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]

Re: System call comparison

From: Mike Stump <mrs at apple dot com>
To: Neil Booth <neil at daikokuya dot co dot uk>
Cc: gcc-patches at gcc dot gnu dot org, Nicola Pero <nicola at brainstorm dot co dot uk>, Geoffrey Keating <geoffk at apple dot com>, gcc at gcc dot gnu dot org, Stan Shebs <shebs at apple dot com>, Zack Weinberg <zack at codesourcery dot com>, David Ayers <d dot ayers at inode dot at>
Date: Mon, 4 Aug 2003 12:50:53 -0700
Subject: Re: System call comparison

On Sunday, August 3, 2003, at 02:25 PM, Neil Booth wrote:

Here are the results for compiling current toplev.c:
		Before		After
Total ex PCH	1416		973		(-31%)
Total ex PCH	1347		888		(-34%)

Looks nice, thanks. Next thing to think about would be 100 search dirs, 10 files apiece, 10,000 #includes. While these numbers aren't very realistic, if you measure the speed improvements on more normal things, I suspect they will win. In code around here, we see a 30-50% compile time speed increase with this change. :-(

Essentially the change is to readdir once for each -I directory and populate a cache that can rule out trying certain include paths.

We should dust it off and submit it... If you want to work on it, play with our patch, just ask and we can send you the code we have... I'm working on the framework code now, which is more important, as it provides the semantics, which trump speed (for now).

Follow-Ups:
- Re: System call comparison
  - From: Neil Booth

References:
- System call comparison
  - From: Neil Booth

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