Date: Thu, 20 Sep 2018 13:34:27 -0400
From: Stephen Frost <sfrost@snowman.net>
To: Andres Freund <andres@anarazel.de>
Cc: Tom Lane <tgl@sss.pgh.pa.us>, Amit Khandekar <amitdkhan.pg@gmail.com>,
	Andreas Joseph Krogh <andreas@visena.com>,
	Robert Haas <robertmhaas@gmail.com>,
	Alvaro Herrera <alvherre@2ndquadrant.com>,
	"Jonathan S. Katz" <jkatz@postgresql.org>,
	PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Subject: Re: Query is over 2x slower with jit=on
Message-ID: <20180920173426.GS4184@tamriel.snowman.net>
References: 
 <CAJ3gD9cv2cbJ2t5zRZiWapNeRdgpM70++f80NbLwbEpM7uX2AQ@mail.gmail.com>
 <20180910160909.qf3h3ttztvbyyuds@alap3.anarazel.de>
 <CAJ3gD9fJcYofmWCe9hJ9Y6dUYU-qpkR=Bii+NhYufE7Mny9AHw@mail.gmail.com>
 <CAJ3gD9cWBKnzXkhu30njaY1au3NO0XD=r4N3GcC_nwFbw5GYYw@mail.gmail.com>
 <27812.1537221015@sss.pgh.pa.us>
 <20180919224047.mcdjwis56ywqzhzs@alap3.anarazel.de>
 <11291.1537414012@sss.pgh.pa.us>
 <20180920033922.zi6n6vk7ilxaseu3@alap3.anarazel.de>
 <20180920130721.GM4184@tamriel.snowman.net>
 <20180920172701.dtiwldlgkkhpvykz@alap3.anarazel.de>
MIME-Version: 1.0
Content-Type: multipart/signed; micalg=pgp-sha512;
	protocol="application/pgp-signature"; boundary="yckM7CAr5H2WMv0u"
Content-Disposition: inline
In-Reply-To: <20180920172701.dtiwldlgkkhpvykz@alap3.anarazel.de>
User-Agent: Mutt/1.5.24 (2015-08-30)
Precedence: bulk


--yckM7CAr5H2WMv0u
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

Greetings,

* Andres Freund (andres@anarazel.de) wrote:
> On 2018-09-20 09:07:21 -0400, Stephen Frost wrote:
> > * Andres Freund (andres@anarazel.de) wrote:
> > > FWIW, not that I want to do that now, but at some point it might make
> > > sense to sub-divide this into things like number of "expressions",
> > > "tuple deforming", "plans", ...  Just mentioning that if somebody wan=
ts
> > > to comment on reformatting this as well, if we're tinkering anyway.
> >=20
> > I'd actually think we'd maybe want some kind of 'verbose' mode which
> > shows exactly what got JIT'd and what didn't- one of the questions that
> > I think people will be asking is "why didn't X get JIT'd?" and I don't
> > think that's very easy to figure out currently.
>=20
> That seems largely a separate discussion / feature though, right?  I'm
> not entirely clear what precisely you mean with "why didn't X get
> JIT'd?" - currently that's a whole query decision.

There are things that can't be JIT'd currently and, I'm thinking anyway,
we'd JIT what we are able to and report out what was JIT'd in some
fashion which would inform a user what was able to be JIT'd and what
wasn't.  I'll admit, perhaps this is more of a debugging tool than
something that a end-user would find useful, unless there's possibly
some way that a user could effect what happens.

> > > I'm pretty certain you're right :).  There's already arguments around
> > > making optimization more gradual (akin to O1,2,3).
> >=20
> > That certainly sounds like it'd be very neat to have though I wonder how
> > well we'll be able to automatically plan out which optimization level to
> > use when..
>=20
> Well, that's not really different from having to decide whether to use
> JIT or not.  I suspect that once / if we get caching and/or background
> JIT compilation, we can get a lot more creative around this.

yeah, I definitely understand that, cacheing, in particular, I would
expect to play a very large role.

Thanks!

Stephen

--yckM7CAr5H2WMv0u
Content-Type: application/pgp-signature; name="signature.asc"

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
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=Cg8W
-----END PGP SIGNATURE-----

--yckM7CAr5H2WMv0u--