Path: chuka.playstation.co.uk!news From: Andrew Partington Newsgroups: scee.yaroze.freetalk.english Subject: Re: CW code generation differences on Intel/AMD?? Date: Thu, 22 Nov 2001 14:14:52 +0000 Organization: PlayStation Net Yaroze (SCEE) Lines: 92 Message-ID: <3BFD085C.513FFA0E@peregrine.com> References: <3BFA9BE1.810576C8@peregrine.com> <9tgdln$4q118@www.netyaroze-europe.com> <9tgo8m$4q119@www.netyaroze-europe.com> <3BFCDDE9.AC885249@peregrine.com> <9tits4$83m3@www.netyaroze-europe.com> NNTP-Posting-Host: 194.207.193.51 Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 4.7 [en] (WinNT; I) X-Accept-Language: en Aha - I didn't have that patch on the HD I salvaged from my old machine, I just installed the one available on the NY website. I remembered that there were two patches but I couldn't remember where the other one was, I remember installing it on my P120 to see if it fixed the assembler in CW (which it doesn't BTW, but increasing my block sizes to 64x64 from 32x32 in my background handler gave me a massive speedup, (over 200 HSyncs down to around 140 in the worst case with L4/L3 optimisation turned on, and the code sprinkled with FntPrints) so I thought that it probably wasn't worth using this far into GG). I probably never bothered to update it on my old desktop machine, since most of my code was being done on the P120 - that might explain a few things! Guess I can get round a few of the debugger problems with trusty old FntPrint/printf :) Thanks a lot Peter! Andy Peter Armstrong wrote: > > Which CW patch did you try, there are two for the Release 2 disc. The first > updates a couple of compiler and linker dll's, the second updates the IDE > and PSComUtil. > > I've just added the first patch to CW and now the program I was trying is > exiting fine at optimization level 4. > > In case you haven't got the patch it's available at > ftp://ftp.metrowerks.com/pub/updates/CWPSNY2/ny_update.exe > > I don't use this patch myself as it causes problems seeing local variables > in the debugger. > > HTH, > Peter > > "Andrew Partington" wrote in message > news:3BFCDDE9.AC885249@peregrine.com... > > Hi all > > > > I tried the patch and it doesn't fix it... :( > > > > I backed up the project from the Intel machine on Sunday complete with > > .pxe that was build on it, so I was able to do a couple of comparisons > > (i'll have to wait until the weekend until I build it again on that > > machine though) > > > > When I extracted the source etc, and just ran the .pxe without > > recompiling it on the AMD, I could exit fine in most cases (sometimes > > theres a bug where it WILL freeze, but I think thats to do with me being > > lazy and not clamping my map scroll values, it happened occasionally > > with previous versions if you played it long enough). The .pxe was > > 0x9000 bytes in length according to siocons, BUT the scanner was not > > being displayed for some reason... > > > > When I recompiled it with the AMD (same settings as on the other > > machine), the length of the .pxe was longer, and it locked up on exit as > > I suspected it would, but the scanner was working fine! No source code > > changes were made. > > > > I tried completely removing the scanner code (assuming it was the > > culprit somehow) - no joy. > > > > None of this is really a problem, I can just drop the optimisation level > > down to 2 and exit OK, its just strange that the code seems to be being > > built differently... unless its my code getting a bit wobbly! > > > > Would it affect the chances of GG appearing on the OPSM coverdisk if it > > didn't allow you to exit back to the coverdisk's menu? [as in lacking a > > menu option to let you exit, rather than just crash people's > > PlayStations for them :)] In which case it IS a problem and i'll be > > sticking to the P120! > > > > Cheers guys > > > > > > Andy > > > > > > Chris Wallace wrote: > > > > > > When I was using Siocons to run CW pxe files it went back to the CIP > screen > > > fine, but never through PScomUtil. > > > However, I can't use siocons at all with WinXP. (or with ME for that > matter) > > > So I couldnt compare the Athlon with the Pentium anyway.. =\