Path: utzoo!attcan!uunet!mcsun!ukc!dcl-cs!aber-cs!rupert!pcg From: pcg@rupert.cs.aber.ac.uk (Piercarlo Grandi) Newsgroups: comp.windows.x Subject: Re: X11R3 on a Sparcstation1 Message-ID: Date: 18 Dec 89 17:31:38 GMT References: <1989Dec12.163847.7802@ux1.cso.uiuc.edu> <8912121932.AA25369@xenon.lcs.mit.edu> Sender: pcg@aber-cs.UUCP Organization: Coleg Prifysgol Cymru Lines: 22 In-reply-to: keith@EXPO.LCS.MIT.EDU's message of 12 Dec 89 19:32:27 GMT In article <8912121932.AA25369@xenon.lcs.mit.edu> keith@EXPO.LCS.MIT.EDU (Keith Packard) writes: R3 cfbbitblt.c was particularily nasty to compilers. It expands a several-hundred line cpp macro 16 times; which breaks many optimizers (including the SPARC). My PMAX spent several hours compiling this one file, as it didn't have the requested 40 meg of physical memory. Would not it be easier to turn the several hundred line macro into a procedure, or, if there is a hot path in the macro, maybe a macro that only contains it and the rest packaged off in a function? There has been a lot of discussion in comp.lang.c/c++ on inlining, and I wonder if there was any real reason for the extensive use of very large macros in the server. Any significant speedup? R4 won't have these problems (and will be "just a bit" faster). What did you do to it? Offlined the macros, offlined them selectively, ???? -- Piercarlo "Peter" Grandi | ARPA: pcg%cs.aber.ac.uk@nsfnet-relay.ac.uk Dept of CS, UCW Aberystwyth | UUCP: ...!mcvax!ukc!aber-cs!pcg Penglais, Aberystwyth SY23 3BZ, UK | INET: pcg@cs.aber.ac.uk