Return to BSD News archive
Path: euryale.cc.adfa.oz.au!newshost.anu.edu.au!harbinger.cc.monash.edu.au!news.mira.net.au!news.mel.connect.com.au!munnari.OZ.AU!news.ecn.uoknor.edu!news.eng.convex.com!newshost.convex.com!newsgate.duke.edu!news-server.ncren.net!news_server.cs.unc.edu!not-for-mail
From: leech@cs.unc.edu (Jon Leech)
Newsgroups: comp.graphics.api.misc,comp.graphics.algorithms,comp.windows.x.i386unix,gnu.misc.discuss,comp.unix.bsd.freebsd.misc,comp.os.linux.development.apps
Subject: Hoo's faster? (was Re: Freeware community needs 3d library NOW)
Followup-To: comp.graphics.api.misc
Date: 26 Jun 1996 01:21:36 -0400
Organization: UNC PixelFlow Project
Lines: 45
Message-ID: <4qqhd0$dh0@watt.cs.unc.edu>
References: <4qlf04$7pj@guysmiley.blarg.net> <4qn08b$ntl@fido.asd.sgi.com> <4qpd9c$9at@hell.team17.com> <4qpq2i$ah@fido.asd.sgi.com>
NNTP-Posting-Host: watt.cs.unc.edu
Xref: euryale.cc.adfa.oz.au comp.graphics.api.misc:633 comp.graphics.algorithms:28980 comp.windows.x.i386unix:20358 gnu.misc.discuss:26715 comp.unix.bsd.freebsd.misc:22145 comp.os.linux.development.apps:18060
In article <4qpq2i$ah@fido.asd.sgi.com>,
Allen Akin <akin@tuolumne.asd.sgi.com> wrote:
>It's a commonly-held belief (heavily promoted by the Direct3D folks,
>among others) that OpenGL is slow. A more accurate statement is that
>the original Microsoft release of software-only OpenGL was slow. It's
>a mistake to use that data point to draw a conclusion about the
>performance of a tuned implementation.
>
>There is no technical barrier that prevents either a fast
>software-only version of OpenGL or a fast hardware-accelerated version
>of OpenGL. You'll see a tuned software-only version soon; Microsoft
>and SGI are working on it actively. Hardware-accelerated versions
>have been available for quite a while, and in fact the best available
>3D graphics performance on both PCs and workstations is achieved with
>OpenGL.
If someone back in the early 80s said that "the best floating point
performance is achieved with Fortran", it did not imply that FORTRAN was a
superior language to all others - merely that FORTRAN was what Cray chose to
devote resources to supporting on the X-MP 48, and that was the fastest
hardware at the time (replace with appropriate machine and language for
timeframe of interest).
So the question of whether a hardware-accelerated OpenGL is faster than
a hardware-accelerated something else is open. XYZZY corporation could ship
a 50 Mtri/s platform tomorrow which only supports HOOPS, for all any of us
know. Personally I think it depends so much on the hardware that it's not a
useful question - nobody I can think of develops graphics hardware without
regard to the intended software layer (though we come pretty close at times
:-) - but it's sure to come up again and again anyway. Haven't the
Renderware folks complained from time to time about their inability to get
down to the metal on SGI machines, or am I misremembering?
I am curious to see if SGI will at some point support NT and a Direct3D
layer for it, and if that will run on top of OpenGL.
>This has been discussed very frequently in the past, so if you're not
>convinced about the technical claims, and want to open the issue
>again, please follow-up to comp.graphics.api.opengl only.
Actually I'm following up to c.g.a.misc, since it is not about OpenGL
more than anything else.
Jon
__@/