home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.unix.sysv386
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!moe.ksu.ksu.edu!ux1.cso.uiuc.edu!bradley.bradley.edu!guru
- From: guru@camelot.bradley.edu (Jerry Whelan)
- Subject: Re: Weitek 4167 and SVR4
- Message-ID: <1992Nov7.230308.11734@bradley.bradley.edu>
- Keywords: weitek, svr4, speed
- Sender: news@bradley.bradley.edu
- Nntp-Posting-Host: stasi.bradley.edu
- Organization: The Smelly Ditch
- References: <1992Nov6.010249.26474@netcom.com> <1218@bazooka.amb.org>
- Date: Sat, 7 Nov 92 23:03:08 GMT
- Lines: 23
-
- ti@bazooka.UUCP (Ti Kan) writes:
- -} cgates@netcom.com (Craig Gates) writes:
- -} >I'm considering buying and installing a Weitek 4167 in my 486DX 33Mhz.
- -} >Would the speedup be worth the money? I see in some ads that the coprocessor
- -} >can be had for $350-500.
- -}
- -} Even if the OS supports the Weitek, it is not likely that your applications
- -} were compiled to use the Weitek chip. The Weitek is not Intel 387 compatible
- -} and applications needs to be compiled with special compilers that generate
- -} Weitek instructions in order to take advantage of the chip.
-
- The math libraries for AT&T's version of SVR4 (and thus probably
- all vendors' versions) support the Weitek as we well as the 387 chipsets.
- You don't need to recompile or anything as the decision on what instructions
- to use is done at run time.
- However, as the original poster says he works primarily with X11
- and designs user interface stuff, he probably is not doing much floating
- point work (X by itself has very few, if any, functions that do floating
- point calculations) and thus would not see much speedup if he were to
- install a math coprocessor.
-
- -------------------------------------------------------------------------------
- ``written by a drunken insane pathological liar'' guru@stasi.bradley.edu
-