RE: [PATCH v7 2/2] mm: support large folios swap-in for sync io devices

["Sridhar, Kanchana P" <kanchana.p.sridhar@intel.com>]

Hi Barry,

> -----Original Message-----
> From: Barry Song <21cnbao@gmail.com>
> Sent: Wednesday, August 28, 2024 7:25 PM
> To: Sridhar, Kanchana P <kanchana.p.sridhar@intel.com>
> Cc: akpm@linux-foundation.org; baolin.wang@linux.alibaba.com;
> chrisl@kernel.org; david@redhat.com; hanchuanhua@oppo.com;
> hannes@cmpxchg.org; hch@infradead.org; hughd@google.com;
> kaleshsingh@google.com; kasong@tencent.com; linux-
> kernel@vger.kernel.org; linux-mm@kvack.org; mhocko@suse.com;
> minchan@kernel.org; nphamcs@gmail.com; ryan.roberts@arm.com;
> ryncsn@gmail.com; senozhatsky@chromium.org; shakeel.butt@linux.dev;
> shy828301@gmail.com; surenb@google.com; v-songbaohua@oppo.com;
> willy@infradead.org; xiang@kernel.org; Huang, Ying
> <ying.huang@intel.com>; yosryahmed@google.com;
> zhengtangquan@oppo.com; Feghali, Wajdi K <wajdi.k.feghali@intel.com>;
> Gopal, Vinodh <vinodh.gopal@intel.com>
> Subject: Re: [PATCH v7 2/2] mm: support large folios swap-in for sync io
> devices
> 
> On Thu, Aug 29, 2024 at 1:01 PM Kanchana P Sridhar
> <kanchana.p.sridhar@intel.com> wrote:
> >
> > Hi Shakeel,
> >
> > We submitted an RFC patchset [1] with the Intel In-Memory Analytics
> > Accelerator (Intel IAA) sometime back. This introduces a new 'canned-by_n'
> > compression algorithm in the IAA crypto driver.
> >
> > Relative to software compressors, we could get a 10X improvement in zram
> > write latency and 7X improvement in zram read latency.
> >
> > [1]
> https://lore.kernel.org/all/cover.1714581792.git.andre.glover@linux.intel.co
> m/
> 
> Hi Kanchana,
> Thanks for sharing. I understand you’ll need this mTHP swap-in series
> to leverage your
> IAA for parallel decompression, right? Without mTHP swap-in, you won't
> get this 7X
> improvement, right?

Yes, that is correct.

> 
> This is another important use case for the mTHP swap-in series,
> highlighting the strong
> need to start the work from the sync IO device.

Sure, this makes sense!

> 
> I’ll try to find some time to review your patch and explore how we can
> better support both
> software and hardware improvements in zsmalloc/zram with a more
> compatible approach.
> Also, I have a talk[1] at LPC2024—would you mind if I include a
> description of your use
> case?

Sure, this sounds good.

Thanks,
Kanchana

> 
> [1] https://lpc.events/event/18/contributions/1780/
> 
> >
> > Thanks,
> > Kanchana
> 
> Thanks
> Barry