-30 2:43 ` Nai Xia
2012-06-30 5:48 ` Dor Laor
2012-06-30 6:58 ` Nai Xia
2012-06-30 13:04 ` Andrea Arcangeli
2012-06-30 15:19 ` Nai Xia
2012-06-30 19:37 ` Dor Laor
2012-07-01 2:41 ` Nai Xia
2012-06-30 23:55 ` Benjamin Herrenschmidt
2012-07-01 3:10 ` Nai Xia
2012-06-30 8:23 ` Nai Xia
2012-07-02 7:29 ` Rik van Riel
2012-07-02 7:43 ` Nai Xia
2012-06-30 12:48 ` Andrea Arcangeli [this message]
2012-06-30 15:10 ` Nai Xia
2012-07-02 7:36 ` Rik van Riel
2012-07-02 7:56 ` Nai Xia
2012-07-02 8:17 ` Rik van Riel
2012-07-02 8:31 ` Nai Xia
2012-07-05 18:07 ` Rik van Riel
2012-07-05 22:59 ` Andrea Arcangeli
2012-07-06 1:00 ` Nai Xia
2012-06-29 19:04 ` Peter Zijlstra
2012-06-29 20:27 ` Nai Xia
2012-06-29 18:03 ` Rik van Riel
2012-06-28 12:55 ` [PATCH 14/40] autonuma: add page structure fields Andrea Arcangeli
2012-06-29 18:06 ` Rik van Riel
2012-06-28 12:55 ` [PATCH 15/40] autonuma: knuma_migrated per NUMA node queues Andrea Arcangeli
2012-06-29 18:31 ` Rik van Riel
2012-06-28 12:55 ` [PATCH 16/40] autonuma: init knuma_migrated queues Andrea Arcangeli
2012-06-29 18:35 ` Rik van Riel
2012-06-28 12:55 ` [PATCH 17/40] autonuma: autonuma_enter/exit Andrea Arcangeli
2012-06-29 18:37 ` Rik van Riel
2012-06-28 12:55 ` [PATCH 18/40] autonuma: call autonuma_setup_new_exec() Andrea Arcangeli
2012-06-29 18:39 ` Rik van Riel
2012-06-30 5:04 ` Konrad Rzeszutek Wilk
2012-07-12 17:50 ` Andrea Arcangeli
2012-06-28 12:55 ` [PATCH 19/40] autonuma: alloc/free/init sched_autonuma Andrea Arcangeli
2012-06-29 18:52 ` Rik van Riel
2012-06-30 5:10 ` Konrad Rzeszutek Wilk
2012-07-12 17:59 ` Andrea Arcangeli
2012-06-28 12:56 ` [PATCH 20/40] autonuma: alloc/free/init mm_autonuma Andrea Arcangeli
2012-06-29 18:54 ` Rik van Riel
2012-06-30 5:12 ` Konrad Rzeszutek Wilk
2012-07-12 18:08 ` Andrea Arcangeli
2012-07-12 18:17 ` Johannes Weiner
2012-07-13 14:19 ` Christoph Lameter
2012-07-14 17:01 ` Andrea Arcangeli
2012-07-01 15:33 ` Rik van Riel
2012-07-12 18:27 ` Andrea Arcangeli
2012-06-28 12:56 ` [PATCH 21/40] autonuma: avoid CFS select_task_rq_fair to return -1 Andrea Arcangeli
2012-06-29 18:57 ` Rik van Riel
2012-06-29 19:05 ` Peter Zijlstra
2012-06-29 19:07 ` Rik van Riel
2012-06-29 20:48 ` Ingo Molnar
2012-06-28 12:56 ` [PATCH 22/40] autonuma: teach CFS about autonuma affinity Andrea Arcangeli
2012-07-01 16:37 ` Rik van Riel
2012-06-28 12:56 ` [PATCH 23/40] autonuma: sched_set_autonuma_need_balance Andrea Arcangeli
2012-07-01 16:57 ` Rik van Riel
2012-06-28 12:56 ` [PATCH 24/40] autonuma: core Andrea Arcangeli
2012-07-02 4:07 ` Rik van Riel
2012-06-28 12:56 ` [PATCH 25/40] autonuma: follow_page check for pte_numa/pmd_numa Andrea Arcangeli
2012-07-02 4:14 ` Rik van Riel
2012-07-14 16:43 ` Andrea Arcangeli
2012-06-28 12:56 ` [PATCH 26/40] autonuma: default mempolicy follow AutoNUMA Andrea Arcangeli
2012-07-02 4:19 ` Rik van Riel
2012-06-28 12:56 ` [PATCH 27/40] autonuma: call autonuma_split_huge_page() Andrea Arcangeli
2012-07-02 4:22 ` Rik van Riel
2012-06-28 12:56 ` [PATCH 28/40] autonuma: make khugepaged pte_numa aware Andrea Arcangeli
2012-07-02 4:24 ` Rik van Riel
2012-07-12 18:50 ` Andrea Arcangeli
2012-07-12 21:25 ` Rik van Riel
2012-06-28 12:56 ` [PATCH 29/40] autonuma: retain page last_nid information in khugepaged Andrea Arcangeli
2012-07-02 4:33 ` Rik van Riel
2012-06-28 12:56 ` [PATCH 30/40] autonuma: numa hinting page faults entry points Andrea Arcangeli
2012-07-02 4:47 ` Rik van Riel
2012-06-28 12:56 ` [PATCH 31/40] autonuma: reset autonuma page data when pages are freed Andrea Arcangeli
2012-07-02 4:49 ` Rik van Riel
2012-06-28 12:56 ` [PATCH 32/40] autonuma: initialize page structure fields Andrea Arcangeli
2012-07-02 4:50 ` Rik van Riel
2012-06-28 12:56 ` [PATCH 33/40] autonuma: link mm/autonuma.o and kernel/sched/numa.o Andrea Arcangeli
2012-07-02 4:56 ` Rik van Riel
2012-06-28 12:56 ` [PATCH 34/40] autonuma: add CONFIG_AUTONUMA and CONFIG_AUTONUMA_DEFAULT_ENABLED Andrea Arcangeli
2012-07-02 4:58 ` Rik van Riel
2012-06-28 12:56 ` [PATCH 35/40] autonuma: boost khugepaged scanning rate Andrea Arcangeli
2012-07-02 5:12 ` Rik van Riel
2012-06-28 12:56 ` [PATCH 36/40] autonuma: page_autonuma Andrea Arcangeli
2012-06-30 5:24 ` Konrad Rzeszutek Wilk
2012-07-12 19:43 ` Andrea Arcangeli
2012-07-02 6:37 ` Rik van Riel
2012-07-12 19:58 ` Andrea Arcangeli
2012-06-28 12:56 ` [PATCH 37/40] autonuma: page_autonuma change #include for sparse Andrea Arcangeli
2012-07-02 6:22 ` Rik van Riel
2012-06-28 12:56 ` [PATCH 38/40] autonuma: autonuma_migrate_head[0] dynamic size Andrea Arcangeli
2012-07-02 5:15 ` Rik van Riel
2012-06-28 12:56 ` [PATCH 39/40] autonuma: bugcheck page_autonuma fields on newly allocated pages Andrea Arcangeli
2012-07-02 6:40 ` Rik van Riel
2012-06-28 12:56 ` [PATCH 40/40] autonuma: shrink the per-page page_autonuma struct size Andrea Arcangeli
2012-07-02 7:18 ` Rik van Riel
2012-07-12 20:21 ` Andrea Arcangeli
2012-07-09 15:40 ` [PATCH 00/40] AutoNUMA19 Rik van Riel
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=20120630124816.GZ6676@redhat.com \
--to=aarcange@redhat.com \
--cc=Lee.Schermerhorn@hp.com \
--cc=a.p.zijlstra@chello.nl \
--cc=akpm@linux-foundation.org \
--cc=alex.shi@intel.com \
--cc=benh@kernel.crashing.org \
--cc=bharata.rao@gmail.com \
--cc=cl@linux.com \
--cc=danms@us.ibm.com \
--cc=dhillf@gmail.com \
--cc=dlaor@redhat.com \
--cc=don.morris@hp.com \
--cc=efault@gmx.de \
--cc=hannes@cmpxchg.org \
--cc=konrad.wilk@oracle.com \
--cc=laijs@cn.fujitsu.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=mauricfo@linux.vnet.ibm.com \
--cc=mingo@elte.hu \
--cc=mingo@kernel.org \
--cc=nai.xia@gmail.com \
--cc=paulmck@linux.vnet.ibm.com \
--cc=pjt@google.com \
--cc=riel@redhat.com \
--cc=suresh.b.siddha@intel.com \
--cc=tglx@linutronix.de \
--cc=torvalds@linux-foundation.org \
--cc=vatsa@linux.vnet.ibm.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;
as well as URLs for NNTP newsgroup(s).