Path: utzoo!utgpu!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!mailrus!ames!pasteur!helios.ee.lbl.gov!nosc!ucsd!sdcc6!sdcc19!sdcc14!cs64fba From: cs64fba@sdcc14.ucsd.edu (TAI) Newsgroups: comp.sys.atari.st Subject: Re: TOS Replacement Project Message-ID: <359@sdcc14.ucsd.edu> Date: 1 Dec 88 16:56:19 GMT Reply-To: cs64fba@sdcc14.UUCP Organization: University of California, San Diego Lines: 15 Sender:pa1132@sdcc15.ucsd.edu via sdcc14.ucsd.edu Keywords:TOS, STOS, ST-LOG, 68000 Has anyone read the newest issue of ST-log? In an article concerning the X-former II (8-bit Atari Emulator) the author states that the TOS uses non-optimizing 68000 codes "everywhere". This must be one reason for GEM's slowness. Perhaps one approach currently for improving the ST is to write faster, more optimizing routines for components of GEM/TOS (such as AES, VDI, and Line-A routines) and interface these routines to the OS so that programs which call these TOS routines often can use our faster routines instead, like what the program "Turbo ST" is trying to do. This way current programs can remain compitable, and their performance can be improved. This approach also encourages the writing of well-behaved program since they can call our faster routines instead of finding someway to cheat in order to get the performance they need. This method can also serve as a starting point of writing a more complete TOS substitution. Any comments?