mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: John Hubbard <jhubbard@nvidia.com>
To: Ryan Roberts <ryan.roberts@arm.com>,
	Matthew Wilcox <willy@infradead.org>
Cc: Andrew Morton <akpm@linux-foundation.org>,
	Yin Fengwei <fengwei.yin@intel.com>,
	David Hildenbrand <david@redhat.com>, Yu Zhao <yuzhao@google.com>,
	Catalin Marinas <catalin.marinas@arm.com>,
	Anshuman Khandual <anshuman.khandual@arm.com>,
	Yang Shi <shy828301@gmail.com>,
	"Huang, Ying" <ying.huang@intel.com>, Zi Yan <ziy@nvidia.com>,
	Luis Chamberlain <mcgrof@kernel.org>,
	Itaru Kitayama <itaru.kitayama@gmail.com>,
	"Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>,
	David Rientjes <rientjes@google.com>,
	Vlastimil Babka <vbabka@suse.cz>, Hugh Dickins <hughd@google.com>,
	linux-mm@kvack.org, linux-kernel@vger.kernel.org,
	linux-arm-kernel@lists.infradead.org
Subject: Re: [PATCH v6 0/9] variable-order, large folios for anonymous memory
Date: Mon, 13 Nov 2023 09:52:29 -0500	[thread overview]
Message-ID: <abb85115-5790-c292-f27e-3c13b105230d@nvidia.com> (raw)
In-Reply-To: <f1fa098b-210e-41a9-80fc-aec212976610@arm.com>

On 11/13/23 2:19 AM, Ryan Roberts wrote:
> On 13/11/2023 05:18, Matthew Wilcox wrote:
>> On Sun, Nov 12, 2023 at 10:57:47PM -0500, John Hubbard wrote:
>>> I've done some initial performance testing of this patchset on an arm64
>>> SBSA server. When these patches are combined with the arm64 arch contpte
>>> patches in Ryan's git tree (he has conveniently combined everything
>>> here: [1]), we are seeing a remarkable, consistent speedup of 10.5x on
>>> some memory-intensive workloads. Many test runs, conducted independently
>>> by different engineers and on different machines, have convinced me and
>>> my colleagues that this is an accurate result.
>>>
>>> In order to achieve that result, we used the git tree in [1] with
>>> following settings:
>>>
>>>      echo always >/sys/kernel/mm/transparent_hugepage/enabled
>>>      echo recommend >/sys/kernel/mm/transparent_hugepage/anon_orders
>>>
>>> This was on a aarch64 machine configure to use a 64KB base page size.
>>> That configuration means that the PMD size is 512MB, which is of course
>>> too large for practical use as a pure PMD-THP. However, with with these
>>> small-size (less than PMD-sized) THPs, we get the improvements in TLB
>>> coverage, while still getting pages that are small enough to be
>>> effectively usable.
>>
>> That is quite remarkable!
> 
> Yes, agreed - thanks for sharing these results! A very nice Monday morning boost!
> 
>>
>> My hope is to abolish the 64kB page size configuration.  ie instead of

We've found that a 64KB base page size provides better performance for
HPC and AI workloads, than a 4KB base size, at least for these kinds of
servers. In fact, the 4KB config is considered odd and I'd have to
look around to get one. It's mostly a TLB coverage issue because,
again, the problem typically has a very large memory footprint.

So even though it would be nice from a software point of view, there's
a real need for this.

>> using the mixture of page sizes that you currently are -- 64k and
>> 1M (right?  Order-0, and order-4)
> 
> Not quite; the contpte-size for a 64K page size is 2M/order-5. (and yes, it is
> 64K/order-4 for a 4K page size, and 2M/order-7 for a 16K page size. I agree that
> intuitively you would expect the order to remain constant, but it doesn't).
> 
> The "recommend" setting above will actually enable order-3 as well even though
> there is no HW benefit to this. So the full set of available memory sizes here is:
> 
> 64K/order-0, 512K/order-3, 2M/order-5, 512M/order-13

Yes, and to provide some further details about the test runs, I went
so far as to test individual anon_orders (for example, 
anon_orders=0x20), in order to isolate behavior and see what's really
going on.

On this hardware, anything with 2MB page sizes which corresponds to
anon_orders=0x20, as I recall) or larger, gets the 10x boost. It's
an interesting on/off behavior. This particular server design and
workload combination really prefers 2MB pages, even if they are
held together with contpte instead of a real PMD entry.

> 
>> , that 4k, 64k and 2MB (order-0,
>> order-4 and order-9) will provide better performance.
>>
>> Have you run any experiements with a 4kB page size?
> 
> Agree that would be interesting with 64K small-sized THP enabled. And I'd love
> to get to a world were we universally deal in variable sized chunks of memory,
> aligned on 4K boundaries.
> 
> In my experience though, there are still some performance benefits to 64K base
> page vs 4K+contpte; the page tables are more cache efficient for the former case
> - 64K of memory is described by 8 bytes in the former vs 8x16=128 bytes in the
> latter. In practice the HW will still only read 8 bytes in the latter but that's
> taking up a full cache line vs the former where a single cache line stores 8x
> 64K entries. >
> Thanks,
> Ryan
> 

thanks,

-- 
John Hubbard
NVIDIA

  parent reply	other threads:[~2023-11-13 14:52 UTC|newest]

Thread overview: 67+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-09-29 11:44 Ryan Roberts
2023-09-29 11:44 ` [PATCH v6 1/9] mm: Allow deferred splitting of arbitrary anon large folios Ryan Roberts
2023-10-05  8:19   ` David Hildenbrand
2023-09-29 11:44 ` [PATCH v6 2/9] mm: Non-pmd-mappable, large folios for folio_add_new_anon_rmap() Ryan Roberts
2023-09-29 13:45   ` Kirill A. Shutemov
2023-09-29 14:39     ` Ryan Roberts
2023-09-29 11:44 ` [PATCH v6 3/9] mm: thp: Account pte-mapped anonymous THP usage Ryan Roberts
2023-09-29 11:44 ` [PATCH v6 4/9] mm: thp: Introduce anon_orders and anon_always_mask sysfs files Ryan Roberts
2023-09-29 22:55   ` Andrew Morton
2023-10-02 10:15     ` Ryan Roberts
2023-10-07 22:54     ` Michael Ellerman
2023-10-10  0:20       ` Andrew Morton
2023-10-12  9:31         ` David Hildenbrand
2023-10-12 11:07         ` Michael Ellerman
2023-10-11  6:02   ` kernel test robot
2023-09-29 11:44 ` [PATCH v6 5/9] mm: thp: Extend THP to allocate anonymous large folios Ryan Roberts
     [not found]   ` <CGME20231005120507eucas1p13f50fa99f52808818840ee7db194e12e@eucas1p1.samsung.com>
2023-10-05 12:05     ` Daniel Gomez
2023-10-05 12:49       ` Ryan Roberts
2023-10-05 14:59         ` Daniel Gomez
2023-10-27 23:04   ` John Hubbard
2023-10-30 11:43     ` Ryan Roberts
2023-10-30 23:25       ` John Hubbard
2023-11-01 13:56         ` Ryan Roberts
2023-09-29 11:44 ` [PATCH v6 6/9] mm: thp: Add "recommend" option for anon_orders Ryan Roberts
2023-10-06 20:08   ` David Hildenbrand
2023-10-06 22:28     ` Yu Zhao
2023-10-09 11:45       ` Ryan Roberts
2023-10-09 14:43         ` David Hildenbrand
2023-10-09 20:04         ` Yu Zhao
2023-10-10 10:16           ` Ryan Roberts
2023-09-29 11:44 ` [PATCH v6 7/9] arm64/mm: Override arch_wants_pte_order() Ryan Roberts
2023-10-02 15:21   ` Catalin Marinas
2023-10-03  7:32     ` Ryan Roberts
2023-10-03 12:05       ` Catalin Marinas
2023-09-29 11:44 ` [PATCH v6 8/9] selftests/mm/cow: Generalize do_run_with_thp() helper Ryan Roberts
2023-09-29 11:44 ` [PATCH v6 9/9] selftests/mm/cow: Add tests for small-order anon THP Ryan Roberts
2023-10-06 20:06 ` [PATCH v6 0/9] variable-order, large folios for anonymous memory David Hildenbrand
2023-10-09 11:28   ` Ryan Roberts
2023-10-09 16:22     ` David Hildenbrand
2023-10-10 10:47       ` Ryan Roberts
2023-10-13 20:14         ` David Hildenbrand
2023-10-20 12:33   ` Ryan Roberts
2023-10-25 16:24     ` Ryan Roberts
2023-10-25 18:47       ` David Hildenbrand
2023-10-25 19:11         ` Yu Zhao
2023-10-26  9:53           ` Ryan Roberts
2023-10-26 15:19             ` David Hildenbrand
2023-10-25 19:10       ` John Hubbard
2023-10-31 11:50   ` Ryan Roberts
2023-10-31 11:55     ` Ryan Roberts
2023-10-31 12:03       ` David Hildenbrand
2023-10-31 13:13         ` Ryan Roberts
2023-10-31 18:29       ` Yang Shi
2023-11-01 14:02         ` Ryan Roberts
2023-11-01 18:11           ` Yang Shi
2023-10-31 11:58     ` David Hildenbrand
2023-10-31 13:12       ` Ryan Roberts
2023-11-13  3:57 ` John Hubbard
2023-11-13  5:18   ` Matthew Wilcox
2023-11-13 10:19     ` Ryan Roberts
2023-11-13 11:52       ` Kefeng Wang
2023-11-13 12:12         ` Ryan Roberts
2023-11-13 14:52           ` Kefeng Wang
2023-11-13 14:52       ` John Hubbard [this message]
2023-11-13 15:04       ` Matthew Wilcox
2023-11-14 10:57         ` Ryan Roberts
2023-12-05 16:05           ` Matthew Wilcox

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=abb85115-5790-c292-f27e-3c13b105230d@nvidia.com \
    --to=jhubbard@nvidia.com \
    --cc=akpm@linux-foundation.org \
    --cc=anshuman.khandual@arm.com \
    --cc=catalin.marinas@arm.com \
    --cc=david@redhat.com \
    --cc=fengwei.yin@intel.com \
    --cc=hughd@google.com \
    --cc=itaru.kitayama@gmail.com \
    --cc=kirill.shutemov@linux.intel.com \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=mcgrof@kernel.org \
    --cc=rientjes@google.com \
    --cc=ryan.roberts@arm.com \
    --cc=shy828301@gmail.com \
    --cc=vbabka@suse.cz \
    --cc=willy@infradead.org \
    --cc=ying.huang@intel.com \
    --cc=yuzhao@google.com \
    --cc=ziy@nvidia.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

all inboxes | Powered by JetHome®