Re: Change prefetch and read strategies to use range in pg_prewarm ... and raise a question about posix_fadvise WILLNEED

From: Nazir Bilal Yavuz <byavuz81(at)gmail(dot)com>
To: Cédric Villemain <Cedric(dot)Villemain(at)abcsql(dot)com>
Cc: pgsql-hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: Change prefetch and read strategies to use range in pg_prewarm ... and raise a question about posix_fadvise WILLNEED
Date: 2024-03-15 12:12:31
Message-ID: CAN55FZ1XrWS_xr2N=2316NHxMwFaCEfJku-pi9Yvda4LWU+DQw@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Hi,

On Thu, 7 Mar 2024 at 15:26, Cédric Villemain
<Cedric(dot)Villemain(at)abcsql(dot)com> wrote:
>
> On 07/03/2024 12:19, Nazir Bilal Yavuz wrote:
> >
> > I did not test read performance but I am planning to do that soon.

I did not have the time to check other things you mentioned but I
tested the read performance. The table size is 5.5GB, I did 20 runs in
total.

When the OS cache is cleared:

Master -> Median: 2266.293ms, Avg: 2265.5038ms
Patched -> Median: 2166.493ms, Avg: 2183.6208ms

When the buffers are in the OS cache:

Master -> Median: 585.719ms, Avg: 583.5032ms
Patched -> Median: 533.071ms, Avg: 532.7119ms

Patched version is better on both. ~4% when the OS cache is cleared,
~%9 when the buffers are in the OS cache.

--
Regards,
Nazir Bilal Yavuz
Microsoft

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Jim Jones 2024-03-15 13:21:59 Re: Adding comments to help understand psql hidden queries
Previous Message Heikki Linnakangas 2024-03-15 12:10:14 Re: Weird test mixup