Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!usc!snorkelwacker.mit.edu!bloom-beacon!eru!hagbard!sunic!sics.se!ifi!nuug!ugle.unit.no!sigyn.idt.unit.no!erlingh
From: erlingh@idt.unit.no (Erling Henanger)
Newsgroups: comp.sys.atari.st.tech
Subject: GNU-C and TT
Keywords: -m68881
Message-ID: <1991Apr9.190429.6482@ugle.unit.no>
Date: 9 Apr 91 19:04:29 GMT
Sender: news@ugle.unit.no
Reply-To: erlingh@idt.unit.no
Organization: IDT -- Norwegian Institute of Technology
Lines: 15

Hi there !
I've finally got a _real_ TT030, and have just recompiled the entire mint-library (patchlevel 8)
with -m68020 and -m68881 flags. The -m68020 flag did really work, and things does go along a bit
faster, but I'm having real trouble with the math-coprosessor. If i write floating-point code and compile with the -m68881 option the program tends to bomb 2 out of 3 times. It bombs every time if I have the processor-cache on. The code does execute faster, so I guess the 68882 is being used. 
Then I saw there was a file on atari.archive named gasfp.zoo, is this some special version of gas
that is needed for the co-processor to work ? or am I missing something else ??
The most obvious thing would be that the libraries only support 68881 as peripheral I guess, as in the megas, anybody know this for sure ?

Oh, and there's no way that I could make emacs.ttp work anymore :-(

-- 
       _______    _____  	 o	     ____       Erling Henanger
      /___       /____/ /       /   /|  /   /           Norwegian Institute
     /          /\     /       /   / | /   |   ___      of Technology. (NTH)
    ------     /  \   /____   /   /  |/     \____| o    Atari Lives !