Horton

Re: WIP: expression evaluation improvements

Andres Freund <andres@anarazel.de>

View thread View on postgr.es
From: Andres Freund <andres@anarazel.de>
To: Andreas Karlsson <andreas@proxel.se>
Cc: pgsql-hackers@postgresql.org
Date: 2019-10-24T22:53:59Z
Lists: pgsql-hackers 
Hi,

On 2019-10-25 00:43:37 +0200, Andreas Karlsson wrote:
> On 10/23/19 6:38 PM, Andres Freund wrote:
> > In the attached *prototype* patch series there's a lot of incremental
> > improvements (and some cleanups) (in time, not importance order):
> 
> You may already know this but your patch set seems to require clang 9.

I didn't, so thanks!


> I get the below compilation error which is probably cause by
> https://github.com/llvm/llvm-project/commit/90868bb0584f first being
> committed for clang 9 (I run "clang version 7.0.1-8
> (tags/RELEASE_701/final)").
> 
> In file included from gistutil.c:24:
> ../../../../src/include/utils/float.h:103:7: error: invalid output
> constraint '=@ccae' in asm
>                          : "=@ccae"(ret), [clobber_reg]"=&x"(clobber_reg)
>                            ^
> 1 error generated.

I'll probably just drop this patch for now, it's not directly related. I
kind of wanted it on the list, so I have I place I can find it if I
forget :).

I think what really needs to happen instead is to improve the code
generated for __builtin_isinf[_sign]() by gcc/clang. They should proce
the constants like I did, instead of loading from the constant pool
every single time. That adds a fair bit of latency...

Greetings,

Andres Freund

Commits

  1. Add special case fast-paths for strict functions

  2. Replace EEOP_DONE with special steps for return/no return

  3. jit: Reference expression step functions via llvmjit_types.

  4. jit: Remove redundancies in expression evaluation code generation.

  5. expression eval: Don't redundantly keep track of AggState.

  6. jit: Reference functions by name in IOCOERCE steps.

  7. expression eval, jit: Minor code cleanups.