X-pop3-spooler: POP3MAIL 2.1.0 b 3 961213 -bs- Delivered-To: pcg AT goof DOT com Message-ID: <19980127171936.22692@cerebro.laendle> Date: Tue, 27 Jan 1998 17:19:36 +0100 From: Marc Lehmann To: Falk Hueffner Cc: beastium Subject: Re: pgcc for non-ix86-Architectures References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.88 In-Reply-To: ; from Falk Hueffner on Tue, Jan 27, 1998 at 11:48:28AM +0100 X-Operating-System: Linux version 2.1.81 (root AT cerebro) (gcc version pgcc-2.91.05 980122 (gcc-2.8.0 release)) Status: RO Content-Length: 1508 Lines: 34 On Tue, Jan 27, 1998 at 11:48:28AM +0100, Falk Hueffner wrote: > hi, > > is there any point to use pgcc instead ofegcs on architectures other > than ix86, e. g. hppa? Are there any optimizations that will benefit > for all architectures? yes, many of them.. but not many people have checked this out. I have reports that pgcc sometimes speeds up calculations on alphas (compared to egcs), but I still have to investigate further. The first problem is to actually _build_ pgcc, and some bugs already have been catched. Then there might be problems with compiling programs using pgcc, since some of the optimizations might make assumptions about the rtl code (which is architecture dependent). if all this works (please try it out!), then we have to look after _which_ optimizations are beneficial and which are not (say, because they are too pentium-centric). On non-x86 architectures, all pgcc-specific optimizations (that can be turned off) are turned off by default, with any optimization level, so they have to be enabled by hand (using -fxxx switches, see the faq). -----==- | ----==-- _ | ---==---(_)__ __ ____ __ Marc Lehmann +-- --==---/ / _ \/ // /\ \/ / pcg AT goof DOT com |e| -=====/_/_//_/\_,_/ /_/\_\ --+ The choice of a GNU generation | |