Re: Why JIT speed improvement is so modest?

From: Merlin Moncure <mmoncure(at)gmail(dot)com>
To: Konstantin Knizhnik <k(dot)knizhnik(at)postgrespro(dot)ru>
Cc: PostgreSQL Hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: Why JIT speed improvement is so modest?
Date: 2019-11-25 15:24:29
Message-ID: CAHyXU0xdckw6JkV9B-YrOh33KGFLrQwD_FVSEXFvZ7st-M59ZA@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Mon, Nov 25, 2019 at 9:09 AM Konstantin Knizhnik
<k(dot)knizhnik(at)postgrespro(dot)ru> wrote:
> JIT was not able to significantly (times) increase speed on Q1 query?
> Experiment with VOPS shows that used aggregation algorithm itself is not
> a bottleneck.
> Profile also give no answer for this question.
> Any ideas?

Well, in the VOPS variant around 2/3 of the time is spent in routines
that are obviously aggregation. In the JIT version, it's around 20%.
So this suggests that the replacement execution engine is more
invasive. I would also guess (!) that the VOPS engine optimizes fewer
classes of query plan. ExecScan for example, looks to be completely
optimized out VOPS but is still utilized in the JIT engine.

I experimented with Vitessa a couple of years back and this was
consistent with my recollection.

merlin

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Masahiko Sawada 2019-11-25 16:12:17 Re: [HACKERS] Block level parallel vacuum
Previous Message Konstantin Knizhnik 2019-11-25 15:09:29 Why JIT speed improvement is so modest?