This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]
Other format:	[Raw text]

[fortran, patch] LEADZ and TRAILZ for integer(kind=16)

From: FX <fxcoudert at gmail dot com>
To: Fortran List <fortran at gcc dot gnu dot org>
Cc: gcc-patches at gcc dot gnu dot org
Date: Tue, 7 Sep 2010 10:16:07 +0200
Subject: [fortran, patch] LEADZ and TRAILZ for integer(kind=16)

Attached patch improves the code generated for the LEADZ and TRAILZ intrinsics. For integer(kind=16), which is larger than C long long type, no builtin is present in the middle-end. Currently, we called library clz128/ctz128 functions in libgfortran.

The patch allows to directly generate code inline, which does basic arithmetic and call the middle-end long long builtins. This leads to better and faster code.

The question I have is: what should be done with the libgfortran functions? I think, for backward compatibility, we need to keep them, but I may be wrong.


Bootstrapped and regtested on x86_64-linux. OK to commit? (potentially with libgfortran functions removal, if deemed suitable)

FX

Attachment: clz_ctz.ChangeLog
Description: Binary data

Attachment: clz_ctz.diff
Description: Binary data

Follow-Ups:
- Re: [fortran, patch] LEADZ and TRAILZ for integer(kind=16)
  - From: Tobias Burnus
- Re: [fortran, patch] LEADZ and TRAILZ for integer(kind=16)
  - From: Thomas Koenig

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]