From: Andreas Seltenreich <seltenreich@gmx.de>
To: Andres Freund <andres@anarazel.de>
Cc: pgsql-hackers@postgresql.org
Subject: Re: Improving executor performance
In-Reply-To: <20160714181710.sh4nt3ylzp3f55wy@alap3.anarazel.de> (Andres
	Freund's message of "Thu, 14 Jul 2016 11:17:10 -0700")
References: <20160624232953.beub22r6yqux4gcp@alap3.anarazel.de>
	<20160714011850.bd5zhu35szle3n3c@alap3.anarazel.de>
	<878tx42j1o.fsf@credativ.de>
	<20160714181710.sh4nt3ylzp3f55wy@alap3.anarazel.de>
User-Agent: Gnus/5.130012 (Ma Gnus v0.12) Emacs/24.4 (gnu/linux)
Message-ID: <87r3aw0w6p.fsf@credativ.de>
Date: Thu, 14 Jul 2016 23:03:10 +0200
MIME-Version: 1.0
Content-Type: text/plain
Precedence: bulk
Sender: pgsql-hackers-owner@postgresql.org

Andres Freund writes:

> The problem is that the previous form has a lot of ad-hoc analysis
> strewn in. The interesting part is getting rid of all that. That's what
> the new ExecInitExpr2() does. The target form can be both evaluated more
> efficiently in the dispatch manner in the patch, and quite simply
> converted to a JIT - without duplicating the analysis code.  I did write
> a small ad-hoc x86 jit, and it was really quite straightforward this
> way.

Ja, I see the advantage when doing ad-hoc-JIT compilation.

> What did you do with JIT and expression evaluation? You basically just
> replaced the toplevel ExprState note with a different evalfunc, pointing
> into your code?

That's the plan, yes.  I'm sorry there's no publishable code yet on the
the postgres side of things.  Using libFirm[1], the plan is to.

1. Automatically generate Firm-IR for the static C code around
   expression evaluation as well operators in the system catalog.

2. Construct IR for expression trees (essentially all the function calls
   the executor would do).

3. Push libFirm's optimize button.  At this stage, most of the
   dispatching goes away by inlining the calls including functions from
   the catalog implementing operators.

4. Generate code and replace the toplevel expression node with a funcall
   node.

I did implement this recipe with a toy Forth interpreter to see whether
libFirm was up to the job (Nobody's done JIT with libFirm before).  The
results were favorable[2].  Currently, a student at credativ is working
on applying these techniques to postgres.

regards,
Andreas

Footnotes: 
[1]  http://libfirm.org/

[2]  https://korte.credativ.com/~ase/firm-postgres-jit-forth.pdf


-- 
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers