Sim racing and new technologies

We have read over and over again that sim racing often demand high end *** machine. One reason for the GPL engine not being used for the next Nascar game is just that. But I was reading the last PC Gamer mag and got an idea. In an article, they mention a new technology called MRM (multi-resolution mesh). It's gonna be used in Team Fortress 2 game. What is does is change the level of details of characters depending on the range they are at. Closer characters are made of 800 polygons while the ones in the background use only 300 polygons. That allows the game to run a bit faster. Why not do the same thing with racing games? Given the fact that FPS often serve as benchmarks for frame rates and graphics, wouldn't be logical to take advantage of the same technologies? Philster

Good idea, but the technology is just becomming available. It won't help gpl any and since the upcomming (someday) release of the gpl based Nascar game will use the gpl code, I don't think it could be incorparated without a costly rewrite. Still, there is that dead TRANS AM game out there...anyone??

Quoted text (17 lines)

>We have read over and over again that sim racing often demand high end >*** machine. One reason for the GPL engine not being used for the >next Nascar game is just that. >But I was reading the last PC Gamer mag and got an idea. In an article, >they mention a new technology called MRM (multi-resolution mesh). It's >gonna be used in Team Fortress 2 game. What is does is change the level >of details of characters depending on the range they are at. Closer >characters are made of 800 polygons while the ones in the background use >only 300 polygons. That allows the game to run a bit faster. Why not do >the same thing with racing games? >Given the fact that FPS often serve as benchmarks for frame rates and >graphics, wouldn't be logical to take advantage of the same >technologies? >Philster

The problem is that the technology you refer to is for graphics. Graphics is not the problem with GPL, the physics of the AI opponents is the problem (with high processing demand). Would we want the AI opponents to be able to drive outside the laws of physics (that we have to abide by) when they are on the other side of the track or far down the road, but make them accurate only when they are close to us? I don't think this has any application to "simulations," but it is obviously great for reducing the graphics load in some situations. Marc.

Part of GPL's extensive hardware demands is that it is optimized for only two types of video cards - Rendition and 3dFX. Otherwise, the CPU is bogged down with having to both do the physics and then render it, which (to do at 36 fps) would require immense processing power. Most people (myself included) seem to be running processors at a MINIMUM of 350 MHz with a Rendition or 3dFX board to be able to get 36 fps. If they wrote a version of it for the Matrox G200 (I mean a GOOD version, not Matrox' lame patch) it would be faster than my Rendition. -- Kirk Lane Tempe, AZ ICQ: 28171652 RMRL #119 >We have read over and over again that sim racing often demand high end >*** machine. One reason for the GPL engine not being used for the >next Nascar game is just that. >But I was reading the last PC Gamer mag and got an idea. In an article, >they mention a new technology called MRM (multi-resolution mesh). It's >gonna be used in Team Fortress 2 game. What is does is change the level >of details of characters depending on the range they are at. Closer >characters are made of 800 polygons while the ones in the background use >only 300 polygons. That allows the game to run a bit faster. Why not do >the same thing with racing games? >Given the fact that FPS often serve as benchmarks for frame rates and >graphics, wouldn't be logical to take advantage of the same >technologies? >Philster

Quite right. The physics determining the dynamics are amazingly complex in GPL. We must also remember that this is the first sim that simulates dynamics from a 3d perspective IE the up / down (z axis) movements as well. Mark

Quoted text (15 lines)

> Part of GPL's extensive hardware demands is that it is optimized for only > two types of video cards - Rendition and 3dFX. Otherwise, the CPU is bogged > down with > having to both do the physics and then render it, which (to do at 36 fps) > would require immense processing power. Most people (myself included) seem > to be running processors at a MINIMUM of 350 MHz with a Rendition or 3dFX > board to be able to get 36 fps. If they wrote a version of it for the > Matrox G200 (I mean a GOOD version, not Matrox' lame patch) it would be > faster than my Rendition. > -- > Kirk Lane > Tempe, AZ > ICQ: 28171652 > RMRL #119

Hmmm... I'm not sure I understand your reply. I was not wondering why GPL was so CPU intensive, but rather how we could have great games with great graphics and still be able to run it on the average pc. Philster

Quoted text (17 lines)

> >We have read over and over again that sim racing often demand high end > >*** machine. One reason for the GPL engine not being used for the > >next Nascar game is just that. > >But I was reading the last PC Gamer mag and got an idea. In an article, > >they mention a new technology called MRM (multi-resolution mesh). It's > >gonna be used in Team Fortress 2 game. What is does is change the level > >of details of characters depending on the range they are at. Closer > >characters are made of 800 polygons while the ones in the background use > >only 300 polygons. That allows the game to run a bit faster. Why not do > >the same thing with racing games? > >Given the fact that FPS often serve as benchmarks for frame rates and > >graphics, wouldn't be logical to take advantage of the same > >technologies? > >Philster

Quoted text (9 lines)

> Quite right. The physics determining the dynamics are amazingly complex in > GPL. We must also remember that this is the first sim that simulates > dynamics from a 3d perspective IE the up / down (z axis) movements as well. > Mark > > The problem is that the technology you refer to is for graphics. Graphics > > is not the problem with GPL, the physics of the AI opponents is the

My question to you then: why do the frame rate like a rock when you change from 640X480 then to 800X600 (no to mention 1024X768)? Don't tell me the physics engine is tied to the graphics engine! Also, the details "slider" does have an impact on the frame rates too. Any comments? Philster

On Sun, 13 Jun 1999 15:24:47 GMT, "Marc Collins" >The problem is that the technology you refer to is for graphics. Graphics >is not the problem with GPL, the physics of the AI opponents is the problem >(with high processing demand). Would we want the AI opponents to be able to >drive outside the laws of physics (that we have to abide by) when they are >on the other side of the track or far down the road, but make them accurate >only when they are close to us? As someone who just recently upgraded from a Voodoo1 to a Voodoo3, I can attest that graphics is indeed a chokepoint for GPL! Martin Nigel Mansell RIP!

Should have worded that one better. Of course it is a fact that graphics make up a certain percentage of demand on the hardware required to run this sim. However you must consider the fact that this is in relative terms to previous sims, a much smaller proportion of all system demands in percentage terms. The extra computing power required to process all the additional data in an extremely complex physics engine literally blows the roof off lesser specified machines. Add to this, higher resolution and detail bias turned all the way up and you are raising the threshold even further. Your original post mentions MRM. This idea is an excellent one. I believe that something along the same lines has already been implemented in GPL, hence the slower framerates when running closer to other cars as opposed to running at a distance. Regards Mark > > Quite right. The physics determining the dynamics are amazingly complex in > > GPL. We must also remember that this is the first sim that simulates > > dynamics from a 3d perspective IE the up / down (z axis) movements as well. > > Mark > > > The problem is that the technology you refer to is for graphics. Graphics

Quoted text (8 lines)

> > > is not the problem with GPL, the physics of the AI opponents is the > My question to you then: why do the frame rate like a rock when you > change from 640X480 then to 800X600 (no to mention 1024X768)? Don't tell > me the physics engine is tied to the graphics engine! Also, the details > "slider" does have an impact on the frame rates too. Any comments? > Philster

On Sun, 13 Jun 1999 22:22:29 +0100, "Mark Seery" >Quite right. The physics determining the dynamics are amazingly complex in >GPL. We must also remember that this is the first sim that simulates >dynamics from a 3d perspective IE the up / down (z axis) movements as well. My cars in GP2 have been bouncing up and down for the last 3 years (GP2 released in summer '96). JoH ======================================== Jo Helsen EDP Operations BF Belgium ========================================

On Sun, 13 Jun 1999 10:37:23 -0400, Philster

Quoted text (15 lines)

>We have read over and over again that sim racing often demand high end >*** machine. One reason for the GPL engine not being used for the >next Nascar game is just that. >But I was reading the last PC Gamer mag and got an idea. In an article, >they mention a new technology called MRM (multi-resolution mesh). It's >gonna be used in Team Fortress 2 game. What is does is change the level >of details of characters depending on the range they are at. Closer >characters are made of 800 polygons while the ones in the background use >only 300 polygons. That allows the game to run a bit faster. Why not do >the same thing with racing games? >Given the fact that FPS often serve as benchmarks for frame rates and >graphics, wouldn't be logical to take advantage of the same >technologies?

Actually, in GP2 (now released 3 years ago and probably the last sim without 3D-card support) a similar principle was used. A last resort to make the thing playable...Cars at a distance are displayed using 2Dsprites. It was quite well implemented and only discovered when people released tools to change camera-angles: the sprites weren't adapted to these new angles and immediately became (much too) apparent...... JoH ======================================== Jo Helsen EDP Operations BF Belgium ========================================

Alas, just a canned effect in GP2. The vertical movement in GP2 is not modelled in terms of vehicle handling dynamics. Mark > On Sun, 13 Jun 1999 22:22:29 +0100, "Mark Seery" > >Quite right. The physics determining the dynamics are amazingly complex in > >GPL. We must also remember that this is the first sim that simulates > >dynamics from a 3d perspective IE the up / down (z axis) movements as well.

Quoted text (9 lines)

> My cars in GP2 have been bouncing up and down for the last 3 years > (GP2 released in summer '96). > JoH > ======================================== > Jo Helsen EDP Operations BF Belgium > ========================================

Simple: Make the average PC include a 3dFX board :) -- Kirk Lane Tempe, AZ ICQ: 28171652 RMRL #119

Well, I don't think you can talk in absolute, discrete terms. It's never black or white. The hight of the vertical movement in GP2 depends on the speed at least (just try touching wheels with another car at different speeds). There's also a difference between high curbs and low curbs etc. Remember: in a simulation everything is "canned" to some extent, even in GPL! To save processor power (they're not actually calculating the airflow-dynamics around the car, are they???) and also because our laws of physics are approximations after all... :-) JoH

Quoted text (24 lines)

>Alas, just a canned effect in GP2. The vertical movement in GP2 is not >modelled in terms of vehicle handling dynamics. >Mark >> On Sun, 13 Jun 1999 22:22:29 +0100, "Mark Seery" >> >Quite right. The physics determining the dynamics are amazingly complex >in >> >GPL. We must also remember that this is the first sim that simulates >> >dynamics from a 3d perspective IE the up / down (z axis) movements as >well. >> My cars in GP2 have been bouncing up and down for the last 3 years >> (GP2 released in summer '96). >> JoH >> ======================================== >> Jo Helsen EDP Operations BF Belgium >> ========================================

======================================== Jo Helsen EDP Operations BF Belgium ========================================