>From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!EU.net!blackbush.xlink.net!newsfeed.freenet.de!nntp-cust.primenet.com!natasha.rmi.net!den-news1.rmi.net!not-for-mail Thu Mar 8 10:27:44 2001 Message-ID: <3AA02BDF.46B91AAF@valinux.com> From: Brian Paul Organization: VA Linux Systems, Inc. X-Mailer: Mozilla 4.72 [en] (X11; U; Linux 2.4.0 i686) X-Accept-Language: en MIME-Version: 1.0 Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Lines: 19 Date: Fri, 02 Mar 2001 16:25:19 -0700 NNTP-Posting-Host: 207.174.125.34 X-Complaints-To: news@rmi.net X-Trace: den-news1.rmi.net 983574762 207.174.125.34 (Fri, 02 Mar 2001 16:12:42 MST) NNTP-Posting-Date: Fri, 02 Mar 2001 16:12:42 MST Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!EU.net!blackbush.xlink.net!newsfeed.freenet.de!nntp-cust.primenet.com!natasha.rmi.net!den-news1.rmi.net!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90566 Tom Impelluso wrote: > > Hi All, > > Can someone advise me on the history of OpenGL? > > Specifically... as I understand... > > OpenGL emerged in the 80's as the selected graphics api paradigm. Not in the 80's! I think Kurt Akely said he started OpenGL in 1989. I believe the first public release of the 1.0 spec was in 1992. OpenGL is a direct descendant of SGI's IRIS GL. I don't know when IRIS GL was first deployed. -Brian >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!newsfeed.berkeley.edu!ucberkeley!logbridge.uoregon.edu!news.maxwell.syr.edu!sunqbc.risq.qc.ca!newsflash.concordia.ca!not-for-mail Thu Mar 8 10:27:44 2001 From: V-man Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: Fri, 2 Mar 2001 18:56:01 -0500 Organization: Concordia University, Montreal, Canada Lines: 33 Message-ID: References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <3AA02BDF.46B91AAF@valinux.com> NNTP-Posting-Host: alcor.concordia.ca Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII NNTP-Posting-User: v_melkon In-Reply-To: <3AA02BDF.46B91AAF@valinux.com> Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!newsfeed.berkeley.edu!ucberkeley!logbridge.uoregon.edu!news.maxwell.syr.edu!sunqbc.risq.qc.ca!newsflash.concordia.ca!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90567 THat makes it quite new then. So what was the first 3d api? I remember some educational show about NASA having some kind of VR system that was wireframe only and you could walk up the stairs (or it was an escalator) And while I'm at it, what machines were used in the making of Star Wars (the firts one) and what software was there back then that could do that. V-man On Fri, 2 Mar 2001, Brian Paul wrote: > Tom Impelluso wrote: > > > > Hi All, > > > > Can someone advise me on the history of OpenGL? > > > > Specifically... as I understand... > > > > OpenGL emerged in the 80's as the selected graphics api paradigm. > > Not in the 80's! > > I think Kurt Akely said he started OpenGL in 1989. I believe the > first public release of the 1.0 spec was in 1992. > > OpenGL is a direct descendant of SGI's IRIS GL. I don't know when > IRIS GL was first deployed. > > -Brian > >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!newshub2.rdc1.sfba.home.com!news.home.com!news1.rdc1.md.home.com.POSTED!not-for-mail Thu Mar 8 10:27:44 2001 Message-ID: <3AA2E27A.4CA4AED2@my-deja.com> From: Samuel Paik Organization: spatial dexyne X-Mailer: Mozilla 4.61 [en] (Win98; U) X-Accept-Language: en MIME-Version: 1.0 Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Lines: 52 Date: Mon, 05 Mar 2001 00:38:53 GMT NNTP-Posting-Host: 24.23.166.69 X-Complaints-To: abuse@home.net X-Trace: news1.rdc1.md.home.com 983752733 24.23.166.69 (Sun, 04 Mar 2001 16:38:53 PST) NNTP-Posting-Date: Sun, 04 Mar 2001 16:38:53 PST Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!newshub2.rdc1.sfba.home.com!news.home.com!news1.rdc1.md.home.com.POSTED!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90586 Tom Impelluso wrote: > OpenGL emerged in the 80's as the selected graphics api paradigm. > Can someone, just to enable me to understand the context... tell > me the names of the competing paradigms? Renderman, maybe? > or a few others? Early 90s. OpenGL is a [3D] graphics API. It isn't a paradigm, but a concrete example of a paradigm. [*] At the time of its introduction, its main competition was PHIGS/PHIGS PLUS (and PEX--PHIGS extension to X). Generally speaking, both are from the same paradigm of procedural graphics. Some major differences included PHIGS emphasized editable display lists ("structures") while OpenGL emphasized dynamic "immediate-mode" rendering; PHIGS was structured with nearly all features optional while OpenGL had a large required core of functionality; PHIGS specified (but did not require...) a very rich set of features while OpenGL is primarily a generalization of a pipelined z-buffer. > Why did OpenGL emerge victorious? This is a tough question, but I believe it has as much to do with SGI's prominent role in the non-simulator [**] 3D graphics industry as with technical issues. Another is that PHIGS was annoying to write to as the very different feature sets and bindings between vendors often made it hard to write portable software (a situation that Direct3D was in for awhile until most implementations supported a similar large core...). Sam [*] I started writing a long bit about paradigms in 3D but cut it out. At a high level, the paradigm split is between procedural graphics and declarative graphics. The difference is in a procedural world you tell the graphics system what to do vs. in a declarative world, you describing your scene and then telling the graphics system to produce a rendering. This distinction is often blurred for ease of use (towards declarative systems) and performance (toward procedural systems). Examples of procedural systems: Pascal, OpenGL, PHIGS, D3DIM Examples of declarative systems: Prolog, HOOPS, Renderman, most ray tracers, D3DRM [**] The flight simulation industry is a quite different market that SGI didn't really start getting a noticeable foothold in until fairly late. -- Samuel S. Paik | http://www.webnexus.com/users/paik/ 3D and multimedia, architecture and implementation >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!usc.edu!howland.erols.net!news-peer-europe.sprintlink.net!news.sprintlink.net!newsfeed.sunet.se!news01.sunet.se!news.chalmers.se!not-for-mail Thu Mar 8 10:27:44 2001 From: "Marcus Lindblom" Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: Mon, 5 Mar 2001 02:42:39 +0100 Organization: Chalmers University of Technology Lines: 52 Message-ID: <97uqtt$pec$1@nyheter.chalmers.se> References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <3AA02BDF.46B91AAF@valinux.com> <97r23i$ksc$1@nyheter.chalmers.se> <4f24ato89mdpkffha94ai8tgccqqlct63s@4ax.com> NNTP-Posting-Host: dynamic-193-140.dialup.chalmers.se X-Trace: nyheter.chalmers.se 983756541 26060 129.16.193.140 (5 Mar 2001 01:42:21 GMT) X-Complaints-To: abuse@chalmers.se NNTP-Posting-Date: 5 Mar 2001 01:42:21 GMT X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.50.4133.2400 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400 Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!usc.edu!howland.erols.net!news-peer-europe.sprintlink.net!news.sprintlink.net!newsfeed.sunet.se!news01.sunet.se!news.chalmers.se!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90588 "Jim M" <}Jimbo{@brit-cit.demon.co.uk> wrote in message news:4f24ato89mdpkffha94ai8tgccqqlct63s@4ax.com... > On Sat, 3 Mar 2001 16:20:34 +0100, "Marcus Lindblom" > wrote: > > >"V-man" wrote in message > >news:Pine.OSF.4.31.0103021852180.6873-100000@alcor.concordia.ca... > >> > >> And while I'm at it, what machines were used in the making of Star Wars > >> (the firts one) and what software was there back then that could do that. > > > >There were no computers involved in making the first SW movies! > > None that did tremendous CGI sequences, although I could imagine > computers used for camera-tracking of models, which later get masked > onto the final film. The end death-star sequence with all that camera > movement through all those models would be a pain if it wasn't > roboticly controlled. Could be, but I doubt it as computers were _quite_ expensive back then and at least episode IV was quite a low budget thing. But I lack hard facts here, so this amounts to pure speculation. > >One of the first movies that had a significant amount of CG > >in it was 'The Last Starfighter' .. lovely flatshaded polygons > >without texturemapping! It came in 1984 or so, I think. > > All the space sequences were CGI and didn't it look wonderful back > then. Nothing compared with the likes of Jurrasic Park but it did the > job. How can you compare it to Jurrasic Park, which came oodles of years later, when computing power had increased by orders of magnitude? > >(Tron is another candidate for first CG enhanced movie, > >haven't seen that though.) > > Not seen Tron !!! That can't be right. I though everone has seen > Tron... :) This came out 1978/9 ish and was a mixture of live action > overscanned with CGI as well as pure CG sequences + proper live > sequences. Still waiting for a full production model of a light-bike > :) I know and I'm ashamed. I'm going to correct that fact as soon as possible. :) /Marcus >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!nntp.primenet.com!nntp.gblx.net!newsfeed.direct.ca!look.ca!news-hub.cableinet.net!internal-news-hub.cableinet.net!news2.cableinet.net.POSTED!not-for-mail Thu Mar 8 10:27:44 2001 From: "Charles E Hardwidge" Newsgroups: comp.graphics.api.opengl References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <3AA02BDF.46B91AAF@valinux.com> <97r23i$ksc$1@nyheter.chalmers.se> <4f24ato89mdpkffha94ai8tgccqqlct63s@4ax.com> <97uqtt$pec$1@nyheter.chalmers.se> Subject: Re: history of OpenGL Lines: 19 X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.50.4133.2400 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400 Message-ID: <6AGo6.2762$th.1027364@news2.cableinet.net> Date: Mon, 05 Mar 2001 06:43:14 GMT NNTP-Posting-Host: 213.48.112.210 X-Complaints-To: http://www.blueyonder.co.uk/abuse X-Trace: news2.cableinet.net 983774594 213.48.112.210 (Mon, 05 Mar 2001 06:43:14 GMT) NNTP-Posting-Date: Mon, 05 Mar 2001 06:43:14 GMT Organization: blueyonder (post doesn't reflect views of blueyonder) Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!nntp.primenet.com!nntp.gblx.net!newsfeed.direct.ca!look.ca!news-hub.cableinet.net!internal-news-hub.cableinet.net!news2.cableinet.net.POSTED!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90601 >Could be, but I doubt it as computers were _quite_ expensive back >then and at least episode IV was quite a low budget thing. I remember computers only being used for camera tracking. If cost wasn't the determining factor, quality probably was. CGI was in it's infancy and models were the superior tech. I still stand by models as a technology. CGI still doesn't cut it when the going get's rough, although at the end of the day, this is largely determined by what you are attempting to achieve. Going completely OT. Apparantly the farmhouse seen on the hill in Gladiator (Ridly Scott), was generated by CGI. This is an example of it's usage at it's best. Good location, good light, and CGI that enhanced rather than intruded. Sorry, just a hobby horse of mine. :) -- Charles E. Hardwidge http://www.hardwidge.org.uk >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!newshub2.rdc1.sfba.home.com!news.home.com!newsfeed.direct.ca!look.ca!news-hub.cableinet.net!internal-news-hub.cableinet.net!news2.cableinet.net.POSTED!not-for-mail Thu Mar 8 10:27:44 2001 From: "Charles E Hardwidge" Newsgroups: comp.graphics.api.opengl References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <3AA2E27A.4CA4AED2@my-deja.com> Subject: Re: history of OpenGL Lines: 9 X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.50.4133.2400 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400 Message-ID: <4CGo6.2763$th.1027268@news2.cableinet.net> Date: Mon, 05 Mar 2001 06:45:20 GMT NNTP-Posting-Host: 213.48.112.210 X-Complaints-To: http://www.blueyonder.co.uk/abuse X-Trace: news2.cableinet.net 983774720 213.48.112.210 (Mon, 05 Mar 2001 06:45:20 GMT) NNTP-Posting-Date: Mon, 05 Mar 2001 06:45:20 GMT Organization: blueyonder (post doesn't reflect views of blueyonder) Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!newshub2.rdc1.sfba.home.com!news.home.com!newsfeed.direct.ca!look.ca!news-hub.cableinet.net!internal-news-hub.cableinet.net!news2.cableinet.net.POSTED!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90602 >[*] I started writing a long bit about paradigms in 3D but cut it out. Damn. That was getting interesting. -- Charles E. Hardwidge http://www.hardwidge.org.uk >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!newshub2.rdc1.sfba.home.com!news.home.com!news1.rdc1.md.home.com.POSTED!not-for-mail Thu Mar 8 10:27:44 2001 Message-ID: <3AA35282.C7D1A689@my-deja.com> From: Samuel Paik Organization: spatial dexyne X-Mailer: Mozilla 4.61 [en] (Win98; U) X-Accept-Language: en MIME-Version: 1.0 Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <3AA2E27A.4CA4AED2@my-deja.com> <4CGo6.2763$th.1027268@news2.cableinet.net> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Lines: 49 Date: Mon, 05 Mar 2001 08:36:47 GMT NNTP-Posting-Host: 24.23.166.69 X-Complaints-To: abuse@home.net X-Trace: news1.rdc1.md.home.com 983781407 24.23.166.69 (Mon, 05 Mar 2001 00:36:47 PST) NNTP-Posting-Date: Mon, 05 Mar 2001 00:36:47 PST Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!newshub2.rdc1.sfba.home.com!news.home.com!news1.rdc1.md.home.com.POSTED!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90605 Charles E Hardwidge wrote: > > >[*] I started writing a long bit about paradigms in 3D but cut it out. > > Damn. That was getting interesting. Ok, what do you want me to talk about? I'm not sure what I can write about which isn't already in textbooks or is mere ephemera of a changing market. And really, I got in too late and was too peripheral to have personally witnessed a lot the earlier history of 3D graphics, so a lot of what I would say is really second hand or mere impressions. (I started out working on Dore', a cross-platform 3D API, at Intelligent Light in the fall of 1989 as a co-op student. They still exist and the code I worked on may still exist in their main product ) Again, I started writing a bunch of stuff then decided to cut almost all of it out. Let me know what you think I should write about and if I have time and inclination I'll try to write some stuff here. Going back to the original poster, back in the late 1980s, there was really only one "open" 3D standard, PHIGS+, and lots of proprietary vendor APIs (IRIS GL for SGI, Starbase for HP, GMR3D for Apollo) and several proprietary cross platform APIs (e.g. HOOPS, Dore') While PHIGS+ had a large advantage, having actually made it through the ISO standards process, it had lots of problems which impeded its cross-platform capabilities and the standards committee made it very slow to adapt to rapidly advancing hardware capabilities. If you think the OpenGL ARB is slow, you haven't watched an ISO standards committee. About that time, workstation vendors were consolidating on X as their 2D graphics system, instead of the mess of proprietary 2D graphics systems they had before, and PEX, the PHIGS Extension to X, appeared which helped on some of the cross-platform problems of PHIGS+. PEX and the APIs which drove it, PEXlib and the various PHIGS implementations, was what OpenGL had to go up against in the early 1990s. I believe OpenGL started out as a more or less clean sheet way to rationalize SGI's 3D API situation. IRIS GL had lots of device-specific (and maybe even somewhat incompatible) variants and was full of deprecated functionality. For some reason, they decided it was worthwhile to open up the API to the rest of the industry which lead to rather direct competition with PEX, where by around 1995, it was obvious that OpenGL had prevailed. Sam Paik >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!torn!newsfeed.direct.ca!look.ca!news-hub.cableinet.net!internal-news-hub.cableinet.net!news2.cableinet.net.POSTED!not-for-mail Thu Mar 8 10:27:44 2001 From: "Charles E Hardwidge" Newsgroups: comp.graphics.api.opengl References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <3AA2E27A.4CA4AED2@my-deja.com> <4CGo6.2763$th.1027268@news2.cableinet.net> <3AA35282.C7D1A689@my-deja.com> Subject: Re: history of OpenGL Lines: 14 X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.50.4133.2400 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400 Message-ID: Date: Mon, 05 Mar 2001 08:55:51 GMT NNTP-Posting-Host: 213.48.112.210 X-Complaints-To: http://www.blueyonder.co.uk/abuse X-Trace: news2.cableinet.net 983782551 213.48.112.210 (Mon, 05 Mar 2001 08:55:51 GMT) NNTP-Posting-Date: Mon, 05 Mar 2001 08:55:51 GMT Organization: blueyonder (post doesn't reflect views of blueyonder) Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!torn!newsfeed.direct.ca!look.ca!news-hub.cableinet.net!internal-news-hub.cableinet.net!news2.cableinet.net.POSTED!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90606 >Ok, what do you want me to talk about? I'm not sure what I can write >about which isn't already in textbooks or is mere ephemera of a changing >market. Sorry Sam, I was thinking aloud. The historical guff was interesting, as the early API's passed me by. Perhaps your longer view and greater experience can shed some light on what the future might hold? (The further you look back, the more you know of today, etcetera). -- Charles E. Hardwidge http://www.hardwidge.org.uk >From fc.hp.com!news Thu Mar 8 10:27:45 2001 Path: fc.hp.com!news From: "Paul Martz" Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: Mon, 5 Mar 2001 07:53:41 -0700 Organization: Hewlett-Packard Fort Collins Site Lines: 82 Approved: paul_martz@hp.com Message-ID: <98099p$6lc$1@fcnews.fc.hp.com> References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> NNTP-Posting-Host: fcpm87756.fc.hp.com X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.50.4133.2400 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400 Xref: fc.hp.com comp.graphics.api.opengl:90620 I believe SGI viewed PEX as its major competition at the time that OpenGL first hit the streets. As others have said, 1992 sounds about right. I was a PEX developer at the time, and I recall a rather lengthy posting from SGI to the comp.graphics.api.pex newsgroup which formally stated why OpenGL was superior to PEX. (One of the stated reasons was that OpenGL had built-in 2D support in the way of glDrawPixels, glReadPixels, and glCopyPixels -- operations which, to this day, perform quite poorly on many implementations. And the ARB has clearly seen these functions as deficient for doing 2D graphics, thus the imaging subset was born.) (Another advantage of OpenGL over PEX was "no subsetting". That is, a PEX implementation could be called "PEX" as long as it implemented one of three models of operation, or subsets. An app written using one subset would not run on a PEX implementation that used another subset. Sadly, the current state of OpenGL with several vendor specific extensions, bears strong resemblance to PEX's subset model.) Why was SGI out to get PEX? If I remember right, PEX was a consortium of big workstation vendors that wanted to elbow in on SGI's hold of the 3D market. Rumor has it that SGI was never invited to join the PEX consortium. SGI didn't care to see its user base migrate off to PEX, so "revamped" Iris GL to make it an open standard. PEX still hangs on to this day -- there are still a few large ISVs that have not migrated off of PEX/PHIGS to OpenGL. PHIGS, of course, had the backing of the US Government; I don't follow PHIGS anymore so whether or not this is still true, I don't know. Why did OpenGL "win"? In many ways, it is a more pleasant API to work with, either as an app writer or as an API implementer/developer. While PHIGS/PEX is "CISC", OpenGL is more "RISC". OpenGL was designed after the needs of SGI's large ISV base, so has pretty much all the features an app would need. Even PEX/PHIGS heavyweights like Evans & Sutherland quickly realized that they had to do OpenGL to stay in the market. -- -Paul Martz (paul_martz@hp.com) Hewlett Packard TCD Personal Workstations Lab OpenGL FAQ: http://www.opengl.org/developers/faqs/technical.html http://www.frii.com/~martz/oglfaq "Tom Impelluso" wrote in message news:3AA01F4A.7F3442BA@caesar.sdsu.edu... > Hi All, > > Can someone advise me on the history of OpenGL? > > Specifically... as I understand... > > OpenGL emerged in the 80's as the selected graphics api paradigm. > > Can someone, just to enable me to understand the context... tell > me the names of the competing paradigms? Renderman, maybe? > or a few others? > > > Why did OpenGL emerge victorious? > > And, with regard to Object Oriented... > Yes, I am aware that OpenInventor emerged... > and VRML was based, in part, on that... > > So, with regard to OO/Web I see see that Java3D may emerge victorious > (could be wrong here... but I would prefer not to start a debate thread > on that), > so what other tecnologies are competing with Java3D? > > thanks > tom > > > (again, I do not wish to start a debate on which is the best... I simply > > would like to know what the competitors of OpenGL *were*, and what > the competitors for Java3D *are*) > > > >From fc.hp.com!news Thu Mar 8 10:27:45 2001 Path: fc.hp.com!news From: "Paul Martz" Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: Mon, 5 Mar 2001 08:04:15 -0700 Organization: Hewlett-Packard Fort Collins Site Lines: 35 Approved: paul_martz@hp.com Message-ID: <9809ti$6ud$1@fcnews.fc.hp.com> References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <3AA02BDF.46B91AAF@valinux.com> NNTP-Posting-Host: fcpm87756.fc.hp.com X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.50.4133.2400 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400 Xref: fc.hp.com comp.graphics.api.opengl:90621 "V-man" wrote in message news:Pine.OSF.4.31.0103021852180.6873-100000@alcor.concordia.ca... > THat makes it quite new then. So what was the first 3d api? Define "API". In the late 60s, Ivan Sutherland developed some simple vector graphics, "sketch" I think it was called. You could draw boxes or arches freehand with a light pen and the computer would straighten your lines for you. E&S's first product was the LDS -- Line Drawing System, right around 1970. IIRC right, E&S's PS300 terminals supported an AVS-like programming paradigm; the 1st edition Foley and Van Dam has info on this, if you can find one in a used bookstore or coworker's shelf. All this happened prior to SGI splitting off from E&S. > And while I'm at it, what machines were used in the making of Star Wars > (the firts one) and what software was there back then that could do that. Computers were used to control camera position. Star Wars was 100% models, blue screens, and matte paintings. There are some scenes where computers were clearly used, but only for simple wireframe renderings -- The "battle training" scene where the rebels are told how to blow up the Death Star, for example. In the first "Star Trek" movie, wireframe computer images present on the bridge's control screens were done by Steve McAllister at E&S -- not sure what hardware he used, though. -- -Paul Martz (paul_martz@hp.com) Hewlett Packard TCD Personal Workstations Lab OpenGL FAQ: http://www.opengl.org/developers/faqs/technical.html http://www.frii.com/~martz/oglfaq >From fc.hp.com!news Thu Mar 8 10:27:45 2001 Path: fc.hp.com!news From: "Paul Martz" Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: Mon, 5 Mar 2001 08:18:32 -0700 Organization: Hewlett-Packard Fort Collins Site Lines: 29 Approved: paul_martz@hp.com Message-ID: <980aoc$7bb$1@fcnews.fc.hp.com> References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <3AA2E27A.4CA4AED2@my-deja.com> <4CGo6.2763$th.1027268@news2.cableinet.net> <3AA35282.C7D1A689@my-deja.com> NNTP-Posting-Host: fcpm87756.fc.hp.com X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.50.4133.2400 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400 Xref: fc.hp.com comp.graphics.api.opengl:90623 "Samuel Paik" wrote in message news:3AA35282.C7D1A689@my-deja.com... > I believe OpenGL started out as a more or less clean sheet > way to rationalize SGI's 3D API situation. IRIS GL had lots of > device-specific (and maybe even somewhat incompatible) variants > and was full of deprecated functionality. For some reason, they > decided it was worthwhile to open up the API to the rest of the > industry which lead to rather direct competition with PEX, > where by around 1995, it was obvious that OpenGL had prevailed. >From what I recall, general consensus in the PEX community was that the ONLY reason OpenGL existed was so SGI could compete on the "open spec" playing field with PEX, and thus prevent further erosion of its market share. The fact that it helped clean up a splintered Iris GL was perceived as a convenient side-effect. Disclaimer: The paragraph above merely sums up my recollection of the general sentiment of the PEX community in the early 1990s and is not intended as the personal opinion of anyone in particular, nor is it intended as a slam against SGI, who did the 3D community a great service by developing OpenGL. -- -Paul Martz (paul_martz@hp.com) Hewlett Packard TCD Personal Workstations Lab OpenGL FAQ: http://www.opengl.org/developers/faqs/technical.html http://www.frii.com/~martz/oglfaq >From fc.hp.com!news Thu Mar 8 10:27:45 2001 Path: fc.hp.com!news From: "Paul Martz" Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: Mon, 5 Mar 2001 08:22:40 -0700 Organization: Hewlett-Packard Fort Collins Site Lines: 16 Approved: paul_martz@hp.com Message-ID: <980b04$7j9$1@fcnews.fc.hp.com> References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> NNTP-Posting-Host: fcpm87756.fc.hp.com X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.50.4133.2400 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400 Xref: fc.hp.com comp.graphics.api.opengl:90625 As a historical note, a PEX prototype was first demonstrated at siggraph 88 in three booths: DEC, Sun, and HP, featuring a multi-player networked 3D "pong" style game, running on all three vendor's platforms. The app was written using PHIGS calls layered on top of PEX. PEX memorabilia: http://www.frii.com/~martz/pex/ -- -Paul Martz (paul_martz@hp.com) Hewlett Packard TCD Personal Workstations Lab OpenGL FAQ: http://www.opengl.org/developers/faqs/technical.html http://www.frii.com/~martz/oglfaq >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!arclight.uoregon.edu!newsfeed.mathworks.com!dispose.news.demon.net!news.demon.co.uk!demon!brit-cit.demon.co.uk!not-for-mail Thu Mar 8 10:27:45 2001 From: Jim M <}Jimbo{@brit-cit.demon.co.uk> Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: Mon, 05 Mar 2001 22:49:21 +0000 Message-ID: References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <3AA02BDF.46B91AAF@valinux.com> <97r23i$ksc$1@nyheter.chalmers.se> <4f24ato89mdpkffha94ai8tgccqqlct63s@4ax.com> <97uqtt$pec$1@nyheter.chalmers.se> Reply-To: {Jim$M}@brit-cit.demon.co.uk NNTP-Posting-Host: brit-cit.demon.co.uk X-NNTP-Posting-Host: brit-cit.demon.co.uk:194.222.176.191 X-Trace: news.demon.co.uk 983832617 nnrp-09:4265 NO-IDENT brit-cit.demon.co.uk:194.222.176.191 X-Complaints-To: abuse@demon.net X-Newsreader: Forte Agent 1.7/32.534 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Lines: 35 Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!arclight.uoregon.edu!newsfeed.mathworks.com!dispose.news.demon.net!news.demon.co.uk!demon!brit-cit.demon.co.uk!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90653 On Mon, 5 Mar 2001 02:42:39 +0100, "Marcus Lindblom" wrote: [More OT ranting] >> >One of the first movies that had a significant amount of CG >> >in it was 'The Last Starfighter' .. lovely flatshaded polygons >> >without texturemapping! It came in 1984 or so, I think. >> >> All the space sequences were CGI and didn't it look wonderful back >> then. Nothing compared with the likes of Jurrasic Park but it did the >> job. > >How can you compare it to Jurrasic Park, which came oodles of years >later, when computing power had increased by orders of magnitude? Erm... "Nothing compared..." ? Anyway, oodles of years doesn't seem like much. I went to see both of those (+ tron) when they had their cinema releases, and I think I only went on the promise of glorious CGI eye-candy. Similarities: (1979) TRON: mainstream CGI at it's best. (1984) TLS: mainstream CGI at it's best. (1993) JP: mainstream CGI at it's best. (2000) Gladiator: mainstream CGI at it's best. :) I feel a sad pattern emerging... Jimbo -- @ Derbyshire >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!logbridge.uoregon.edu!xmission!nnrp.xmission!not-for-mail Thu Mar 8 10:27:45 2001 From: legalize+jeeves@mail.xmission.com (Rich) Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: 6 Mar 2001 09:02:12 -0700 Organization: multi-cellular, biological Lines: 28 Message-ID: <9831m4$ai3$1@xmission.xmission.com> References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <3AA2E27A.4CA4AED2@my-deja.com> Reply-To: (Rich) legalize+jeeves@mail.xmission.com NNTP-Posting-Host: xmission.xmission.com X-Trace: news.xmission.com 983894533 25215 198.60.22.20 (6 Mar 2001 16:02:13 GMT) X-Complaints-To: abuse@xmission.com NNTP-Posting-Date: 6 Mar 2001 16:02:13 GMT X-Reply-Etiquette: No copy by email, please Mail-Copies-To: never Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!logbridge.uoregon.edu!xmission!nnrp.xmission!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90698 [Please do not mail me a copy of your followup] Samuel Paik spake the secret code <3AA2E27A.4CA4AED2@my-deja.com> thusly: >Some major differences included >PHIGS emphasized editable display lists ("structures") while OpenGL >emphasized dynamic "immediate-mode" rendering; PHIGS was structured >with nearly all features optional while OpenGL had a large required >core of functionality; PHIGS specified (but did not require...) a very >rich set of features while OpenGL is primarily a generalization of a >pipelined z-buffer. PHIGS/PHIGS-PLUS is an ISO standard, apparently very heavily used by the oil & gas industries who made their investments in computer graphics long before SGI existed as a gleam in Jim Clark's eye. OGL was born in response to PEX, and PEX did provide a PHIGS API, but it also provided a protocol-level API that didn't require the use of structures. The "optional" bits of PHIGS/PHIGS-PLUS (and PEX by extension) made for a horrible mess. The plethora of OGL extensions seems to be making the same sort of quagmire for OGL. Its a side effect of OGL's success, unfortunately. -- Ask me about my upcoming book on Direct3D from Addison-Wesley! Direct3D Book Home Fractals! >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!newsxfer.eecs.umich.edu!news.cc.utah.edu!xmission!nnrp.xmission!not-for-mail Thu Mar 8 10:27:45 2001 From: legalize+jeeves@mail.xmission.com (Rich) Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: 6 Mar 2001 09:07:23 -0700 Organization: multi-cellular, biological Lines: 39 Message-ID: <9831vr$b3f$1@xmission.xmission.com> References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <98099p$6lc$1@fcnews.fc.hp.com> Reply-To: (Rich) legalize+jeeves@mail.xmission.com NNTP-Posting-Host: xmission.xmission.com X-Trace: news.xmission.com 983894844 25295 198.60.22.20 (6 Mar 2001 16:07:24 GMT) X-Complaints-To: abuse@xmission.com NNTP-Posting-Date: 6 Mar 2001 16:07:24 GMT X-Reply-Etiquette: No copy by email, please Mail-Copies-To: never Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!newsxfer.eecs.umich.edu!news.cc.utah.edu!xmission!nnrp.xmission!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90699 [Please do not mail me a copy of your followup] "Paul Martz" spake the secret code <98099p$6lc$1@fcnews.fc.hp.com> thusly: >[...] I recall a rather lengthy posting from SGI to >the comp.graphics.api.pex newsgroup which formally stated why OpenGL was >superior to PEX. It was an ongoing flamewar, actually, with the core assertion that somehow immediate mode was -always- superior to a display list. The SGI marketing folks were doing their best to make PEX look inferior. In reality, they were closer to each other than either side probably wanted to admit. SGI would say stupid things like (imagine two cavemen from a Gary Larson Far Side comic) "Thag! Display lists bad! Immediate mode good!" until it was pointed out that the SGI benchmarks that obtained the maximum performance out of the machine used display lists. Ooops, better change flaming tactics... >Why was SGI out to get PEX? PEX was the first multivendor viable alternative 3D graphics API that had any chance of competing with Iris GL. SGI only licensed Iris GL once -- to IBM -- and I understand the fee was hefty. >[...] PHIGS, of course, had the backing of the US Government That's probably because government contracts often have language in there requiring that ISO standards be used wherever possible. I think nowadays people work around that by saying "I can't do X in the ISO standard PHIGS/PHIGS-PLUS" which gives them the freedom to use whatever they want to achieve X. PHIGS/PHIGS-PLUS is an ISO standard and not just a US gov't thing. -- Ask me about my upcoming book on Direct3D from Addison-Wesley! Direct3D Book Home Fractals! >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!logbridge.uoregon.edu!xmission!nnrp.xmission!not-for-mail Thu Mar 8 10:27:45 2001 From: legalize+jeeves@mail.xmission.com (Rich) Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: 6 Mar 2001 09:08:22 -0700 Organization: multi-cellular, biological Lines: 15 Message-ID: <98321m$b6p$1@xmission.xmission.com> References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <3AA02BDF.46B91AAF@valinux.com> <9809ti$6ud$1@fcnews.fc.hp.com> Reply-To: (Rich) legalize+jeeves@mail.xmission.com NNTP-Posting-Host: xmission.xmission.com X-Trace: news.xmission.com 983894903 25300 198.60.22.20 (6 Mar 2001 16:08:23 GMT) X-Complaints-To: abuse@xmission.com NNTP-Posting-Date: 6 Mar 2001 16:08:23 GMT X-Reply-Etiquette: No copy by email, please Mail-Copies-To: never Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!logbridge.uoregon.edu!xmission!nnrp.xmission!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90700 [Please do not mail me a copy of your followup] "Paul Martz" spake the secret code <9809ti$6ud$1@fcnews.fc.hp.com> thusly: >In the first "Star Trek" movie, wireframe computer images present on the >bridge's control screens were done by Steve McAllister at E&S -- not sure >what hardware he used, though. I believe it was an Evans & Sutherland PS/390. -- Ask me about my upcoming book on Direct3D from Addison-Wesley! Direct3D Book Home Fractals! >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!logbridge.uoregon.edu!xmission!nnrp.xmission!not-for-mail Thu Mar 8 10:27:45 2001 From: legalize+jeeves@mail.xmission.com (Rich) Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: 6 Mar 2001 10:54:46 -0700 Organization: multi-cellular, biological Lines: 14 Message-ID: <983896$lia$1@xmission.xmission.com> References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <3AA02BDF.46B91AAF@valinux.com> Reply-To: (Rich) legalize+jeeves@mail.xmission.com NNTP-Posting-Host: xmission.xmission.com X-Trace: news.xmission.com 983901288 27037 198.60.22.20 (6 Mar 2001 17:54:48 GMT) X-Complaints-To: abuse@xmission.com NNTP-Posting-Date: 6 Mar 2001 17:54:48 GMT X-Reply-Etiquette: No copy by email, please Mail-Copies-To: never Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!logbridge.uoregon.edu!xmission!nnrp.xmission!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90708 [Please do not mail me a copy of your followup] Brian Paul spake the secret code <3AA02BDF.46B91AAF@valinux.com> thusly: >OpenGL is a direct descendant of SGI's IRIS GL. I don't know when >IRIS GL was first deployed. It was first deployed when Silicon Graphics was started. -- Ask me about my upcoming book on Direct3D from Addison-Wesley! Direct3D Book Home Fractals! >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!logbridge.uoregon.edu!xmission!nnrp.xmission!not-for-mail Thu Mar 8 10:27:45 2001 From: legalize+jeeves@mail.xmission.com (Rich) Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: 6 Mar 2001 11:00:34 -0700 Organization: multi-cellular, biological Lines: 20 Message-ID: <9838k2$m2t$1@xmission.xmission.com> References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <3AA02BDF.46B91AAF@valinux.com> <97r23i$ksc$1@nyheter.chalmers.se> Reply-To: (Rich) legalize+jeeves@mail.xmission.com NNTP-Posting-Host: xmission.xmission.com X-Trace: news.xmission.com 983901635 27123 198.60.22.20 (6 Mar 2001 18:00:35 GMT) X-Complaints-To: abuse@xmission.com NNTP-Posting-Date: 6 Mar 2001 18:00:35 GMT X-Reply-Etiquette: No copy by email, please Mail-Copies-To: never Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!logbridge.uoregon.edu!xmission!nnrp.xmission!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90711 [Please do not mail me a copy of your followup] "Marcus Lindblom" spake the secret code <97r23i$ksc$1@nyheter.chalmers.se> thusly: >Define first 3d api. If you mean the first generic cross-platform api >that has had a huge breakthrough, it's probably OpenGL. Define "huge breakthrough"; I suspect your definition is colored by the recent history of PC graphics. Computer graphics as a commercially viable technology predates the PC, probably by about 10 years. The first generic cross-platform API that was widely used was GKS, the precursor to PHIGS/PHIGS-PLUS. -- Ask me about my upcoming book on Direct3D from Addison-Wesley! Direct3D Book Home Fractals! >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!paloalto-snf1.gtei.net!news.gtei.net!newsfeed.stanford.edu!canoe.uoregon.edu!logbridge.uoregon.edu!xmission!nnrp.xmission!not-for-mail Thu Mar 8 10:27:45 2001 From: legalize+jeeves@mail.xmission.com (Rich) Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: 6 Mar 2001 11:04:58 -0700 Organization: multi-cellular, biological Lines: 30 Message-ID: <9838sa$mfk$1@xmission.xmission.com> References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <3AA02BDF.46B91AAF@valinux.com> <9809ti$6ud$1@fcnews.fc.hp.com> Reply-To: (Rich) legalize+jeeves@mail.xmission.com NNTP-Posting-Host: xmission.xmission.com X-Trace: news.xmission.com 983901903 27242 198.60.22.20 (6 Mar 2001 18:05:03 GMT) X-Complaints-To: abuse@xmission.com NNTP-Posting-Date: 6 Mar 2001 18:05:03 GMT X-Reply-Etiquette: No copy by email, please Mail-Copies-To: never Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!paloalto-snf1.gtei.net!news.gtei.net!newsfeed.stanford.edu!canoe.uoregon.edu!logbridge.uoregon.edu!xmission!nnrp.xmission!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90712 [Please do not mail me a copy of your followup] "Paul Martz" spake the secret code <9809ti$6ud$1@fcnews.fc.hp.com> thusly: >All this happened prior to SGI splitting off from E&S. SGI didn't "split off from E&S". Jim Clark was an Evans & Sutherland employee and approached Dave Evans about making a graphics workstation. Reputedly Dr. Evans' response was "workstations are a fad". Clark had been working on a procedural oriented "graphics library" API while at E&S. At the time, E&S was using a dataflow driven language on its PS/390 product. Dataflow is a good paradigm for large data blocks, but not for itty bitty things like ints and floats. At any rate, Clark left E&S to become a professor at Stanford and form Silicon Graphics Incorporated -- now officially named "SGI". This whole story repeats itself with uncanny similarity when you substitute "PostScript" for "graphics library" and Dave Warnock for Jim Clark. E&S invented so many things, its sad that they haven't achieved as much commercial success as others with their ideas. The frame buffer, PostScript and "GL" all originated at E&S. -- Ask me about my upcoming book on Direct3D from Addison-Wesley! Direct3D Book Home Fractals! >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!newsfeed.mathworks.com!sunqbc.risq.qc.ca!cyclone2.usenetserver.com!news-out.usenetserver.com!telocity-west!TELOCITY!newsrump.sjc.telocity.net!not-for-mail Thu Mar 8 10:27:45 2001 Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <98099p$6lc$1@fcnews.fc.hp.com> <9831vr$b3f$1@xmission.xmission.com> From: akin3@pobox.com Reply-To: akin3@pobox.com Message-ID: User-Agent: slrn/0.9.6.2 (Linux) Lines: 78 X-Trace: NzggTm9BdXRoVXNlciBURUxPQ0lUWS1SRUFERVJTIDY0LjE5NS4xNzAuMTA1ICBUdWUsIDA2IE1h!ciAyMDAxIDEyOjM4OjA0IFBTVA== X-Abuse-Info: Please forward ALL headers when reporting abuse. X-Complaints-To: abuse@telocity.net NNTP-Posting-Date: Tue, 06 Mar 2001 12:38:04 PST Date: Tue, 06 Mar 2001 20:38:04 GMT Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!newsfeed.mathworks.com!sunqbc.risq.qc.ca!cyclone2.usenetserver.com!news-out.usenetserver.com!telocity-west!TELOCITY!newsrump.sjc.telocity.net!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90723 I was going to stay out of this because I no longer read c.g.a.o regularly, but I guess a few personal observations are in order. I managed the software group at DEC that produced the device-dependent code and microcode for the first commercial implementation of PEX (on the VAXStation 3520/3540). We worked with other groups at DEC, Sun, and elsewhere that were active in creating PEX and PEXlib. (Andy Vesper remembers all this, probably better than I do.) Later, I managed the OpenGL group at SGI and went on to become one of the managers in the high-end (RealityEngine, InfiniteReality) product group there. One of my jobs while in the OpenGL group was to monitor Usenet and make a convincing case for OpenGL to the technical community. Having experience with both PEX and OpenGL made it easier to do this during the time PEX and OpenGL were competing actively. On 6 Mar 2001 09:07:23 -0700, Rich wrote: | "Paul Martz" spake the secret code | <98099p$6lc$1@fcnews.fc.hp.com> thusly: | | >[...] I recall a rather lengthy posting from SGI to | >the comp.graphics.api.pex newsgroup which formally stated why OpenGL was | >superior to PEX. Paul's probably thinking of the white paper I wrote for the PEX-vs-OpenGL SIGGRAPH debate, which was eventually posted on Usenet as well as being distributed in hardcopy at the debate. I think I still have a copy of this lying around somewhere. | It was an ongoing flamewar, actually, with the core assertion that | somehow immediate mode was -always- superior to a display list. ... You're right that it was a long war, but as a participant on the SGI side (maybe the most frequent participant on the SGI side), I disagree with your characterization of the "core assertion." PEX structures aren't the same as OpenGL display lists; they're editable, which makes for some significant issues. I'm sure I asserted that immediate mode was superior to frequently-edited PEX structures, but I also sure I asserted that OpenGL display lists were superior to infrequently-edited PEX structures. There was a great deal of technical discussion (including some benchmarks) to provide evidence for that position. | ... The | SGI marketing folks were doing their best to make PEX look inferior. Odds are good that you didn't see any of the very few Usenet posts from the OpenGL marketing guy (there was only one at the time). In those days Usenet was pretty much exclusively an engineering forum as far as SGI was concerned. Quite a few engineers participated in the OpenGL-vs-PEX discussion, though. | ... SGI would say stupid things like (imagine two | cavemen from a Gary Larson Far Side comic) "Thag! Display lists bad! | Immediate mode good!" until it was pointed out that the SGI benchmarks | that obtained the maximum performance out of the machine used display | lists. Ooops, better change flaming tactics... You might want to review those old messages. I think you'd find a few surprises. :-) As for the question asked elsewhere -- why did SGI create OpenGL in the first place? -- that was also discussed in some detail on the net in the old days. The main reason was that software vendors said they had to expand their hardware base in order to survive, so they gave SGI the choice of opening up GL in some form or they'd drop support for SGI hardware and go for the higher-volume workstations. PEXlib was one of several alternative APIs that would have competed for those software vendors if OpenGL hadn't been created. Allen (Note: Remove the random digit from my return address to reply to me.) >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!csulb.edu!newshub.sdsu.edu!not-for-mail Thu Mar 8 10:27:45 2001 From: Tom Impelluso Newsgroups: comp.graphics.api.opengl Subject: history of OpenGL Date: Fri, 02 Mar 2001 14:31:39 -0800 Organization: San Diego State University Lines: 35 Message-ID: <3AA01F4A.7F3442BA@caesar.sdsu.edu> NNTP-Posting-Host: caesar.sdsu.edu Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Trace: gondor.sdsu.edu 983572294 7965 130.191.163.149 (2 Mar 2001 22:31:34 GMT) X-Complaints-To: news@newshub.sdsu.edu NNTP-Posting-Date: 2 Mar 2001 22:31:34 GMT X-Mailer: Mozilla 4.76C-SGI [en] (X11; U; IRIX64 6.5 IP30) X-Accept-Language: en Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!csulb.edu!newshub.sdsu.edu!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90561 Hi All, Can someone advise me on the history of OpenGL? Specifically... as I understand... OpenGL emerged in the 80's as the selected graphics api paradigm. Can someone, just to enable me to understand the context... tell me the names of the competing paradigms? Renderman, maybe? or a few others? Why did OpenGL emerge victorious? And, with regard to Object Oriented... Yes, I am aware that OpenInventor emerged... and VRML was based, in part, on that... So, with regard to OO/Web I see see that Java3D may emerge victorious (could be wrong here... but I would prefer not to start a debate thread on that), so what other tecnologies are competing with Java3D? thanks tom (again, I do not wish to start a debate on which is the best... I simply would like to know what the competitors of OpenGL *were*, and what the competitors for Java3D *are*) >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!newsxfer.eecs.umich.edu!news.cc.utah.edu!xmission!nnrp.xmission!not-for-mail Thu Mar 8 10:27:46 2001 From: legalize+jeeves@mail.xmission.com (Rich) Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: 6 Mar 2001 14:16:49 -0700 Organization: multi-cellular, biological Lines: 66 Message-ID: <983k41$arf$1@xmission.xmission.com> References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <98099p$6lc$1@fcnews.fc.hp.com> <9831vr$b3f$1@xmission.xmission.com> Reply-To: (Rich) legalize+jeeves@mail.xmission.com NNTP-Posting-Host: xmission.xmission.com X-Trace: news.xmission.com 983913410 435 198.60.22.20 (6 Mar 2001 21:16:50 GMT) X-Complaints-To: abuse@xmission.com NNTP-Posting-Date: 6 Mar 2001 21:16:50 GMT X-Reply-Etiquette: No copy by email, please Mail-Copies-To: never Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!newsxfer.eecs.umich.edu!news.cc.utah.edu!xmission!nnrp.xmission!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90726 [Please do not mail me a copy of your followup] akin3@pobox.com spake the secret code thusly: >You're right that it was a long war, but as a participant on the SGI >side (maybe the most frequent participant on the SGI side), I disagree >with your characterization of the "core assertion." I was one of the participants as well, and I remember this blanket assertion coming up repeatedly until it was deflated by pointing out that the peak benchmarks on a machine like the VGX could only be obtained by using a display list. When Kurt Akeley published the benchmark in response to a 'put up or shut up' challenge I made about the 1MTri/sec VGX claim, it indeed showed that the benchmark used GL display lists (OpenGL didn't exist yet) to achieve the peak number. I had gotten sick of the local sales rep repeatedly babbling about 1 M Tri/sec when he was showing a demo of only about 60,000 triangles that were texture mapped, environment mapped, etc. When I specifically asked him to show me any program that achieved 1 million triangles all I got back was a bunch of mumbling. He didn't have any demos that could achieve the peak numbers; it turned out that only a very specific set of circumstances could obtain the peak numbers -- as Kurt Akeley's benchmark showed -- and that typical numbers an application would see were far, far less than 1 Million tri/sec. >PEX structures >aren't the same as OpenGL display lists; they're editable, which makes >for some significant issues. Yes, the fact that display lists under OGL are write-only is a significant difference compared to the PEX structures. But the SGI marketing guy (Bill Glazier was his name, I think?) was spewing out a tremendous amount of FUD about display lists and PEX/PHIGS at the time. >Odds are good that you didn't see any of the very few Usenet posts >from the OpenGL marketing guy (there was only one at the time). At the time I remember seeing each and every one of the very few usenet posts from the marketing guy. >| ... SGI would say stupid things like (imagine two >| cavemen from a Gary Larson Far Side comic) "Thag! Display lists bad! >| Immediate mode good!" until it was pointed out that the SGI benchmarks >| that obtained the maximum performance out of the machine used display >| lists. Ooops, better change flaming tactics... > >You might want to review those old messages. I think you'd find a few >surprises. :-) In this case when I say "SGI" I'm referring only to the marketing guy. (Isn't it funny how reasonable comments aren't remembered and only the flaming dufus assertions are the ones that stick in your mind?) The engineers were reasonable, unless they were just repeating the marketing schpiel. Unfortunately there was some of that, but most of the engineers were saying reasonable things and most of the unreasonable things they were saying about PEX were because they were ignorant of the technical details and relied on the marketingspeak to fill in the gaps. -- Ask me about my upcoming book on Direct3D from Addison-Wesley! Direct3D Book Home Fractals! >From fc.hp.com!news Thu Mar 8 10:27:46 2001 Path: fc.hp.com!news From: "Paul Martz" Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: Thu, 8 Mar 2001 07:38:09 -0700 Organization: Hewlett-Packard Fort Collins Site Lines: 10 Approved: paul_martz@hp.com Message-ID: <9885gm$lr$1@fcnews.fc.hp.com> References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> NNTP-Posting-Host: fcpm87756.fc.hp.com X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.50.4133.2400 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400 Xref: fc.hp.com comp.graphics.api.opengl:90822 OT: Is there an efficient way to download an entire thread to file(s)? Private email responses, please. -- -Paul Martz (paul_martz@hp.com) Hewlett Packard TCD Personal Workstations Lab OpenGL FAQ: http://www.opengl.org/developers/faqs/technical.html http://www.frii.com/~martz/oglfaq >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!usc.edu!howland.erols.net!newsfeed.skycache.com!Cidera!newsie2.cent.net!not-for-mail Thu Mar 8 10:27:46 2001 From: "Andrew F. Vesper" Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: Tue, 06 Mar 2001 17:09:45 -0500 Organization: Mostly disorganized Lines: 31 Message-ID: <3AA56029.D63BFACE@acm.org> References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <3AA02BDF.46B91AAF@valinux.com> <97r23i$ksc$1@nyheter.chalmers.se> <9838k2$m2t$1@xmission.xmission.com> NNTP-Posting-Host: 206.190.210.196 Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Trace: news.cent.net 983916818 27856 206.190.210.196 (6 Mar 2001 22:13:38 GMT) X-Complaints-To: usenet@news.cent.net NNTP-Posting-Date: Tue, 6 Mar 2001 22:13:38 +0000 (UTC) X-Mailer: Mozilla 4.75 [en] (Win98; U) X-Accept-Language: en Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!usc.edu!howland.erols.net!newsfeed.skycache.com!Cidera!newsie2.cent.net!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90727 Rich wrote: > The first generic cross-platform API that was widely used was GKS, > the precursor to PHIGS/PHIGS-PLUS. GKS was much more widely used in Europe than in the USA, both in 2D and 3D forms. If I recall correctly, it became an ISO standard before PHIGS. However, I don't believe that GKS is the true precursor to PHIGS, but rather the Siggraph "Core" library gave birth to both GKS and PHIGS. Searching Siggraph.org, I found: http://www.siggraph.org/publications/newsletter//v32n1/columns/carson.html This doesn't contradict my belief, but it doesn't particularly reinforce it either. I recall having a "Core" library available for the Digital personal computer that was based on the PDP-11. (As opposed to the Digital personal computer based on the PDP-8 and the Digital personal computer based on the Intel 8080.) -- Andy V (OpenGL Alpha Geek) "In order to make progress, one must leave the door to the unknown ajar." Richard P. Feynman, quoted by Jagdish Mehra in _The Beat of a Different Drum_. Paul Martz's OpenGL FAQ: http://www.opengl.org/developers/faqs/technical.html >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!logbridge.uoregon.edu!xmission!nnrp.xmission!not-for-mail Thu Mar 8 10:27:46 2001 From: legalize+jeeves@mail.xmission.com (Rich) Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: 6 Mar 2001 15:32:16 -0700 Organization: multi-cellular, biological Lines: 22 Message-ID: <983ohg$hm0$1@xmission.xmission.com> References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <97r23i$ksc$1@nyheter.chalmers.se> <9838k2$m2t$1@xmission.xmission.com> <3AA56029.D63BFACE@acm.org> Reply-To: (Rich) legalize+jeeves@mail.xmission.com NNTP-Posting-Host: xmission.xmission.com X-Trace: news.xmission.com 983917937 1715 198.60.22.20 (6 Mar 2001 22:32:17 GMT) X-Complaints-To: abuse@xmission.com NNTP-Posting-Date: 6 Mar 2001 22:32:17 GMT X-Reply-Etiquette: No copy by email, please Mail-Copies-To: never Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!logbridge.uoregon.edu!xmission!nnrp.xmission!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90728 [Please do not mail me a copy of your followup] "Andrew F. Vesper" spake the secret code <3AA56029.D63BFACE@acm.org> thusly: >However, I don't believe that GKS is the true precursor to PHIGS, but >rather the Siggraph "Core" library gave birth to both GKS and PHIGS. You're probalby right on that. When I first started doing 3D graphics I was briefly exposed to GKS and it was a painful experience mostly due to the lack of documentation and nobody seemed to know how to make the library work at all with the terminals we had. (I vaguely seem to recall they were some kind of Tektronix graphics terminals connected to a central VAX, not true workstations.) Most of what I know about 3D graphics pre-1988 consists of the results of my journeys into "computing archaeology". -- Ask me about my upcoming book on Direct3D from Addison-Wesley! Direct3D Book Home Fractals! >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!newsfeed.Austria.EU.net!newsfeed.kpnqwest.at!nautilus.visp-europe.psi.com!newsfeed00.sul.t-online.de!t-online.de!newsfeed.freenet.de!news-out.nuthinbutnews.com!news-in-austin.nuthinbutnews.com!telocity-west!TELOCITY!newsrump.sjc.telocity.net!not-for-mail Thu Mar 8 10:27:46 2001 Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <98099p$6lc$1@fcnews.fc.hp.com> <9831vr$b3f$1@xmission.xmission.com> <983k41$arf$1@xmission.xmission.com> From: akin6@pobox.com Reply-To: akin6@pobox.com Message-ID: User-Agent: slrn/0.9.6.2 (Linux) Lines: 27 X-Trace: MjcgTm9BdXRoVXNlciBURUxPQ0lUWS1SRUFERVJTIDY0LjE5NS4xNzAuMTA1ICBUdWUsIDA2IE1h!ciAyMDAxIDE0OjQwOjM5IFBTVA== X-Abuse-Info: Please forward ALL headers when reporting abuse. X-Complaints-To: abuse@telocity.net NNTP-Posting-Date: Tue, 06 Mar 2001 14:40:39 PST Date: Tue, 06 Mar 2001 22:40:39 GMT Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!newsfeed.Austria.EU.net!newsfeed.kpnqwest.at!nautilus.visp-europe.psi.com!newsfeed00.sul.t-online.de!t-online.de!newsfeed.freenet.de!news-out.nuthinbutnews.com!news-in-austin.nuthinbutnews.com!telocity-west!TELOCITY!newsrump.sjc.telocity.net!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90729 On 6 Mar 2001 14:16:49 -0700, Rich wrote: | | >You're right that it was a long war, but as a participant on the SGI | >side (maybe the most frequent participant on the SGI side), I disagree | >with your characterization of the "core assertion." | | I was one of the participants as well, and I remember this blanket | assertion coming up repeatedly until it was deflated by pointing out | that the peak benchmarks on a machine like the VGX could only be | obtained by using a display list. When Kurt Akeley published the | benchmark in response to a 'put up or shut up' challenge I made about | the 1MTri/sec VGX claim, it indeed showed that the benchmark used | GL display lists (OpenGL didn't exist yet) to achieve the peak number. Ah. Perhaps you're thinking about a different Usenet discussion than I was? The OpenGL-vs-PEX debate I had in mind obviously took place after OpenGL existed. I dug up the old PEX-vs-OpenGL debate document and skimmed it. It does talk about immediate-mode vs. structure-store at some length, going over data duplication, reformatting, etc. It also talks about the advantages of OpenGL display lists over PEX structures, pretty much as I noted in the previous post. Allen (Please remove the random digit from my return address to reply to me.) >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!logbridge.uoregon.edu!xmission!nnrp.xmission!not-for-mail Thu Mar 8 10:27:46 2001 From: legalize+jeeves@mail.xmission.com (Rich) Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: 6 Mar 2001 16:03:53 -0700 Organization: multi-cellular, biological Lines: 56 Message-ID: <983qcp$l73$1@xmission.xmission.com> References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <983k41$arf$1@xmission.xmission.com> Reply-To: (Rich) legalize+jeeves@mail.xmission.com NNTP-Posting-Host: xmission.xmission.com X-Trace: news.xmission.com 983919834 2207 198.60.22.20 (6 Mar 2001 23:03:54 GMT) X-Complaints-To: abuse@xmission.com NNTP-Posting-Date: 6 Mar 2001 23:03:54 GMT X-Reply-Etiquette: No copy by email, please Mail-Copies-To: never Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!logbridge.uoregon.edu!xmission!nnrp.xmission!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90733 [Please do not mail me a copy of your followup] akin6@pobox.com spake the secret code thusly: >Ah. Perhaps you're thinking about a different Usenet discussion than >I was? The OpenGL-vs-PEX debate I had in mind obviously took place >after OpenGL existed. The benchmark that Kurt posted did result from a different discussion thread. However, it was concurrent at the time of what I called "API Wars, Volume 1". (Volume 2 was D3D vs. OGL. Quicktime3D was so uninteresting to the graphics mainstream, we didn't even debate it.) The reason I think they were concurrent is that when I went to SIGGRAPH that year (Vegas? My neurons are fuzzy.), Bill Glazier made a point of giving me a long sleeve OGL T-shirt. Assuming I remembered his name correctly. However, it could have been a year or two earlier. My first SIGGRAPH was 1990 in Dallas. >I dug up the old PEX-vs-OpenGL debate document and skimmed it. It >does talk about immediate-mode vs. structure-store at some length, >going over data duplication, reformatting, etc. It also talks about >the advantages of OpenGL display lists over PEX structures, pretty >much as I noted in the previous post. I think Mark Kilgard summarized all these points in his "PEX vs. OGL" paper on the SGI web site that I linked to in my web page discussing D3D vs. OGL. Considering the network transparency of PEX, its probably a more apples-to-apples comparison to compare PEX to GLX where the GL client is running remotely but displaying locally. Realistically, nobody really coded for this scenario because the amount of data that needs to be chucked around for useful 3D graphics makes it quicker to render remotely and just send the rendered image to be displayed locally when the scene is complex.. I think Los Alamos did some research in the past couple of years on using this sort of paradigm for distributed graphics. With 3D web content, the whole issue of transporting 3D scene data efficiently across a low-bandwidth connection has returned. This has spurred lots of research in mesh compression, which didn't exist 10 years ago. Also Low-Bandwidth X came after all this as well. LBX is great for the X protocol over a dialup connection but still doesn't have enough compression power to deal with 3D scenes. In fact, I don't know what LBX does with extension protocol requests, but it probably just compresses them with no additional semantic compression. LBX uses its knowledge of the semantics of the core X protocol to achieve some additional compression beyond straight LZW compression of protocol requests, I believe. -- Ask me about my upcoming book on Direct3D from Addison-Wesley! Direct3D Book Home Fractals! >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!newsfeed.mathworks.com!nntp-out.monmouth.com!newspeer.monmouth.com!newsfeed.mad.ttd.net!telenews.teleline.es!not-for-mail Thu Mar 8 10:27:46 2001 Message-ID: <3AA56C81.F7F5B4B2@egg.chips.and.spam.com> From: fungus X-Mailer: Mozilla 4.7 [en] (WinNT; U) X-Accept-Language: en MIME-Version: 1.0 Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <98099p$6lc$1@fcnews.fc.hp.com> <9831vr$b3f$1@xmission.xmission.com> <983k41$arf$1@xmission.xmission.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Lines: 60 Date: Tue, 06 Mar 2001 23:09:11 GMT NNTP-Posting-Host: 213.0.67.223 X-Complaints-To: usenet@teleline.es X-Trace: telenews.teleline.es 983920151 213.0.67.223 (Wed, 07 Mar 2001 00:09:11 MET) NNTP-Posting-Date: Wed, 07 Mar 2001 00:09:11 MET Organization: Clientes_Teleline Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!newsfeed.mathworks.com!nntp-out.monmouth.com!newspeer.monmouth.com!newsfeed.mad.ttd.net!telenews.teleline.es!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90732 Rich wrote: > > I was one of the participants as well, and I remember this blanket > assertion coming up repeatedly until it was deflated by pointing out > that the peak benchmarks on a machine like the VGX could only be > obtained by using a display list. > I threw a VGX into the bin about a month ago (literally!), along with an even older SGI GT machine. I was moving house and decided my "museum" just wasn't worth it. That VGX nearly drove me mad with the noise it made, I had to sit right next to it so it would have been about 100dB. In winter it kept me cozy and warm, in summer it would overheat and switch itself off via a little sensor in the top of the machine. I had to work at night to get any work done. > I had gotten sick of the local sales rep repeatedly babbling about 1 > M Tri/sec when he was showing a demo of only about 60,000 triangles > that were texture mapped, environment mapped, etc. That sounds about right. > typical numbers an application would see > were far, far less than 1 Million tri/sec. > I've still got my personal benchmarks in a text file: On a good day, without the texture mapping, you could get about 200,000 tiny triangles out of a VGX. As soon as you turned the texture on the performance dropped to about 50,000 triangles (in a display list you'd get a little bit more - all those bgnpolygon()/v3f()/v3f()/v3f()/endpolygon() calls were a serious overhead on a 32MHz machine) The pixel fill rate without texture was 200 megapixels/sec which was decent enough (even by today's standards!) but as soon as you turned texturing on it dropped to about 17 megapixels/sec with no perspective correction. The VGXT added perspective correction and doubled the textured fill rate. Still, with only 512k of texture memory you couldn't do an awful lot of texturing, just some basic patterns on the ground/water. Remember that this was a machine which cost hundreds of thousands of dollars. Kids today... they don't know they're born.... -- <\___/> / O O \ \_____/ FTB. >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!telocity-west!TELOCITY!newsrump.sjc.telocity.net!not-for-mail Thu Mar 8 10:27:46 2001 Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <983k41$arf$1@xmission.xmission.com> <983qcp$l73$1@xmission.xmission.com> From: akin9@pobox.com Reply-To: akin9@pobox.com Message-ID: User-Agent: slrn/0.9.6.2 (Linux) Lines: 45 X-Trace: NDUgTm9BdXRoVXNlciBURUxPQ0lUWS1SRUFERVJTIDY0LjE5NS4xNzAuMTA1ICBUdWUsIDA2IE1h!ciAyMDAxIDE4OjM2OjQwIFBTVA== X-Abuse-Info: Please forward ALL headers when reporting abuse. X-Complaints-To: abuse@telocity.net NNTP-Posting-Date: Tue, 06 Mar 2001 18:36:40 PST Date: Wed, 07 Mar 2001 02:36:40 GMT Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!telocity-west!TELOCITY!newsrump.sjc.telocity.net!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90745 On 6 Mar 2001 16:03:53 -0700, Rich wrote: | | >Ah. Perhaps you're thinking about a different Usenet discussion than | >I was? The OpenGL-vs-PEX debate I had in mind obviously took place | >after OpenGL existed. | | The benchmark that Kurt posted did result from a different discussion | thread. However, it was concurrent at the time of what I called "API | Wars, Volume 1". The timing must have been close. What I think of as the PEX/OpenGL wars started right after the SIGGRAPH debate in 1992, in Chicago. For the discussions in which I participated, I'm pretty sure SGI people didn't take a rabid "immediate-mode at any cost!" position. (If that had been a widely-held opinion inside SGI, then IrisGL wouldn't have "objects" and OpenGL wouldn't have display lists.) But there's probably no way for us to figure out why you and I have such different recollections until the Deja archives come back online. | ... Considering the network transparency of PEX, its | probably a more apples-to-apples comparison to compare PEX to GLX | where the GL client is running remotely but displaying locally. Not sure I follow you, but I'm also not sure it matters much. The direct-rendering case dominates most applications nowadays. | | ...quicker to render remotely and just send the rendered image | to be displayed locally when the scene is complex.. I think Los Alamos | did some research in the past couple of years on using this sort of | paradigm for distributed graphics. Yep. Still an interesting approach. | LBX is great for the X protocol over a dialup connection but still | doesn't have enough compression power to deal with 3D scenes. ... Ever read John Danskin's dissertation (Princeton) on compression techniques for network-transparent graphics? Worthwhile if that's a topic of interest to you... Allen (Remove the random digit from my return address to reply to me.) >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!newshub2.rdc1.sfba.home.com!news.home.com!news1.rdc1.mi.home.com.POSTED!not-for-mail Thu Mar 8 10:27:46 2001 From: "Dave Brondsema" Newsgroups: comp.graphics.api.opengl References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <983k41$arf$1@xmission.xmission.com> <983qcp$l73$1@xmission.xmission.com> Subject: Re: history of OpenGL Lines: 20 X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.50.4522.1200 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200 Message-ID: Date: Wed, 07 Mar 2001 03:12:50 GMT NNTP-Posting-Host: 24.11.123.12 X-Complaints-To: abuse@home.net X-Trace: news1.rdc1.mi.home.com 983934770 24.11.123.12 (Tue, 06 Mar 2001 19:12:50 PST) NNTP-Posting-Date: Tue, 06 Mar 2001 19:12:50 PST Organization: Excite@Home - The Leader in Broadband http://home.com/faster Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!newshub2.rdc1.sfba.home.com!news.home.com!news1.rdc1.mi.home.com.POSTED!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90748 wrote in message news:slrn9ab7lo.bom.akin@nospam.net... > On 6 Mar 2001 16:03:53 -0700, Rich wrote: > But > there's probably no way for us to figure out why you and I have such > different recollections until the Deja archives come back online. > Allen > > (Remove the random digit from my return address to reply to me.) why not search at http://groups.google.com ? You're automatically forwarded there from http://www.deja.com/usenet All the archives are available for searching, but feature-limited. Dave Brondsema >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!newshub2.rdc1.sfba.home.com!news.home.com!news1.rdc1.md.home.com.POSTED!not-for-mail Thu Mar 8 10:27:46 2001 Message-ID: <3AA5AB09.EF9398BA@my-deja.com> From: Samuel Paik Organization: spatial dexyne X-Mailer: Mozilla 4.61 [en] (Win98; U) X-Accept-Language: en MIME-Version: 1.0 Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <3AA02BDF.46B91AAF@valinux.com> <9809ti$6ud$1@fcnews.fc.hp.com> <9838sa$mfk$1@xmission.xmission.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Lines: 12 Date: Wed, 07 Mar 2001 03:18:47 GMT NNTP-Posting-Host: 24.23.166.69 X-Complaints-To: abuse@home.net X-Trace: news1.rdc1.md.home.com 983935127 24.23.166.69 (Tue, 06 Mar 2001 19:18:47 PST) NNTP-Posting-Date: Tue, 06 Mar 2001 19:18:47 PST Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!newshub2.rdc1.sfba.home.com!news.home.com!news1.rdc1.md.home.com.POSTED!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90749 Rich wrote: > This whole story repeats itself with uncanny similarity when you > substitute "PostScript" for "graphics library" and Dave Warnock for > Jim Clark. You mean John Warnock. I'm pretty sure Postscript is a descendent of Xerox Interpress, which was one of the few concrete products based on Xerox PARC work. Did he come from E&S before Xerox? [could be... I see that J. Warnock has a publication, probably a thesis, from U. of Utah...] Sam Paik >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!newshub2.rdc1.sfba.home.com!news.home.com!news1.rdc1.md.home.com.POSTED!not-for-mail Thu Mar 8 10:27:46 2001 Message-ID: <3AA5AFA3.5DA1450E@my-deja.com> From: Samuel Paik Organization: spatial dexyne X-Mailer: Mozilla 4.61 [en] (Win98; U) X-Accept-Language: en MIME-Version: 1.0 Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <983k41$arf$1@xmission.xmission.com> <983qcp$l73$1@xmission.xmission.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Lines: 9 Date: Wed, 07 Mar 2001 03:38:22 GMT NNTP-Posting-Host: 24.23.166.69 X-Complaints-To: abuse@home.net X-Trace: news1.rdc1.md.home.com 983936302 24.23.166.69 (Tue, 06 Mar 2001 19:38:22 PST) NNTP-Posting-Date: Tue, 06 Mar 2001 19:38:22 PST Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!newshub2.rdc1.sfba.home.com!news.home.com!news1.rdc1.md.home.com.POSTED!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90750 Dave Brondsema wrote: > why not search at http://groups.google.com ? You're automatically forwarded > there from http://www.deja.com/usenet All the archives are available for > searching, but feature-limited. First, the full Deja archive isn't online yet. Second, the full Deja archive only goes back to 1995--most of this happened around 1992. Sam >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!logbridge.uoregon.edu!xmission!nnrp.xmission!not-for-mail Thu Mar 8 10:27:47 2001 From: legalize+jeeves@mail.xmission.com (Rich) Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: 6 Mar 2001 22:07:26 -0700 Organization: multi-cellular, biological Lines: 45 Message-ID: <984fme$s0i$1@xmission.xmission.com> References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <983qcp$l73$1@xmission.xmission.com> Reply-To: (Rich) legalize+jeeves@mail.xmission.com NNTP-Posting-Host: xmission.xmission.com X-Trace: news.xmission.com 983941648 8269 198.60.22.20 (7 Mar 2001 05:07:28 GMT) X-Complaints-To: abuse@xmission.com NNTP-Posting-Date: 7 Mar 2001 05:07:28 GMT X-Reply-Etiquette: No copy by email, please Mail-Copies-To: never Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!logbridge.uoregon.edu!xmission!nnrp.xmission!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90751 [Please do not mail me a copy of your followup] akin9@pobox.com spake the secret code thusly: >The timing must have been close. What I think of as the PEX/OpenGL >wars started right after the SIGGRAPH debate in 1992, in Chicago. You're talking about the panel discussion where Marty Hess from Sun (I think) played the videotape of the two old ladies arguing graphics APIs over their white picket fence? ;-) My "little grey cells" think that the discussion in question happened before the SIGGRAPH panel because: a) E&S came out with their PEX box in '89/90, when Paul Martz and I both worked in the server group there. API Wars, Volume 1 seemed to start around the same time, about 2 years before OGL was announced. b) I remember that the VGX/T had come out around the same time and the issue of its benchmark claims was hot on everyone's lips, and apparently other people had been trying to get ahold of this "1 M tri/sec" benchmark from SGI for quite some time, with no success until I challenged them on the claim. Kurt Akeley responded saying that they took their benchmark numbers very seriously and immediately published the benchmark that validated their claim. c) This was all before anyone knew of OpenGL (the VGX/T benchmark was in Iris GL, not Open GL) and those of us at E&S considered Iris GL our competition. Yes, I think we are remembering two different threads. The issue of "display list vs. immediate mode" has been around forever in graphics it seems, and the sweet spot shifts from here to there and back again as we roll around the wheel of reincarnation. Since the E&S box predated OpenGL, we were hot and heavy into the "display list debate" before OGL even appeared on the scene and the comp.graphics.api.* newsgroup hierarchy didn't even exist yet! Most of these threads were taking place in comp.graphics.misc, I think. -- Ask me about my upcoming book on Direct3D from Addison-Wesley! Direct3D Book Home Fractals! >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!logbridge.uoregon.edu!xmission!nnrp.xmission!not-for-mail Thu Mar 8 10:27:47 2001 From: legalize+jeeves@mail.xmission.com (Rich) Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: 6 Mar 2001 22:11:29 -0700 Organization: multi-cellular, biological Lines: 30 Message-ID: <984fu1$srq$1@xmission.xmission.com> References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <9809ti$6ud$1@fcnews.fc.hp.com> <9838sa$mfk$1@xmission.xmission.com> <3AA5AB09.EF9398BA@my-deja.com> Reply-To: (Rich) legalize+jeeves@mail.xmission.com NNTP-Posting-Host: xmission.xmission.com X-Trace: news.xmission.com 983941891 8350 198.60.22.20 (7 Mar 2001 05:11:31 GMT) X-Complaints-To: abuse@xmission.com NNTP-Posting-Date: 7 Mar 2001 05:11:31 GMT X-Reply-Etiquette: No copy by email, please Mail-Copies-To: never Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!logbridge.uoregon.edu!xmission!nnrp.xmission!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90752 [Please do not mail me a copy of your followup] Samuel Paik spake the secret code <3AA5AB09.EF9398BA@my-deja.com> thusly: >You mean John Warnock. Oops, brain cramp there, yes I mean John Warnock. >I'm pretty sure Postscript is a descendent >of Xerox Interpress, which was one of the few concrete products >based on Xerox PARC work. Did he come from E&S before Xerox? Yes, he came from E&S. What he had while at E&S wasn't even called "PostScript" yet. I'm not sure of what the exact state of his idea was when he left E&S, but I do know that he approached E&S about the idea and they turned him down. What I heard was that he got to take his idea that he developed at E&S and go commercial with it in his own company in return for some shares in the company he started. The company was Adobe. Granted, I haven't heard this from John Warnock personally, but this story was often told at E&S by "old hands" that I trusted and they weren't the kind of people known for exagerrating to make a point (like I am ;-). -- Ask me about my upcoming book on Direct3D from Addison-Wesley! Direct3D Book Home Fractals! >From fc.hp.com!news Thu Mar 8 10:27:47 2001 Path: fc.hp.com!news From: "Paul Martz" Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: Wed, 7 Mar 2001 08:16:54 -0700 Organization: Hewlett-Packard Fort Collins Site Lines: 25 Approved: paul_martz@hp.com Message-ID: <985jd9$dt1$1@fcnews.fc.hp.com> References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <98099p$6lc$1@fcnews.fc.hp.com> <9831vr$b3f$1@xmission.xmission.com> <983k41$arf$1@xmission.xmission.com> NNTP-Posting-Host: fcpm87756.fc.hp.com X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.50.4133.2400 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400 Xref: fc.hp.com comp.graphics.api.opengl:90765 Allen -- Thanks for joining the discussion. This might be a valuable thread to save. "Rich" wrote in message news:983k41$arf$1@xmission.xmission.com... > akin3@pobox.com spake the secret code > thusly: > >Odds are good that you didn't see any of the very few Usenet posts > >from the OpenGL marketing guy (there was only one at the time). > > At the time I remember seeing each and every one of the very few > usenet posts from the marketing guy. :-) Me too. If I am remembering the right guy, he was particularly fond of calling it "the OpenGL" instead of just "OpenGL", a nomenclature that I see lives on the the spec, but has been dropped from everyday usage. -- -Paul Martz (paul_martz@hp.com) Hewlett Packard TCD Personal Workstations Lab OpenGL FAQ: http://www.opengl.org/developers/faqs/technical.html http://www.frii.com/~martz/oglfaq >From fc.hp.com!news Thu Mar 8 10:27:47 2001 Path: fc.hp.com!news From: "Paul Martz" Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: Wed, 7 Mar 2001 08:34:55 -0700 Organization: Hewlett-Packard Fort Collins Site Lines: 51 Approved: paul_martz@hp.com Message-ID: <985kf2$ecn$1@fcnews.fc.hp.com> References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <983k41$arf$1@xmission.xmission.com> <983qcp$l73$1@xmission.xmission.com> NNTP-Posting-Host: fcpm87756.fc.hp.com X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.50.4133.2400 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400 Xref: fc.hp.com comp.graphics.api.opengl:90766 wrote in message news:slrn9ab7lo.bom.akin@nospam.net... > On 6 Mar 2001 16:03:53 -0700, Rich wrote: > | >Ah. Perhaps you're thinking about a different Usenet discussion than > | >I was? The OpenGL-vs-PEX debate I had in mind obviously took place > | >after OpenGL existed. > | > | The benchmark that Kurt posted did result from a different discussion > | thread. However, it was concurrent at the time of what I called "API > | Wars, Volume 1". > > The timing must have been close. What I think of as the PEX/OpenGL > wars started right after the SIGGRAPH debate in 1992, in Chicago. My memory on this is foggy. Rich and I were coworkers at E&S at the time, and he was a PhD student at UofU. I remember the SGI "demo truck" coming around to the UofU campus, and Rich going in (on his own, of course, though those of us left behind at E&S definitely viewed him as a corporate spy gathering valuable information for us :-). He told us all about the truck after his visit, and I recall that he didn't see a single demo showing off the famed 1m tris/sec (or wasn't it 1.1m tris/sec?) capability of the SGI hardware. Of the Usenet OpenGL-v-PEX debate, I only recall posting lots of obviously biased and probably stupid things that I would later live to regret. Lesson learned: don't flame anyone on Usenet, one day you might end up working with them. One point of the argument that comes to mind was the PEX community bemoaning the overhead of a function call per vertex, until it was you, Allen, I believe, who described the trick of inlining the vertex call and trapping bus accesses so that glVertex3f could be implemented as follows: void glVertex3f(GLfloat x, GLfloat y, GLfloat z) { volatile GLfloat a; a = x; a = y; a = z; } Once SGI pointed out that glVertex could be implemented in such a way that it amounted to no more that 3 floating point stores at the machine instruction level, a lot of the PEX people shut up. -- -Paul Martz (paul_martz@hp.com) Hewlett Packard TCD Personal Workstations Lab OpenGL FAQ: http://www.opengl.org/developers/faqs/technical.html http://www.frii.com/~martz/oglfaq >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!logbridge.uoregon.edu!xmission!nnrp.xmission!not-for-mail Thu Mar 8 10:27:47 2001 From: legalize+jeeves@mail.xmission.com (Rich) Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: 7 Mar 2001 11:46:31 -0700 Organization: multi-cellular, biological Lines: 96 Message-ID: <985vm7$35i$1@xmission.xmission.com> References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <983qcp$l73$1@xmission.xmission.com> <985kf2$ecn$1@fcnews.fc.hp.com> Reply-To: (Rich) legalize+jeeves@mail.xmission.com NNTP-Posting-Host: xmission.xmission.com X-Trace: news.xmission.com 983990793 22424 198.60.22.20 (7 Mar 2001 18:46:33 GMT) X-Complaints-To: abuse@xmission.com NNTP-Posting-Date: 7 Mar 2001 18:46:33 GMT X-Reply-Etiquette: No copy by email, please Mail-Copies-To: never Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!logbridge.uoregon.edu!xmission!nnrp.xmission!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90779 [Please do not mail me a copy of your followup] "Paul Martz" spake the secret code <985kf2$ecn$1@fcnews.fc.hp.com> thusly: >[...] I remember the SGI "demo truck" coming >around to the UofU campus, and [...] There was the "demo truck", but they didn't have a VGX in that truck, IIRC. (It probably would have overheated inside!) The VGX/T was first demonstrated at the University of Utah's CS department before the truck ever came around, although I happened to miss that particular demo. The 1M tri/sec challenge arose out of a separate mini visualization forum at the U., where the local SGI sales rep Richard Grossen demonstrated the VGX/T. He showed the environment mapped F16 model, I believe, while talking into a loudspeaker and saying "1 million triangles per second!!!" over and over and over again. I just asked him to put some proof behind his words and show me a million triangles in -any- demo. He couldn't because the only way to achieve this was with Kurt Akeley's (as yet unpublished) benchmark program that didn't even use texturing! The benchmark did achieve maximum throughput on the device, but it also showed that what an application was likely to do would never achieve this maximum throughput. Evans & Sutherland used 10-pixel, lit (1 light, I think), shaded, depth-buffered, depth-cueued triangles as its benchmark workhorse at the time. If I recall correctly (and I probably don't on some details), the SGI benchmark used unlit, untextured, non depth-cueued, Z buffered triangles arranged in a particular fashion into a tristrip in a display list. I don't remember the exact pixel dimensions of the triangles, but 10-25 pixels per triangle was common for benchmarks at the time. I can't recall if all the triangles were within the viewport once rendered, I seem to recall that most of them are clipped -- meaning that the benchmark demonstrated peak performance on the transformation part of the pipeline more than it demonstrated the rasterization portion of the pipeline. This was our beef with the SGI benchmarks all along -- they always talked about maximum peak possible performance (while conveniently omitting that no real world application would ever draw anything so contorted and thus would always end up with significantly less performance than the peak), whereas our benchmarks always talked about typical application performance. Naturally we each wanted to show our machines in the best light, so we used the biggest numbers we each could get without lying. The "fine art of benchmarking" has changed one bit; all the tactics are still the same and the goal is always to get a higher number than the other guy, no matter how contorted or unrealistic your benchmark has to be in order to get the higher number. Its just the two machines were designed very differently; the E&S machine was designed so that its peak performance was very close to typical application performance. Essentially everything in the system was balanced. The VGX on the other hand could achieve much higher peak performance than the ESV, even if the typical application perfomance was comparable. Of course the VGX had a texturing option, which the ESV didn't have. (Although I managed to write a demo using color-interpolated antialiased polylines that had the SGI guys doing a double-take on our box thinking we'd snuck in texturing! That was how good the AA quality of E&S' polylines was -- if you just drew them close enough and refined the polyline grid sufficiently, it could easily fool you into thinking it was a textured polygon.) The whole issue of benchmarking and peak vs. typical performance were some of the major driving forces behind the creation of the GPC benchmark suite from the SPEC organization, which has now evolved into the ViewPerf benchmarks for OGL. The tactic of playing fast and loose with benchmarks is something that still goes on today and its always important to take performance numbers with a grain of salt unless you understand the context of the benchmark claim. Just saying X triangles per second is almost meaningless without more information about what -kind- of triangles were drawn, among other factors. >Of the Usenet OpenGL-v-PEX debate, I only recall posting lots of obviously >biased and probably stupid things that I would later live to regret. Lesson >learned: don't flame anyone on Usenet, one day you might end up working with >them. Don't feel too bad, we were young guns then :-). I've always been happy to yield to someone who can show me that I'm wrong. But you have to show me :-). This is better done by engineers than marketing droids in suits. -- Ask me about my upcoming book on Direct3D from Addison-Wesley! Direct3D Book Home Fractals! >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!telocity-west!TELOCITY!newsrump.sjc.telocity.net!not-for-mail Thu Mar 8 10:27:47 2001 Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <983k41$arf$1@xmission.xmission.com> <983qcp$l73$1@xmission.xmission.com> <985kf2$ecn$1@fcnews.fc.hp.com> From: akin1@pobox.com Reply-To: akin1@pobox.com Message-ID: User-Agent: slrn/0.9.6.2 (Linux) Lines: 19 X-Trace: MTkgTm9BdXRoVXNlciBURUxPQ0lUWS1SRUFERVJTIDY0LjE5NS4xNzAuMTA1ICBXZWQsIDA3IE1h!ciAyMDAxIDExOjU0OjIwIFBTVA== X-Abuse-Info: Please forward ALL headers when reporting abuse. X-Complaints-To: abuse@telocity.net NNTP-Posting-Date: Wed, 07 Mar 2001 11:54:20 PST Date: Wed, 07 Mar 2001 19:54:20 GMT Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!telocity-west!TELOCITY!newsrump.sjc.telocity.net!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90781 On Wed, 7 Mar 2001 08:34:55 -0700, Paul Martz wrote: | | ... don't flame anyone on Usenet, one day you might end up working with | them. Good advice. I ignore it occasionally, but usually regret it. | One point of the argument that comes to mind was the PEX community bemoaning | the overhead of a function call per vertex, until it was you, Allen, I | believe, who described the trick of inlining the vertex call and trapping | bus accesses... I remember mentioning it a couple of times. The first published description I recall is in Kurt's article "The Silicon Graphics 4D/240GTX Superworkstation," in IEEE CG&A, July 1989. Allen (Please remove the random digit from my return address to reply to me.) >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!usc.edu!howland.erols.net!newspharm.inet.tele.dk.MISMATCH!news.tele.dk!195.158.233.21!news1.ebone.net!news.ebone.net!newsfeed.sunet.se!news01.sunet.se!news.chalmers.se!news.gu.se!not-for-mail Thu Mar 8 10:27:47 2001 From: "Marcus Lindblom" Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: Wed, 7 Mar 2001 23:27:27 +0100 Organization: Gothenburg University Lines: 29 Message-ID: <986cjr$l3d$1@news.gu.se> References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <3AA02BDF.46B91AAF@valinux.com> <97r23i$ksc$1@nyheter.chalmers.se> <9838k2$m2t$1@xmission.xmission.com> NNTP-Posting-Host: fizzgig.olofshojd.studenthem.gu.se Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 8bit X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.50.4133.2400 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400 Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!usc.edu!howland.erols.net!newspharm.inet.tele.dk.MISMATCH!news.tele.dk!195.158.233.21!news1.ebone.net!news.ebone.net!newsfeed.sunet.se!news01.sunet.se!news.chalmers.se!news.gu.se!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90793 "Rich" wrote in message news:9838k2$m2t$1@xmission.xmission.com... > [Please do not mail me a copy of your followup] > > "Marcus Lindblom" spake the secret code > <97r23i$ksc$1@nyheter.chalmers.se> thusly: > > >Define first 3d api. If you mean the first generic cross-platform api > >that has had a huge breakthrough, it's probably OpenGL. > > Define "huge breakthrough"; I suspect your definition is colored by > the recent history of PC graphics. Computer graphics as a > commercially viable technology predates the PC, probably by about 10 > years. It is. I really shouldn't be speculating like this, luckily I did say probably, so... :) > The first generic cross-platform API that was widely used was GKS, > the precursor to PHIGS/PHIGS-PLUS. Ah.. PHIGS.. it's all over Foley-vanDam .... that's the only place I've heard of it.. /Macke >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!newsxfer.eecs.umich.edu!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!sunqbc.risq.qc.ca!newsflash.concordia.ca!not-for-mail Thu Mar 8 10:27:47 2001 From: V-man Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: Wed, 7 Mar 2001 19:24:42 -0500 Organization: Concordia University, Montreal, Canada Lines: 28 Message-ID: References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <983k41$arf$1@xmission.xmission.com> <983qcp$l73$1@xmission.xmission.com> <985kf2$ecn$1@fcnews.fc.hp.com> NNTP-Posting-Host: alcor.concordia.ca Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII NNTP-Posting-User: v_melkon In-Reply-To: Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!newsxfer.eecs.umich.edu!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!sunqbc.risq.qc.ca!newsflash.concordia.ca!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90791 I would like to read that article. Is it on the net or some file format? V-man On Wed, 7 Mar 2001 akin1@pobox.com wrote: > On Wed, 7 Mar 2001 08:34:55 -0700, Paul Martz wrote: > | > | ... don't flame anyone on Usenet, one day you might end up working with > | them. > > Good advice. I ignore it occasionally, but usually regret it. > > | One point of the argument that comes to mind was the PEX community bemoaning > | the overhead of a function call per vertex, until it was you, Allen, I > | believe, who described the trick of inlining the vertex call and trapping > | bus accesses... > > I remember mentioning it a couple of times. The first published > description I recall is in Kurt's article "The Silicon Graphics > 4D/240GTX Superworkstation," in IEEE CG&A, July 1989. > > Allen > > (Please remove the random digit from my return address to reply to me.) > >From fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!logbridge.uoregon.edu!xmission!nnrp.xmission!not-for-mail Thu Mar 8 10:27:47 2001 From: legalize+jeeves@mail.xmission.com (Rich) Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: 7 Mar 2001 17:27:53 -0700 Organization: multi-cellular, biological Lines: 25 Message-ID: <986jm9$5q7$1@xmission.xmission.com> References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <985kf2$ecn$1@fcnews.fc.hp.com> Reply-To: (Rich) legalize+jeeves@mail.xmission.com NNTP-Posting-Host: xmission.xmission.com X-Trace: news.xmission.com 984011274 28345 198.60.22.20 (8 Mar 2001 00:27:54 GMT) X-Complaints-To: abuse@xmission.com NNTP-Posting-Date: 8 Mar 2001 00:27:54 GMT X-Reply-Etiquette: No copy by email, please Mail-Copies-To: never Path: fc.hp.com!news.cup.hp.com!hpb10302.boi.hp.com!hp-corv.cv.hp.com!sdd.hp.com!enews.sgi.com!logbridge.uoregon.edu!xmission!nnrp.xmission!not-for-mail Xref: fc.hp.com comp.graphics.api.opengl:90792 [Please do not mail me a copy of your followup] akin1@pobox.com spake the secret code thusly: >| One point of the argument that comes to mind was the PEX community bemoaning >| the overhead of a function call per vertex, until it was you, Allen, I >| believe, who described the trick of inlining the vertex call and trapping >| bus accesses... > >I remember mentioning it a couple of times. The first published >description I recall is in Kurt's article "The Silicon Graphics >4D/240GTX Superworkstation," in IEEE CG&A, July 1989. This is a very cool trick, but it certainly does through Purify and other memory leak detectors into a frenzy! OGL would initialize buffers by DMA and Purify would always report an uninitialized memory read by the client because it doesn't snoop DMA and it thought the buffer was uninitialized since it never saw a CPU instruction that initialized the buffer. -- Ask me about my upcoming book on Direct3D from Addison-Wesley! Direct3D Book Home Fractals! >From fc.hp.com!news Thu Mar 8 10:27:47 2001 Path: fc.hp.com!news From: "Paul Martz" Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: Thu, 8 Mar 2001 07:33:08 -0700 Organization: Hewlett-Packard Fort Collins Site Lines: 15 Approved: paul_martz@hp.com Message-ID: <988579$gh$1@fcnews.fc.hp.com> References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> <983k41$arf$1@xmission.xmission.com> <983qcp$l73$1@xmission.xmission.com> <985kf2$ecn$1@fcnews.fc.hp.com> NNTP-Posting-Host: fcpm87756.fc.hp.com X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.50.4133.2400 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400 Xref: fc.hp.com comp.graphics.api.opengl:90820 "V-man" wrote in message news:Pine.OSF.4.31.0103071924050.16822-100000@alcor.concordia.ca... > > I would like to read that article. Is it on the net or some file format? Allen was kind enough to forward it to me for inclusion in the FAQ. I'll try to have it online in the near future. -- -Paul Martz (paul_martz@hp.com) Hewlett Packard TCD Personal Workstations Lab OpenGL FAQ: http://www.opengl.org/developers/faqs/technical.html http://www.frii.com/~martz/oglfaq >From fc.hp.com!news Thu Mar 8 10:27:48 2001 Path: fc.hp.com!news From: "Paul Martz" Newsgroups: comp.graphics.api.opengl Subject: Re: history of OpenGL Date: Thu, 8 Mar 2001 07:36:40 -0700 Organization: Hewlett-Packard Fort Collins Site Lines: 11 Approved: paul_martz@hp.com Message-ID: <9885dt$i7$1@fcnews.fc.hp.com> References: <3AA01F4A.7F3442BA@caesar.sdsu.edu> NNTP-Posting-Host: fcpm87756.fc.hp.com X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.50.4133.2400 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400 Xref: fc.hp.com comp.graphics.api.opengl:90821 Can anyone recall the first public demonstration of OpenGL? I presume it was done on an SGI box, but which one? And how many other graphics vendors were in on the demo / announcement? -- -Paul Martz (paul_martz@hp.com) Hewlett Packard TCD Personal Workstations Lab OpenGL FAQ: http://www.opengl.org/developers/faqs/technical.html http://www.frii.com/~martz/oglfaq