WSC

A part of the interview saying : WR: Most of these screenshots are from a Pentium II 450MHz with a Voodoo3 2000. The full-scene anti-aliased shots were taken on the same system, but with a Voodoo5 5500. The game runs at about 45 frames per second on that system, but no real optimizing has been done yet, so there's room for improvement. So is this only one car at the time or optimized for the Glide or just plain bull, coz if it isn't I'll have around 200 fps on my current system... -- Oliver

From what I remember(shaddup youse guys!!!) the West team started coding in glide, but last summer they were gonna completely swap over the video to D3d and or opengl... I can't quite recall if they were going to keep glide support in or not. HOWEVER. Anyone who thinks they'll get super speedy framerates with a PII 450 is drinking motor oil. One only has to take a look at how N4 gobbles cpus to imagine how much horsepower WSC will need. dave henrie "Slick" <[email hidden]> wrote in message news:97etg3$au1$[email hidden]...

Quoted text (9 lines)

> A part of the interview saying : > > WR: Most of these screenshots are from a Pentium II 450MHz with a Voodoo3 > 2000. The full-scene anti-aliased shots were taken on the same system, but > with a Voodoo5 5500. The game runs at about 45 frames per second on that > system, but no real optimizing has been done yet, so there's room for > improvement. > > So is this only one car at the time or optimized for the Glide or just

plain

Quoted text (10 lines)

> bull, coz if it isn't I'll have around 200 fps on my current system... > > > -- > > Oliver > > > >

On Tue, 27 Feb 2001 03:41:00 GMT, "Dave Henrie" <[email hidden]> wrote:

Quoted text (8 lines)

> From what I remember(shaddup youse guys!!!) the West team started >coding in glide, but last summer they were gonna completely swap over >the video to D3d and or opengl... I can't quite recall if they were going >to >keep glide support in or not. HOWEVER. Anyone who thinks they'll get >super speedy framerates with a PII 450 is drinking motor oil. One only has >to take a look at how N4 gobbles cpus to imagine how much horsepower >WSC will need.

Well, I'm sure it will be demanding, but I've seen some games where the graphics are astounding, but are also very efficient so it may not be as demanding as N4 and still look far better.

In article <97etg3$au1$[email hidden]>, Slick <[email hidden]> wrote: >So is this only one car at the time or optimized for the Glide or just plain >bull, coz if it isn't I'll have around 200 fps on my current system... I think I read somewhere that some or all of the shots we have seen this far have been taken without the actual physics engine code implemented. Once they get that monster working, I'm sure you can forget framerates like that on a P2-450 or anything on the market at the moment. Might be possible with a single car, but with many AI cars around using the same sophisticated physics engine, you are just going to need lots of CPU power. -Juha -- //[email hidden]

On Tue, 27 Feb 2001 03:41:00 GMT, "Dave Henrie" <[email hidden]> wrote:

Quoted text (9 lines)

> From what I remember(shaddup youse guys!!!) the West team started >coding in glide, but last summer they were gonna completely swap over >the video to D3d and or opengl... I can't quite recall if they were going >to >keep glide support in or not. HOWEVER. Anyone who thinks they'll get >super speedy framerates with a PII 450 is drinking motor oil. One only has >to take a look at how N4 gobbles cpus to imagine how much horsepower >WSC will need. >dave henrie

That doesn't say much, F1RC is graphically imo so much better looking than N4 (and any other sim for that matter), and while I still am struggling to keep a decent fps in N4 (I have to run in 800res, not more than 20 AI cars, and quite some details turned down) while F1RC, with -all- options at max, in 1024 res, runs superb, even at the start with 22 cars. On my PIII500 GFII. I suspect that the physics model of N4 is much more advanced than F1RC's, but I do think Papy's graphic engine could use a little improvement here and there. Andre "Michael feels the same way about safety in auto racing as I do. Great minds think alike." - David G Fisher

"Jagg" <[email hidden]> wrote in message news:[email hidden]... > On Tue, 27 Feb 2001 03:41:00 GMT, "Dave Henrie" <[email hidden]> > wrote: > > > From what I remember(shaddup youse guys!!!) the West team started > >coding in glide, but last summer they were gonna completely swap over > >the video to D3d and or opengl... I can't quite recall if they weregoing > >to > >keep glide support in or not. HOWEVER. Anyone who thinks they'll get > >super speedy framerates with a PII 450 is drinking motor oil. One onlyhas > >to take a look at how N4 gobbles cpus to imagine how much horsepower > >WSC will need. > > Well, I'm sure it will be demanding, but I've seen some games where > the graphics are astounding, but are also very efficient so it may not > be as demanding as N4 and still look far better. Exactly - GPL looks fantastic - and will run on VERY modest hardware, TOCA 2 looks nice, similar hardware req. Err...Pity about GP3 Doug

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 well, it's not only physics engine or graphic complexity taht makes the FPS, the programmer does it also, if you take for example the PSX games (or N64, or DC, or whatever game system) the hardware is ALWAYS the same and the games are getting better, more complicated, more fps, more realistic. My personal idea is that the programmers who had experience with fixed systems ("old" computers, ATARI ST, AMIGA, consoles) are much better than the "PC" programmers who just hope for the customer to buy a bigger, faster computer to run the game. Sometimes I wonder about the POWER we have in our machines, do you realize what a 1G CPU means ? Do you realize what 128 Megs rams are ? Do you have any idea what the programmers of SEGA could do with a dreamcast with a 1G cpu and 128Megs rams, not to mention a geforce2 ? I think that PC programmers are a bit "sloppy" and do NOT try at all to optimize the code, maybe that is what the west brothers are doing, after all GP2 was running fine (or sort of) with a cyrix 200 and 32Megs, Personally I think that GP3 doesn't NEED the upgrade in compuiong power from GP2 hen you look at the difference in the game itself DIscussion opened "Andre Warringa" <[email hidden]> a �crit dans le message news: [email hidden]...

Quoted text (30 lines)

> On Tue, 27 Feb 2001 03:41:00 GMT, "Dave Henrie" <[email hidden]> > wrote: > > > From what I remember(shaddup youse guys!!!) the West team > > started > >coding in glide, but last summer they were gonna completely swap > >over the video to D3d and or opengl... I can't quite recall if > >they were going to > >keep glide support in or not. HOWEVER. Anyone who thinks they'll > >get super speedy framerates with a PII 450 is drinking motor oil. > >One only has to take a look at how N4 gobbles cpus to imagine how > >much horsepower WSC will need. > >dave henrie > > That doesn't say much, F1RC is graphically imo so much better > looking than N4 (and any other sim for that matter), and while I > still am > struggling to keep a decent fps in N4 (I have to run in 800res, not > more than 20 AI cars, and quite some details turned down) while > F1RC, with -all- options at max, in 1024 res, runs superb, even at > the start with 22 cars. On my PIII500 GFII. > I suspect that the physics model of N4 is much more advanced than > F1RC's, but I do think Papy's graphic engine could use a little > improvement here and there. > > > Andre > > "Michael feels the same way about safety in auto racing as I do. > Great minds think alike." - David G Fisher

-----BEGIN PGP SIGNATURE----- Version: PGPfreeware 7.0.3 for non-commercial use <http://www.pgp.com> iQA/AwUBOpuezghv7FFac7/nEQKDawCgseRXyWB2ugriQsDPVlNQWdyY3scAn3lW 6WcMXP2ZUGuAJV6oocXhWRTH =16WW -----END PGP SIGNATURE-----

>"Dave Henrie" <[email hidden]> wrote > HOWEVER. Anyone who thinks they'll get > super speedy framerates with a PII 450 is drinking motor oil. One onlyhas > to take a look at how N4 gobbles cpus to imagine how much horsepower > WSC will need. Papyrus has never been known to be great code optimisers. Most of their titles are aimed to run well on future PCs. Just do something, load the original Nascar Racing. Bet you will struggle even on a Pentium 3 in SVGA with every graphic in heavy traffic. I'm not kidding. The only title by Papyrus who could stand very well on current PC when released was Nascar Racing 2, imho. Check out F1RC to see a current software who looks great and runs quite fast on normal PC's. Same for NFS:PU. It's all a matter of how they optimise the code. Of course I expect the WSC standards to change quite a bit before it's release, but I don't expect that it won't run well unless you have a 5GHZ PC and a GeForce5 ;-) -- -- Fran�ois M�nard <ymenard> -- May the Downforce be with you... -- http://www.ymenard.com/ -- People think it must be fun to be a genius, but they don't realise how hard it is to put up with all the idiots in the world.

>"Txl" <[email hidden]> wrote > I think that PC programmers are a bit "sloppy" and do NOT try at all > to optimize the code Let's remember that PC programmers aren't helped by MS's poor DirectX Hardware integration. They go as fast as MS makes it possible, which is quite low compared to proprietary console hardware. Just watch for the X-box, you'll see many buggy software. -- -- Fran�ois M�nard <ymenard> -- May the Downforce be with you... -- http://www.ymenard.com/ -- People think it must be fun to be a genius, but they don't realise how hard it is to put up with all the idiots in the world.

On Tue, 27 Feb 2001 18:53:50 GMT, Jagg <[email hidden]> wrote:

Quoted text (11 lines)

>On Tue, 27 Feb 2001 14:33:33 GMT, [email hidden] (Andre >Warringa) wrote: > > >>F1RC -is- a good example, this game is graphically so much more >>advanced than any other racing sim, yet with a full field it runs >>better than -any- other racing sim I own.. and that includes 2-year >>old GPL.. > >True. But, does it also have a sophisticated physics model? I think >not.

Perhaps you missed my previous post in this thread: That doesn't say much, F1RC is graphically imo so much better looking than N4 (and any other sim for that matter), and while I still am struggling to keep a decent fps in N4 (I have to run in 800res, not more than 20 AI cars, and quite some details turned down) while F1RC, with -all- options at max, in 1024 res, runs superb, even at the start with 22 cars. On my PIII500 GFII. I suspect that the physics model of N4 is much more advanced than F1RC's, but I do think Papy's graphic engine could use a little improvement here and there.

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 In a computer the physics model means nothing in terms of computing power, think that your system can do 1,000,000,000 cycles per second, so to compute a mere 256 operations per second which is what the physic model of GPL is like (and we know it's very advanced you have about 1G/256 cycles of computer power to do 1 GPL "frame".... that's a lot, besides that the physics model of GP1on my trusted AtariST with a 6800 8 bits processor running at a fantastic 2.8 Mhz was quite okay, the gfx were crap by today's standards but the physics were okay at the time... Botom line is physics is a piece of cake for computer, gfx is more complicated. "Jagg" <[email hidden]> a �crit dans le message news: [email hidden]...

Quoted text (10 lines)

> On Tue, 27 Feb 2001 14:33:33 GMT, [email hidden] (Andre > Warringa) wrote: > > > >F1RC -is- a good example, this game is graphically so much more > >advanced than any other racing sim, yet with a full field it runs > >better than -any- other racing sim I own.. and that includes > >2-year old GPL.. > > True. But, does it also have a sophisticated physics model? I think

not. -----BEGIN PGP SIGNATURE----- Version: PGPfreeware 7.0.3 for non-commercial use <http://www.pgp.com> iQA/AwUBOpwqEQhv7FFac7/nEQItiACfdzHrTnJiayWi78kVt2NtoBHQtPQAoL+l gUdqDGuGp33ZGENOjwpLinFD =bzmZ -----END PGP SIGNATURE-----

Be careful in how you define "cycles". We've all heard of the mythical "188 Hz" GPL physics engine which does these 188 (I think that's the number) separate calculations per second in its physics engine. That does NOT translate into 188 CPU cycles per second. Each of those 188 calculations is potentially hundreds of lines of code, corresponding to thousands of CPU cycles each. So each second's worth of physics could require (say) 150,000 CPU cycles. I have seen articles proposing PPU's, or Physics Processing Units, which could actually do an "angular momentum" or "rigid body rotation" calculation in one cycle. If those every see the light of day then a 188Hz physics engine really could translate to 188 cycles. "Txl" <[email hidden]> wrote in message news:vSVm6.6603$[email hidden]...

Quoted text (42 lines)

> -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > In a computer the physics model means nothing in terms of computing > power, think that your system can do 1,000,000,000 cycles per second, > so to compute a mere 256 operations per second which is what the > physic model of GPL is like (and we know it's very advanced you have > about 1G/256 cycles of computer power to do 1 GPL "frame".... > > that's a lot, besides that the physics model of GP1on my trusted > AtariST with a 6800 8 bits processor running at a fantastic 2.8 Mhz > was quite okay, the gfx were crap by today's standards but the > physics were okay at the time... > > Botom line is physics is a piece of cake for computer, gfx is more > complicated. > > > "Jagg" <[email hidden]> a �crit dans le message news: > [email hidden]... > > On Tue, 27 Feb 2001 14:33:33 GMT, [email hidden] (Andre > > Warringa) wrote: > > > > > > >F1RC -is- a good example, this game is graphically so much more > > >advanced than any other racing sim, yet with a full field it runs > > >better than -any- other racing sim I own.. and that includes > > >2-year old GPL.. > > > > True. But, does it also have a sophisticated physics model? I think > not. > > -----BEGIN PGP SIGNATURE----- > Version: PGPfreeware 7.0.3 for non-commercial use <http://www.pgp.com> > > iQA/AwUBOpwqEQhv7FFac7/nEQItiACfdzHrTnJiayWi78kVt2NtoBHQtPQAoL+l > gUdqDGuGp33ZGENOjwpLinFD > =bzmZ > -----END PGP SIGNATURE----- > > >

Afaik, GPL's physics are running at 288 Hz. ed_ Rob Adams wrote: > Be careful in how you define "cycles". We've all heard of the mythical"188 > Hz" GPL physics engine which does these 188 (I think that's the number) > separate calculations per second in its physics engine. That does NOT > translate into 188 CPU cycles per second. Each of those 188 calculationsis > potentially hundreds of lines of code, corresponding to thousands of CPU > cycles each. So each second's worth of physics could require (say) 150,000 > CPU cycles. > > I have seen articles proposing PPU's, or Physics Processing Units, which > could actually do an "angular momentum" or "rigid body rotation"calculation

Quoted text (49 lines)

> in one cycle. If those every see the light of day then a 188Hz physics > engine really could translate to 188 cycles. > > "Txl" <[email hidden]> wrote in message > news:vSVm6.6603$[email hidden]... > > -----BEGIN PGP SIGNED MESSAGE----- > > Hash: SHA1 > > > > In a computer the physics model means nothing in terms of computing > > power, think that your system can do 1,000,000,000 cycles per second, > > so to compute a mere 256 operations per second which is what the > > physic model of GPL is like (and we know it's very advanced you have > > about 1G/256 cycles of computer power to do 1 GPL "frame".... > > > > that's a lot, besides that the physics model of GP1on my trusted > > AtariST with a 6800 8 bits processor running at a fantastic 2.8 Mhz > > was quite okay, the gfx were crap by today's standards but the > > physics were okay at the time... > > > > Botom line is physics is a piece of cake for computer, gfx is more > > complicated. > > > > > > "Jagg" <[email hidden]> a �crit dans le message news: > > [email hidden]... > > > On Tue, 27 Feb 2001 14:33:33 GMT, [email hidden] (Andre > > > Warringa) wrote: > > > > > > > > > >F1RC -is- a good example, this game is graphically so much more > > > >advanced than any other racing sim, yet with a full field it runs > > > >better than -any- other racing sim I own.. and that includes > > > >2-year old GPL.. > > > > > > True. But, does it also have a sophisticated physics model? I think > > not. > > > > -----BEGIN PGP SIGNATURE----- > > Version: PGPfreeware 7.0.3 for non-commercial use <http://www.pgp.com> > > > > iQA/AwUBOpwqEQhv7FFac7/nEQItiACfdzHrTnJiayWi78kVt2NtoBHQtPQAoL+l > > gUdqDGuGp33ZGENOjwpLinFD > > =bzmZ > > -----END PGP SIGNATURE----- > > > > > > > >

As far as I know, the physics engine wasn't implemented at the time those screen shots were taken. I have the feeling that once they incorporate those calculations, the frame rates will begin to drop. Again, I don't really know anything for sure but, that has been the speculation that I've heard on some newsgroups. Jah_Warrior86

Thanks, I knew there was an "88" in there somewhere :) "Ed Solheim" <[email hidden]> wrote in message news:YuXm6.4171$[email hidden]...

Quoted text (13 lines)

> Afaik, GPL's physics are running at 288 Hz. > > ed_ > > Rob Adams wrote: > > Be careful in how you define "cycles". We've all heard of the mythical > "188 > > Hz" GPL physics engine which does these 188 (I think that's the number) > > separate calculations per second in its physics engine. That does NOT > > translate into 188 CPU cycles per second. Each of those 188 calculations > is > > potentially hundreds of lines of code, corresponding to thousands of CPU > > cycles each. So each second's worth of physics could require (say)

150,000

Quoted text (56 lines)

> > CPU cycles. > > > > I have seen articles proposing PPU's, or Physics Processing Units, which > > could actually do an "angular momentum" or "rigid body rotation" > calculation > > in one cycle. If those every see the light of day then a 188Hz physics > > engine really could translate to 188 cycles. > > > > "Txl" <[email hidden]> wrote in message > > news:vSVm6.6603$[email hidden]... > > > -----BEGIN PGP SIGNED MESSAGE----- > > > Hash: SHA1 > > > > > > In a computer the physics model means nothing in terms of computing > > > power, think that your system can do 1,000,000,000 cycles per second, > > > so to compute a mere 256 operations per second which is what the > > > physic model of GPL is like (and we know it's very advanced you have > > > about 1G/256 cycles of computer power to do 1 GPL "frame".... > > > > > > that's a lot, besides that the physics model of GP1on my trusted > > > AtariST with a 6800 8 bits processor running at a fantastic 2.8 Mhz > > > was quite okay, the gfx were crap by today's standards but the > > > physics were okay at the time... > > > > > > Botom line is physics is a piece of cake for computer, gfx is more > > > complicated. > > > > > > > > > "Jagg" <[email hidden]> a �crit dans le message news: > > > [email hidden]... > > > > On Tue, 27 Feb 2001 14:33:33 GMT, [email hidden] (Andre > > > > Warringa) wrote: > > > > > > > > > > > > >F1RC -is- a good example, this game is graphically so much more > > > > >advanced than any other racing sim, yet with a full field it runs > > > > >better than -any- other racing sim I own.. and that includes > > > > >2-year old GPL.. > > > > > > > > True. But, does it also have a sophisticated physics model? I think > > > not. > > > > > > -----BEGIN PGP SIGNATURE----- > > > Version: PGPfreeware 7.0.3 for non-commercial use <http://www.pgp.com> > > > > > > iQA/AwUBOpwqEQhv7FFac7/nEQItiACfdzHrTnJiayWi78kVt2NtoBHQtPQAoL+l > > > gUdqDGuGp33ZGENOjwpLinFD > > > =bzmZ > > > -----END PGP SIGNATURE----- > > > > > > > > > > > > > > >

On Tue, 27 Feb 2001 20:54:32 GMT, "ymenard" <[email hidden]> wrote: >Let's remember that PC programmers aren't helped by MS's poor DirectX >Hardware integration. They go as fast as MS makes it possible, which is >quite low compared to proprietary console hardware. Just watch for the >X-box, you'll see many buggy software. Console programmers use Assembly a lot more too.

On Tue, 27 Feb 2001 20:58:20 GMT, Andre Warringa <[email hidden]> wrote:

Quoted text (11 lines)

>Perhaps you missed my previous post in this thread: > >That doesn't say much, F1RC is graphically imo so much better looking >than N4 (and any other sim for that matter), and while I still am >struggling to keep a decent fps in N4 (I have to run in 800res, not >more than 20 AI cars, and quite some details turned down) while F1RC, >with -all- options at max, in 1024 res, runs superb, even at the start >with 22 cars. On my PIII500 GFII. >I suspect that the physics model of N4 is much more advanced than >F1RC's, but I do think Papy's graphic engine could use a little >improvement here and there.

But, that's what I'm getting at. Maybe F1RC isn't putting much of a strain on the cpu because it lacks the sophisticated physics model. I won't know that for a fact until it's released. Just suggesting it. Most arcade racers look good and run fast because they don't have to use cpu calculations on the physics model.

On Tue, 27 Feb 2001 22:28:43 GMT, "Txl" <[email hidden]> wrote: >Botom line is physics is a piece of cake for computer, gfx is more >complicated. The add-in AI, crash detection and so on and I think it does impact the cpu resources a fair amount.

On Tue, 27 Feb 2001 22:28:43 GMT, "Txl" <[email hidden]> wrote: >that's a lot, besides that the physics model of GP1on my trusted >AtariST with a 6800 8 bits processor running at a fantastic 2.8 Mhz >was quite okay, the gfx were crap by today's standards but the >physics were okay at the time... I sure hope your Atari ST had a Motorola 68000 16/32-bit processor running at 8MHz. Unless it was a "special" French model... :) "But in a way, fear is a big part of racing, because if there was nothing to be frightened of, and no limit, any fool could get into a motor car and racing would not exist as a sport." -- Jim Clark

>"Jagg" <[email hidden]> wrote > Console programmers use Assembly a lot more too. Geoff Crammond use it still, and we know how great GP3 was <VBG> -- -- Fran�ois M�nard <ymenard> -- May the Downforce be with you... -- http://www.ymenard.com/ -- People think it must be fun to be a genius, but they don't realise how hard it is to put up with all the idiots in the world.

The physics model (as it was then) was implemented. The tire model had yet to be put in. Both have undergone significant updates since those screenshots and movies were taken. Shane ~Jah_Warrior86~ <[email hidden]> wrote in message news:3DXm6.121016$[email hidden]... > As far as I know, the physics engine wasn't implemented at the time those > screen shots were taken. I have the feeling that once they incorporatethose > calculations, the frame rates will begin to drop. Again, I don't reallyknow > anything for sure but, that has been the speculation that I've heard onsome > newsgroups. > > Jah_Warrior86 > >

Ubi Soft sent you a review copy too? David G Fisher "Andre Warringa" <[email hidden]> wrote in message news:[email hidden]...

Quoted text (25 lines)

> On Tue, 27 Feb 2001 18:53:50 GMT, Jagg <[email hidden]> wrote: > > >On Tue, 27 Feb 2001 14:33:33 GMT, [email hidden] (Andre > >Warringa) wrote: > > > > > >>F1RC -is- a good example, this game is graphically so much more > >>advanced than any other racing sim, yet with a full field it runs > >>better than -any- other racing sim I own.. and that includes 2-year > >>old GPL.. > > > >True. But, does it also have a sophisticated physics model? I think > >not. > > Perhaps you missed my previous post in this thread: > > That doesn't say much, F1RC is graphically imo so much better looking > than N4 (and any other sim for that matter), and while I still am > struggling to keep a decent fps in N4 (I have to run in 800res, not > more than 20 AI cars, and quite some details turned down) while F1RC, > with -all- options at max, in 1024 res, runs superb, even at the start > with 22 cars. On my PIII500 GFII. > I suspect that the physics model of N4 is much more advanced than > F1RC's, but I do think Papy's graphic engine could use a little > improvement here and there.

On Wed, 28 Feb 2001 03:45:52 GMT, Jagg <[email hidden]> wrote:

Quoted text (21 lines)

>On Tue, 27 Feb 2001 20:58:20 GMT, Andre Warringa <[email hidden]> >wrote: > > >>Perhaps you missed my previous post in this thread: >> >>That doesn't say much, F1RC is graphically imo so much better looking >>than N4 (and any other sim for that matter), and while I still am >>struggling to keep a decent fps in N4 (I have to run in 800res, not >>more than 20 AI cars, and quite some details turned down) while F1RC, >>with -all- options at max, in 1024 res, runs superb, even at the start >>with 22 cars. On my PIII500 GFII. >>I suspect that the physics model of N4 is much more advanced than >>F1RC's, but I do think Papy's graphic engine could use a little >>improvement here and there. > >But, that's what I'm getting at. Maybe F1RC isn't putting much of a >strain on the cpu because it lacks the sophisticated physics model. >I won't know that for a fact until it's released. Just suggesting it. >Most arcade racers look good and run fast because they don't have to >use cpu calculations on the physics model.

Allthough the physics model isn't as good as N4, I wonder if a physics model like N4 has really has such a major impact on the graphics.. Sure you have to cut on graphics when the cpu is mostly used for calculating physics, but in another thread I was talking about how programmes cranked out everything possible on computers like the C64 and the Amiga. Today programmers seem to become lazier.. it doesn't run good? Hey, just upgrade your pc! The graphics of F1RC are so much better than N4's graphics, I think Papy would love to get their hands on the graphics engine of F1RC. Andre "Michael feels the same way about safety in auto racing as I do. Great minds think alike." - David G Fisher

On Tue, 27 Feb 2001 18:50:51 GMT, Jagg <[email hidden]> wrote:

Quoted text (11 lines)

>On 27 Feb 2001 14:56:14 GMT, [email hidden] (Juha Kallioinen) >wrote: > > >>I don't know if they are going to use the same engine for AI, I too doubt >>it, but even if they use a simplified physics model, it doesn't necessarily >>mean that it is a 'light' model. > >Doesn't N4 and NH use the same physics model for all cars? I don't >like the idea of racing against cars that don't have the same >limitations as you do.

I'm quite sure the AI uses a dumbed down physics model. If all 42 AI cars do use the same model as the player, ok, then I can understand the simple graphics of N4 :) Andre "Michael feels the same way about safety in auto racing as I do. Great minds think alike." - David G Fisher

On Wed, 28 Feb 2001 08:43:35 GMT, [email hidden] (Andre Warringa) wrote: >but in another thread I was talking about how >programmes cranked out everything possible on computers like the C64 >and the Amiga. Yes, and they were *very* simple graphics and physics engines by todays standards. I don't know what the real deal is. I just know that most arcade games have great eye candy, run fast, but really skimp on everything else. I don know that developers of hardcore flight sims say all the math calculations take up a lot of cpu resources.

On Wed, 28 Feb 2001 03:54:43 GMT, "ymenard" <[email hidden]> wrote: >Geoff Crammond use it still, and we know how great GP3 was <VBG> Wel, it does run smoothly even on lower end machines. :-)

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 I know, that's why I said GPL "FRAME" and not "computer cycly or something like that, anyway the point is that the physics is NOT the cpu power eater, it's the graphics... "Rob Adams" <[email hidden]> a �crit dans le message news: _nWm6.1111$[email hidden]...

Quoted text (62 lines)

> Be careful in how you define "cycles". We've all heard of the > mythical "188 Hz" GPL physics engine which does these 188 (I think > that's the number) separate calculations per second in its physics > engine. That does NOT translate into 188 CPU cycles per second. > Each of those 188 calculations is potentially hundreds of lines of > code, corresponding to thousands of CPU cycles each. So each > second's worth of physics could require (say) 150,000 CPU cycles. > > I have seen articles proposing PPU's, or Physics Processing Units, > which could actually do an "angular momentum" or "rigid body > rotation" calculation in one cycle. If those every see the light of > day then a 188Hz physics engine really could translate to 188 > cycles. > > "Txl" <[email hidden]> wrote in message > news:vSVm6.6603$[email hidden]... > > -----BEGIN PGP SIGNED MESSAGE----- > > Hash: SHA1 > > > > In a computer the physics model means nothing in terms of > > computing power, think that your system can do 1,000,000,000 > > cycles per second, so to compute a mere 256 operations per second > > which is what the physic model of GPL is like (and we know it's > > very advanced you have about 1G/256 cycles of computer power to > > do 1 GPL "frame".... > > > > that's a lot, besides that the physics model of GP1on my trusted > > AtariST with a 6800 8 bits processor running at a fantastic 2.8 > > Mhz was quite okay, the gfx were crap by today's standards but > > the physics were okay at the time... > > > > Botom line is physics is a piece of cake for computer, gfx is > > more complicated. > > > > > > "Jagg" <[email hidden]> a �crit dans le message news: > > [email hidden]... > > > On Tue, 27 Feb 2001 14:33:33 GMT, [email hidden] (Andre > > > Warringa) wrote: > > > > > > > > > >F1RC -is- a good example, this game is graphically so much > > > >more advanced than any other racing sim, yet with a full field > > > >it runs better than -any- other racing sim I own.. and that > > > >includes 2-year old GPL.. > > > > > > True. But, does it also have a sophisticated physics model? I > > > think > > not. > > > > -----BEGIN PGP SIGNATURE----- > > Version: PGPfreeware 7.0.3 for non-commercial use > > <http://www.pgp.com> > > > > iQA/AwUBOpwqEQhv7FFac7/nEQItiACfdzHrTnJiayWi78kVt2NtoBHQtPQAoL+l > > gUdqDGuGp33ZGENOjwpLinFD > > =bzmZ > > -----END PGP SIGNATURE----- > > > > > > >

-----BEGIN PGP SIGNATURE----- Version: PGPfreeware 7.0.3 for non-commercial use <http://www.pgp.com> iQA/AwUBOpzKAwhv7FFac7/nEQLbNQCg0wa+KUZMeDFjL4qlYj7khBIGp28AoKdy PrbXyOhxZ4mr/XPbrKv0wFk+ =Ip43 -----END PGP SIGNATURE-----

Why isn't the physics model as good as N4? Do you have some facts or is this just an opinion? David G Fisher "I think Dave Fisher is a true genius, while Andre Warringa really is a bit of a twat." - Michael Schumacher "Andre Warringa" <[email hidden]> wrote in message news:[email hidden]...

Quoted text (9 lines)

> On Wed, 28 Feb 2001 03:45:52 GMT, Jagg <[email hidden]> wrote: > > >On Tue, 27 Feb 2001 20:58:20 GMT, Andre Warringa <[email hidden]> > >wrote: > Allthough the physics model isn't as good as N4, I wonder if a physics > model like N4 has really has such a major impact on the graphics.. > Andre > > "Michael feels the same way about safety in auto racing as I do. Great

minds > think alike." - David G Fisher

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 not that much, really, PC programmpers are just "lazy" (of course I could NEVER do what they do but still) and instead of them optimizing their code it is us (the buyers) that do this optimization for them, by buying bigger and faster machines. Check out the PSX1 games that are released now, on a 4 year old system, when the best computer you could get was a P133 (maybe) with 32Megs and a PCI video with 2 megs.... You will see a GREAT improvement in the quality both physics and graphics of the game in the PSX (wipeout vs wipeout3 for example) and a fantastic DETERIORATION of the PC (GP1 vs GP3 if you compare the games running on similar systems). But i'm still VERY interested on F1RC...we'll see, hopefully since the guys at UBISOFT do also a LOT of console games (incl. RAYMAN) it is more in the "culture" of the company to optimize code whereas papyrus (or sierra) just hope for us to purchase another extra 300 Mhz of CPU speed. "Jagg" <[email hidden]> a �crit dans le message news: [email hidden]...

Quoted text (9 lines)

> On Tue, 27 Feb 2001 22:28:43 GMT, "Txl" <[email hidden]> wrote: > > > >Botom line is physics is a piece of cake for computer, gfx is more > >complicated. > > The add-in AI, crash detection and so on and I think it does impact > the cpu resources a fair amount. >

-----BEGIN PGP SIGNATURE----- Version: PGPfreeware 7.0.3 for non-commercial use <http://www.pgp.com> iQA/AwUBOpzLawhv7FFac7/nEQLVGgCgooVkG0WnHh9/q60uOWEbgdDOJk0AoMHM MNkDg8jJjNk/GKpvSOKFeMwD =EPlM -----END PGP SIGNATURE-----

On Wed, 28 Feb 2001 09:55:19 GMT, "David G Fisher" <[email hidden]> wrote: >Why isn't the physics model as good as N4? Do you have some facts or is this >just an opinion? It's just an opinion. What leads me to believe that the physics model of F1RC is not quite accurate is this: The tracks of F1RC are very, very close to the real tracks right? Ok, yesterday I did a 1:26 at Melbourne. That's what, 10 seconds faster than Schumi? That doesn't take away the fact that is simply big fun driving in F1RC. Problem is ofcourse that the AI is the worst I've ever seen in any computer game, unless they made some BIGTIME chances on the last days, and I heard from a couple of people that the online play via the net sucks bigtime as well, because there is no prediction model. That would make F1RC a great hotlapping game, unfortunately. >David G Fisher >"I think Dave Fisher is a true genius, while Andre Warringa really is a bit >of a twat." - Michael Schumacher I love you too David :) Andre "The physics model in the final version of 4x4 Evolution will me much better than it is in the demo" - David G Fisher

Ruud van Gaal wrote:

Quoted text (19 lines)

> > On Tue, 27 Feb 2001 13:51:06 -0000, "Tony Whitley" <t o n y . w h i t > l e y @ nStPlAwMorld.com> wrote: > > ... > >Oh! How quickly they forget. Weren't you around when GPL came out? I > >certainly upgraded my PC to run it all, and then again to get it to run > >well. I still don't have all the options on now... I think that it's stupid > >to have a physics engine hogging the processor - it must be trivial to scale > >it down, for example, by changing the update rate (288 Hz in the case of GPL > >I'm told). > > Hehe, you don't want to *know* what downsizing the update rate does > sometimes in my sim. ;-) > Into deep space! > > Ruud van Gaal, GPL Rank +53.25 > Pencil art : http://www.marketgraph.nl/gallery/ > Car simulation: http://www.marketgraph.nl/gallery/racer/

Hehe, I can imagine. Probably the GPL engine is already at the absolute satisfactory minimum, considering the forces involved. -Gregor

On Tue, 27 Feb 2001 23:04:26 GMT, "Rob Adams" <[email hidden]> wrote: ... >I have seen articles proposing PPU's, or Physics Processing Units, which >could actually do an "angular momentum" or "rigid body rotation" calculation >in one cycle. If those every see the light of day then a 188Hz physics >engine really could translate to 188 cycles. Even then, you'd have multiple physics calculations; there's a lot more surrounding these RRB rotation commands. Matrix calculations in 1 cycle could help very much. And a Pacejka formula blaster, hehe, that takes IIRC 2 sin() and 2 arctan calls. For a single tire, and you need 3 of those formulas; for lateral, longitudinal and aligning moment (for your force feedback). Ofcourse these days FPU's seem to give sin() and such away for nearly free! Ruud van Gaal, GPL Rank +53.25 Pencil art : http://www.marketgraph.nl/gallery/ Car simulation: http://www.marketgraph.nl/gallery/racer/

"Ruud van Gaal" <[email hidden]> wrote ... > On Tue, 27 Feb 2001 23:04:26 GMT, "Rob Adams" <[email hidden]> > wrote: > > ... > >I have seen articles proposing PPU's, or Physics Processing Units, which > >could actually do an "angular momentum" or "rigid body rotation"calculation

Quoted text (12 lines)

> >in one cycle. If those every see the light of day then a 188Hz physics > >engine really could translate to 188 cycles. > > Even then, you'd have multiple physics calculations; there's a lot > more surrounding these RRB rotation commands. > Matrix calculations in 1 cycle could help very much. > And a Pacejka formula blaster, hehe, that takes IIRC 2 sin() and 2 > arctan calls. For a single tire, and you need 3 of those formulas; for > lateral, longitudinal and aligning moment (for your force feedback). > > Ofcourse these days FPU's seem to give sin() and such away for nearly > free!

And of course you could always create a table of sines and arctans and just look them up rather than calculating them every time... And if you're using floating point arithmetic all the time no wonder the cpu cycles are swallowed up. -- Tony Whitley GPLRank +9.99 "On some laps I'd get as many as fifteen corners really right, but Moss, he gets them all right on every lap" - Von Trips after the 1961 German Grand Prix.

David G Fisher wrote something: Welcome home, Mr. Kane!

On Wed, 28 Feb 2001 13:29:42 +0100, Gregor Veble <[email hidden]> wrote: >Ruud van Gaal wrote:... >> Hehe, you don't want to *know* what downsizing the update rate does >> sometimes in my sim. ;-) >> Into deep space! >> >Hehe, I can imagine. Probably the GPL engine is already at the absolute >satisfactory minimum, considering the forces involved. I would say so; at 10ms intervals (100Hz), my sim works mostly but a bit choppy. At 1000Hz it works like a charm, but given the GPL rate I'm currently sticking at 250Hz, which works quite ok-ish. Except for the suspension it seems; rather the tire rate, which is much higher than the suspension itself. Will have to dive into that to get rid of the funny effect that falling down means the car will bounce up harder than it went down! ;-) Ruud van Gaal, GPL Rank +53.25 Pencil art : http://www.marketgraph.nl/gallery/ Car simulation: http://www.marketgraph.nl/gallery/racer/

On Wed, 28 Feb 2001 16:24:14 -0000, "Tony Whitley" <t o n y . w h i t l e y @ nStPlAwMorld.com> wrote: > >"Ruud van Gaal" <[email hidden]> wrote ... >> On Tue, 27 Feb 2001 23:04:26 GMT, "Rob Adams" <[email hidden]> >> wrote:... >> Ofcourse these days FPU's seem to give sin() and such away for nearly >> free! >And of course you could always create a table of sines and arctans and just >look them up rather than calculating them every time... Hm, I'd rather have more precision for now than create tables yet. > And if you're using >floating point arithmetic all the time no wonder the cpu cycles are >swallowed up. Ah no, floating point is for free these days. Compares the same as integer stuff almost (considering you'd have to do fixed point arithmetic). As a note, the physics take up a measly 2% (roughly calculated) currently in my sim. The main part is graphics. So if I optimized it to 1 cycle, I still would have gained but 2% speed increase. So I won't, for now. Ruud van Gaal, GPL Rank +53.25 Pencil art : http://www.marketgraph.nl/gallery/ Car simulation: http://www.marketgraph.nl/gallery/racer/

"Ruud van Gaal" <[email hidden]> wrote ... <snip>

Quoted text (7 lines)

> > And if you're using > >floating point arithmetic all the time no wonder the cpu cycles are > >swallowed up. > > Ah no, floating point is for free these days. Compares the same as > integer stuff almost (considering you'd have to do fixed point > arithmetic).

Really? I'm an 8/16 bit embedded programmer, I didn't realise you Pentium guys had it so easy :-) > As a note, the physics take up a measly 2% (roughly calculated) > currently in my sim. The main part is graphics. So if I optimized it > to 1 cycle, I still would have gained but 2% speed increase. So I > won't, for now.Of course, you have to identify /where/ to optimise. 2% does seem to contradict a lot of the stuff said on this ng (and in this thread for that matter). I believe your figures - you're actually doing it. And of course you're following rule #1 of software development: "Get it working first, then try to optimise it". -- Tony Whitley GPLRank +9.99 "On some laps I'd get as many as fifteen corners really right, but Moss, he gets them all right on every lap" - Von Trips after the 1961 German Grand Prix.

There is a huge difference in programming for a fixed hardware console and for the PC. Shane Txl <[email hidden]> wrote in message news:0Y3n6.1072$[email hidden]...

Quoted text (46 lines)

> -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > not that much, really, PC programmpers are just "lazy" (of course I > could NEVER do what they do but still) and instead of them optimizing > their code it is us (the buyers) that do this optimization for them, > by buying bigger and faster machines. > > Check out the PSX1 games that are released now, on a 4 year old > system, when the best computer you could get was a P133 (maybe) with > 32Megs and a PCI video with 2 megs.... > > You will see a GREAT improvement in the quality both physics and > graphics of the game in the PSX (wipeout vs wipeout3 for example) and > a fantastic DETERIORATION of the PC (GP1 vs GP3 if you compare the > games running on similar systems). > > But i'm still VERY interested on F1RC...we'll see, hopefully since > the guys at UBISOFT do also a LOT of console games (incl. RAYMAN) it > is more in the "culture" of the company to optimize code whereas > papyrus (or sierra) just hope for us to purchase another extra 300 > Mhz of CPU speed. > > > "Jagg" <[email hidden]> a �crit dans le message news: > [email hidden]... > > On Tue, 27 Feb 2001 22:28:43 GMT, "Txl" <[email hidden]> wrote: > > > > > > >Botom line is physics is a piece of cake for computer, gfx is more > > >complicated. > > > > The add-in AI, crash detection and so on and I think it does impact > > the cpu resources a fair amount. > > > > -----BEGIN PGP SIGNATURE----- > Version: PGPfreeware 7.0.3 for non-commercial use <http://www.pgp.com> > > iQA/AwUBOpzLawhv7FFac7/nEQLVGgCgooVkG0WnHh9/q60uOWEbgdDOJk0AoMHM > MNkDg8jJjNk/GKpvSOKFeMwD > =EPlM > -----END PGP SIGNATURE----- > > >

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 I am happy to get a confirmation of what I have been saying for ages, the physics DON'T take CPU power, the GFX does. "Ruud van Gaal" <[email hidden]> a �crit dans le message news: [email hidden]...

Quoted text (33 lines)

> On Wed, 28 Feb 2001 16:24:14 -0000, "Tony Whitley" <t o n y . w h i > t l e y @ nStPlAwMorld.com> wrote: > > > > >"Ruud van Gaal" <[email hidden]> wrote ... > >> On Tue, 27 Feb 2001 23:04:26 GMT, "Rob Adams" > >> <[email hidden]> wrote: > ... > >> Ofcourse these days FPU's seem to give sin() and such away for > >> nearly free! > >And of course you could always create a table of sines and arctans > >and just look them up rather than calculating them every time... > > Hm, I'd rather have more precision for now than create tables yet. > > > And if you're using > >floating point arithmetic all the time no wonder the cpu cycles > >are swallowed up. > > Ah no, floating point is for free these days. Compares the same as > integer stuff almost (considering you'd have to do fixed point > arithmetic). > > As a note, the physics take up a measly 2% (roughly calculated) > currently in my sim. The main part is graphics. So if I optimized > it to 1 cycle, I still would have gained but 2% speed increase. So > I > won't, for now. > > > Ruud van Gaal, GPL Rank +53.25 > Pencil art : http://www.marketgraph.nl/gallery/ > Car simulation: http://www.marketgraph.nl/gallery/racer/

-----BEGIN PGP SIGNATURE----- Version: PGPfreeware 7.0.3 for non-commercial use <http://www.pgp.com> iQA/AwUBOp4fNghv7FFac7/nEQKvEgCgutXPZw52diKVKz/KSWqdn1NObgUAoIAc Hpcl7tdNMjBl74nogBaTZPLo =e4aY -----END PGP SIGNATURE-----

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 yes, when you program for a fixed system you cannot count on the user to buy some extra power, so you have to "push" the system and "find" some "tweak" to make it run better, this is called "code optimization" and I wish the buys programming on the PC would learn it someday... "Shane Lowry" <[email hidden]> a �crit dans le message news: 97jpm7$ijq$[email hidden]...

Quoted text (56 lines)

> There is a huge difference in programming for a fixed hardware > console and for the PC. > > Shane > > > Txl <[email hidden]> wrote in message > news:0Y3n6.1072$[email hidden]... > > -----BEGIN PGP SIGNED MESSAGE----- > > Hash: SHA1 > > > > not that much, really, PC programmpers are just "lazy" (of course > > I could NEVER do what they do but still) and instead of them > > optimizing their code it is us (the buyers) that do this > > optimization for them, by buying bigger and faster machines. > > > > Check out the PSX1 games that are released now, on a 4 year old > > system, when the best computer you could get was a P133 (maybe) > > with 32Megs and a PCI video with 2 megs.... > > > > You will see a GREAT improvement in the quality both physics and > > graphics of the game in the PSX (wipeout vs wipeout3 for example) > > and a fantastic DETERIORATION of the PC (GP1 vs GP3 if you > > compare the games running on similar systems). > > > > But i'm still VERY interested on F1RC...we'll see, hopefully > > since the guys at UBISOFT do also a LOT of console games (incl. > > RAYMAN) it is more in the "culture" of the company to optimize > > code whereas papyrus (or sierra) just hope for us to purchase > > another extra 300 Mhz of CPU speed. > > > > > > "Jagg" <[email hidden]> a �crit dans le message news: > > [email hidden]... > > > On Tue, 27 Feb 2001 22:28:43 GMT, "Txl" <[email hidden]> wrote: > > > > > > > > > >Botom line is physics is a piece of cake for computer, gfx is > > > >more complicated. > > > > > > The add-in AI, crash detection and so on and I think it does > > > impact the cpu resources a fair amount. > > > > > > > -----BEGIN PGP SIGNATURE----- > > Version: PGPfreeware 7.0.3 for non-commercial use > > <http://www.pgp.com> > > > > iQA/AwUBOpzLawhv7FFac7/nEQLVGgCgooVkG0WnHh9/q60uOWEbgdDOJk0AoMHM > > MNkDg8jJjNk/GKpvSOKFeMwD > > =EPlM > > -----END PGP SIGNATURE----- > > > > > > >

-----BEGIN PGP SIGNATURE----- Version: PGPfreeware 7.0.3 for non-commercial use <http://www.pgp.com> iQA/AwUBOp4fkwhv7FFac7/nEQKfNwCfc0TmUrASJe8xfLmZ3N+YFllZvSYAoOJO 27ezHPOYxkIsQU1S0jP07CdN =8XfB -----END PGP SIGNATURE-----

Txl wrote:

Quoted text (9 lines)

> > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > yes, when you program for a fixed system you cannot count on the user > to buy some extra power, so you have to "push" the system and "find" > some "tweak" to make it run better, this is called "code > optimization" and I wish the buys programming on the PC would learn > it someday...

You missed the crucial diff between fixed and open systems. If you have a fixed system, you can tweak a lot more. Just by using DirectX a lot of generalization is put into the code, thus not as performant as accessing the hardware directly. On PCs all contact to hardware has to go through abstractions (OS) that hides the details of the hardware, on a console there's just one type of hardware, so no need for abstraction, thus faster. l8er ronny -- This is Linux country. On a quiet night, you can hear Windows NT reboot.

On Wed, 28 Feb 2001 20:50:12 -0000, "Tony Whitley" <t o n y . w h i t l e y @ nStPlAwMorld.com> wrote: >> Ah no, floating point is for free these days. Compares the same as >> integer stuff almost (considering you'd have to do fixed point >> arithmetic). >Really? I'm an 8/16 bit embedded programmer, I didn't realise you Pentium >guys had it so easy :-) Yep, that's why we sometimes gets lazy. :) But it's great to just use float and get away with it.

Quoted text (9 lines)

>> As a note, the physics take up a measly 2% (roughly calculated) >> currently in my sim. The main part is graphics. So if I optimized it >> to 1 cycle, I still would have gained but 2% speed increase. So I >> won't, for now. >Of course, you have to identify /where/ to optimise. 2% does seem to >contradict a lot of the stuff said on this ng (and in this thread for that >matter). I believe your figures - you're actually doing it. And of course >you're following rule #1 of software development: "Get it working first, >then try to optimise it".

Right, and optimize where the pain is. And as long as your fps doesn't go below 50fps, well, then it wouldn't even be worth optimizing. I *did* have a track yesterday which got it to a halt, but then again, the track was divided into only a few models, so my triangle intersection was basically testing ALL triangles in the track, hehe. No wonder it ran like a turtle, and at other times did 70fps. Ruud van Gaal, GPL Rank +53.25 Pencil art : http://www.marketgraph.nl/gallery/ Car simulation: http://www.marketgraph.nl/gallery/racer/

According to your philosophy I can assume you agree with me since you don't respond David? On Wed, 28 Feb 2001 10:06:42 GMT, [email hidden] (Andre Warringa) wrote:

Quoted text (29 lines)

>On Wed, 28 Feb 2001 09:55:19 GMT, "David G Fisher" <[email hidden]> >wrote: > >>Why isn't the physics model as good as N4? Do you have some facts or is this >>just an opinion? > >It's just an opinion. What leads me to believe that the physics model >of F1RC is not quite accurate is this: >The tracks of F1RC are very, very close to the real tracks right? >Ok, yesterday I did a 1:26 at Melbourne. That's what, 10 seconds >faster than Schumi? >That doesn't take away the fact that is simply big fun driving in >F1RC. > >Problem is ofcourse that the AI is the worst I've ever seen in any >computer game, unless they made some BIGTIME chances on the last days, >and I heard from a couple of people that the online play via the net >sucks bigtime as well, because there is no prediction model. >That would make F1RC a great hotlapping game, unfortunately. > >>David G Fisher >>"I think Dave Fisher is a true genius, while Andre Warringa really is a bit >>of a twat." - Michael Schumacher > >I love you too David :) > >Andre > >"The physics model in the final version of 4x4 Evolution will me much better than it is in the demo" - David G Fisher

Andre "The physics model in the final version of 4x4 Evolution will me much better than it is in the demo" - David G Fisher

Since I don't have the full game, there isn't much I can respond to but I'll give it a try. Form what I've heard, if you have the preview code, the AI is not as good as it is in the review code. I've also heard at the HG forums that net play has been excellent, which is the opposite of what you were told. As for the times, we always smash the real world records in sims. If it's too much of a difference at some or all of the tracks, I would guess that it isn't necessarily a flaw in the physics model, but the numbers plugged into it may be off. Even in the demo with 40% anti-skid enabled, there doesn't seem to be an excessive amount of grip in the corners. I've had a lot of fun dancing on the edge because of this. It could be something as simple as the cars having the ability to accelerate too quickly which shouldn't be too hard to fix. All guesses and second hand information though. David G Fisher "I think Dave Fisher is a true genius, while Andre Warringa really is a bit of a twat." - Michael Schumacher "Andre Warringa" <[email hidden]> wrote in message news:[email hidden]...

Quoted text (11 lines)

> According to your philosophy I can assume you agree with me since you > don't respond David? > > > On Wed, 28 Feb 2001 10:06:42 GMT, [email hidden] (Andre > Warringa) wrote: > > >On Wed, 28 Feb 2001 09:55:19 GMT, "David G Fisher" <[email hidden]> > >wrote: > > > >>Why isn't the physics model as good as N4? Do you have some facts or is

this

Quoted text (18 lines)

> >>just an opinion? > > > >It's just an opinion. What leads me to believe that the physics model > >of F1RC is not quite accurate is this: > >The tracks of F1RC are very, very close to the real tracks right? > >Ok, yesterday I did a 1:26 at Melbourne. That's what, 10 seconds > >faster than Schumi? > >That doesn't take away the fact that is simply big fun driving in > >F1RC. > > > >Problem is ofcourse that the AI is the worst I've ever seen in any > >computer game, unless they made some BIGTIME chances on the last days, > >and I heard from a couple of people that the online play via the net > >sucks bigtime as well, because there is no prediction model. > >That would make F1RC a great hotlapping game, unfortunately. > > > >>David G Fisher > >>"I think Dave Fisher is a true genius, while Andre Warringa really is a

bit

Quoted text (7 lines)

> >>of a twat." - Michael Schumacher > > > >I love you too David :) > > > >Andre > > > >"The physics model in the final version of 4x4 Evolution will me much

better than it is in the demo" - David G Fisher > > > Andre > > "The physics model in the final version of 4x4 Evolution will me muchbetter than it is in the demo" - David G Fisher

If any of the WSC guys are watching this thread, could I talk one of you into providing an example of the results file that WSC will be writing when a race is completed? I want to include importing of this file in my SimResults DLL (already supports N4, and code is ready to be instered for GPL and N3/NL). In article <e0An6.134330$[email hidden]>, [email hidden] says...

Quoted text (50 lines)

> Since I don't have the full game, there isn't much I can respond to but I'll > give it a try. > > Form what I've heard, if you have the preview code, the AI is not as good as > it is in the review code. I've also heard at the HG forums that net play has > been excellent, which is the opposite of what you were told. As for the > times, we always smash the real world records in sims. If it's too much of a > difference at some or all of the tracks, I would guess that it isn't > necessarily a flaw in the physics model, but the numbers plugged into it may > be off. Even in the demo with 40% anti-skid enabled, there doesn't seem to > be an excessive amount of grip in the corners. I've had a lot of fun dancing > on the edge because of this. It could be something as simple as the cars > having the ability to accelerate too quickly which shouldn't be too hard to > fix. > > All guesses and second hand information though. > > David G Fisher > "I think Dave Fisher is a true genius, while Andre Warringa really is a bit > of a twat." - Michael Schumacher > > > "Andre Warringa" <[email hidden]> wrote in message > news:[email hidden]... > > According to your philosophy I can assume you agree with me since you > > don't respond David? > > > > > > On Wed, 28 Feb 2001 10:06:42 GMT, [email hidden] (Andre > > Warringa) wrote: > > > > >On Wed, 28 Feb 2001 09:55:19 GMT, "David G Fisher" <[email hidden]> > > >wrote: > > > > > >>Why isn't the physics model as good as N4? Do you have some facts or is > this > > >>just an opinion? > > > > > >It's just an opinion. What leads me to believe that the physics model > > >of F1RC is not quite accurate is this: > > >The tracks of F1RC are very, very close to the real tracks right? > > >Ok, yesterday I did a 1:26 at Melbourne. That's what, 10 seconds > > >faster than Schumi? > > >That doesn't take away the fact that is simply big fun driving in > > >F1RC. > > > > > >Problem is ofcourse that the AI is the worst I've ever seen in any > > >computer game, unless they made some BIGTIME chances on the last days, > > >and I heard from a couple of people that the online play via the net > > >sucks bigtime as well, because there is no prediction model.

Now I just read that HG had to put their review of the final version on hold because there is still yet another version on it's way to them. It looks like no one has had the true final version just yet. David G Fisher "David G Fisher" <[email hidden]> wrote in message news:e0An6.134330$[email hidden]... > Since I don't have the full game, there isn't much I can respond to butI'll > give it a try. > > Form what I've heard, if you have the preview code, the AI is not as goodas > it is in the review code. I've also heard at the HG forums that net playhas > been excellent, which is the opposite of what you were told. As for the > times, we always smash the real world records in sims. If it's too much ofa > difference at some or all of the tracks, I would guess that it isn't > necessarily a flaw in the physics model, but the numbers plugged into itmay > be off. Even in the demo with 40% anti-skid enabled, there doesn't seemto > be an excessive amount of grip in the corners. I've had a lot of fundancing > on the edge because of this. It could be something as simple as the cars > having the ability to accelerate too quickly which shouldn't be too hardto > fix. > > All guesses and second hand information though. > > David G Fisher > "I think Dave Fisher is a true genius, while Andre Warringa really is abit

Quoted text (16 lines)

> of a twat." - Michael Schumacher > > > "Andre Warringa" <[email hidden]> wrote in message > news:[email hidden]... > > According to your philosophy I can assume you agree with me since you > > don't respond David? > > > > > > On Wed, 28 Feb 2001 10:06:42 GMT, [email hidden] (Andre > > Warringa) wrote: > > > > >On Wed, 28 Feb 2001 09:55:19 GMT, "David G Fisher" <[email hidden]> > > >wrote: > > > > > >>Why isn't the physics model as good as N4? Do you have some facts or

is

Quoted text (19 lines)

> this > > >>just an opinion? > > > > > >It's just an opinion. What leads me to believe that the physics model > > >of F1RC is not quite accurate is this: > > >The tracks of F1RC are very, very close to the real tracks right? > > >Ok, yesterday I did a 1:26 at Melbourne. That's what, 10 seconds > > >faster than Schumi? > > >That doesn't take away the fact that is simply big fun driving in > > >F1RC. > > > > > >Problem is ofcourse that the AI is the worst I've ever seen in any > > >computer game, unless they made some BIGTIME chances on the last days, > > >and I heard from a couple of people that the online play via the net > > >sucks bigtime as well, because there is no prediction model. > > >That would make F1RC a great hotlapping game, unfortunately. > > > > > >>David G Fisher > > >>"I think Dave Fisher is a true genius, while Andre Warringa really is

a

Quoted text (16 lines)

> bit > > >>of a twat." - Michael Schumacher > > > > > >I love you too David :) > > > > > >Andre > > > > > >"The physics model in the final version of 4x4 Evolution will me much > better than it is in the demo" - David G Fisher > > > > > > Andre > > > > "The physics model in the final version of 4x4 Evolution will me much > better than it is in the demo" - David G Fisher >

FP is typically as fast or faster than integer ops on the new pentium/et al machines... Actually, these days doing more math is often faster than table lookups because memory access is usually more of a bottleneck than FP ops (especially on a certain L2-cache-less console that will remain unnamed)--a polynomial curve fit can be a win in some cases... even a "Pacejka blaster" is usually lost among big heaps of matrix operations, terrain lookups, dirty socks, etc. Unfortunately the new vector-units/SSE2/P4/etc instruction sets take special per-chip work to take advantage of, because compilers can't do that automatically yet. There's usually a lag before these are adopted by more programs--hand-coding assembly for a P4 is usually not the best use of time while it's new, because it's the PIII's and celerons that need the help. The first rule of optimization is to optimize your worst case first, not the best case, unless you really want to screw over your "minimum platform" base. The new ops are also much easier for tight inner loops like sound processing and vertex transformation ala Direct3D than big long complex physics loops, although again you can optimize your calculations and just spend more time waiting on the memory bus... macro-scale optimizations are usually more important, especially when you're straddling many different CPU ranges that can have conflicting wants. (forest/trees, etc) Such is life. Luckily these days we seem to be pushing more of the graphics math onto dedicated hardware, so we can start to colonize more of the CPU for physics and AI... At last count physics and AI are more than half of the "ee core" usage on our PS2 sku thanks to our friend VU1. Lost are the days of LDA, STA, except for the gameboy developers... "Tony Whitley" <t o n y . w h i t l e y @ nStPlAwMorld.com> wrote in message news:mH9n6.3829$[email hidden]...

Quoted text (7 lines)

> > "Ruud van Gaal" <[email hidden]> wrote ... > > On Tue, 27 Feb 2001 23:04:26 GMT, "Rob Adams" <[email hidden]> > > wrote: > > > > ... > > >I have seen articles proposing PPU's, or Physics Processing Units,

which

Quoted text (15 lines)

> > >could actually do an "angular momentum" or "rigid body rotation" > calculation > > >in one cycle. If those every see the light of day then a 188Hz physics > > >engine really could translate to 188 cycles. > > > > Even then, you'd have multiple physics calculations; there's a lot > > more surrounding these RRB rotation commands. > > Matrix calculations in 1 cycle could help very much. > > And a Pacejka formula blaster, hehe, that takes IIRC 2 sin() and 2 > > arctan calls. For a single tire, and you need 3 of those formulas; for > > lateral, longitudinal and aligning moment (for your force feedback). > > > > Ofcourse these days FPU's seem to give sin() and such away for nearly > > free! > And of course you could always create a table of sines and arctans and

just > look them up rather than calculating them every time... And if you'reusing

Quoted text (7 lines)

> floating point arithmetic all the time no wonder the cpu cycles are > swallowed up. > > -- > Tony Whitley > GPLRank +9.99 > "On some laps I'd get as many as fifteen corners really right, but Moss,

he > gets them all right on every lap" - Von Trips after the 1961 German Grand > Prix. > >

"Dave Pollatsek" <[email hidden]> wrote ... > FP is typically as fast or faster than integer ops on the new pentium/etal > machines... ><snip a lot of interesting stuff about Pentiums/Pentii/Pentia> Thanks for enlightening me, I think I'll stay with my cosy 16 bit processor, I thought they were getting too complicated nowadays! I'll skip the P4 assembler, if it's all right with you. > Lost are the days of LDA, STA, except for the gameboy developers...Ah, those were the days. I think I feel a tear coming... -- Tony Whitley GPLRank +9.99 "On some laps I'd get as many as fifteen corners really right, but Moss, he gets them all right on every lap" - Von Trips after the 1961 German Grand Prix.

On Thu, 01 Mar 2001 22:26:18 GMT, "David G Fisher" <[email hidden]> wrote:

Quoted text (20 lines)

>Since I don't have the full game, there isn't much I can respond to but I'll >give it a try. > >Form what I've heard, if you have the preview code, the AI is not as good as >it is in the review code. I've also heard at the HG forums that net play has >been excellent, which is the opposite of what you were told. As for the >times, we always smash the real world records in sims. If it's too much of a >difference at some or all of the tracks, I would guess that it isn't >necessarily a flaw in the physics model, but the numbers plugged into it may >be off. Even in the demo with 40% anti-skid enabled, there doesn't seem to >be an excessive amount of grip in the corners. I've had a lot of fun dancing >on the edge because of this. It could be something as simple as the cars >having the ability to accelerate too quickly which shouldn't be too hard to >fix. > >All guesses and second hand information though. > >David G Fisher >"I think Dave Fisher is a true genius, while Andre Warringa really is a bit >of a twat." - Michael Schumacher

We'll just have to wait and see then.. the AI is really bad in the beta, like the review says here: http://www.gamesmania.de/cgi-bin/gm/rcdox.pl?layer=gmpc/gmpc&gmpccontent=gmpc_test_simulation_f1racingsim_page1 (german preview) the AI consist of 21 Ukyo Katayamas :) The AI crashes all over the place, can brake unrealistically late and smack you in the rear without mercy. Hats off to Ubi if they could get the AI working the the last couple of days before going gold. Racing in F1RC is superb fun, but I'd hate to see this becoming a second Nascar Heat. If they fix the AI and multiplayer, then this will without doubt be my second favorite racing sim, and clearly the best current F1 sim so far. Andre "The physics model in the final version of 4x4 Evolution will me much better than it is in the demo" - David G Fisher

I wouldn't generalize it so quickly. It really depends on what your physics calculations are. If you go into more detail, the timestep needed can become extremely small. For example, start treating the wheels on a modern F1 or CART car as Newtonian objects (calculate proper acceleration instead of the velocity as immediately adapting to damper/spring system), and you can't get away with much more that a 1/1000 s timestep. Try incuding the flexibility of each suspension link atop of that, and you'll need a timestep a couple orders of magnitude smaller, while the complexity of the calculation increases as well. So, it's all a matter of fidelity. With current games, though, I think your statement could be close to correct. Although, I do have a strange suspicion WSC will somehow tip the ballance, considering what it is that is promised to be modelled. -Gregor Txl wrote:

Quoted text (7 lines)

> > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > I am happy to get a confirmation of what I have been saying for ages, > the physics DON'T take CPU power, the GFX does. >

I se your point, but the question here at the tart was that F1RC was faster in FPS than N4 because the physics in N4 were so much better (which remains to be proved), one thing for sure is that the AI in N4 is not too CPU power consuling as far as I could see on this NG.... "Gregor Veble" <[email hidden]> a �crit dans le message news: [email hidden]...

Quoted text (26 lines)

> I wouldn't generalize it so quickly. It really depends on what your > physics calculations are. If you go into more detail, the timestep > needed can become extremely small. For example, start treating the > wheels on a modern F1 or CART car as Newtonian objects (calculate proper > acceleration instead of the velocity as immediately adapting to > damper/spring system), and you can't get away with much more that a > 1/1000 s timestep. Try incuding the flexibility of each suspension link > atop of that, and you'll need a timestep a couple orders of magnitude > smaller, while the complexity of the calculation increases as well. > > So, it's all a matter of fidelity. With current games, though, I think > your statement could be close to correct. Although, I do have a strange > suspicion WSC will somehow tip the ballance, considering what it is that > is promised to be modelled. > > -Gregor > > Txl wrote: > > > > -----BEGIN PGP SIGNED MESSAGE----- > > Hash: SHA1 > > > > I am happy to get a confirmation of what I have been saying for ages, > > the physics DON'T take CPU power, the GFX does. > > >

On Fri, 02 Mar 2001 11:27:59 +0100, Gregor Veble <[email hidden]> wrote:

Quoted text (9 lines)

>I wouldn't generalize it so quickly. It really depends on what your >physics calculations are. If you go into more detail, the timestep >needed can become extremely small. For example, start treating the >wheels on a modern F1 or CART car as Newtonian objects (calculate proper >acceleration instead of the velocity as immediately adapting to >damper/spring system), and you can't get away with much more that a >1/1000 s timestep. Try incuding the flexibility of each suspension link >atop of that, and you'll need a timestep a couple orders of magnitude >smaller, while the complexity of the calculation increases as well.

WSC uses a timestep much bigger than that (but not too be afraid that that does the physics to waste). So it is possible, and oh gosh, will the physics be light on the CPU then when I get to the same level (in 2005? ;-). Ruud van Gaal, GPL Rank +53.25 Pencil art : http://www.marketgraph.nl/gallery/ Car simulation: http://www.marketgraph.nl/gallery/racer/

I remember in college they had an old PDP-8 that they used in one course--no terminals or anything, just flipping switches and looking at flashing binary lights. You'd write your program on a notepad, hand-assemble it into octal, and then spend 5 minutes flicking switches up and down, run your program, and see the result on the lights, debug, etc. It was great--totally bare bones, really simple instruction set, you literally knew what was at every memory location in the thing (only 4K 12-bit words...) Plus, when you powered up the sucker, the fans would make this cool phase-shifting turbine sound... "Tony Whitley" <t o n y . w h i t l e y @ nStPlAwMorld.com> wrote in message news:GeKn6.743$[email hidden]...

Quoted text (8 lines)

> > "Dave Pollatsek" <[email hidden]> wrote ... > > FP is typically as fast or faster than integer ops on the new pentium/et > al > > machines... > > > <snip a lot of interesting stuff about Pentiums/Pentii/Pentia> > Thanks for enlightening me, I think I'll stay with my cosy 16 bit

processor,

Quoted text (10 lines)

> I thought they were getting too complicated nowadays! I'll skip the P4 > assembler, if it's all right with you. > > > Lost are the days of LDA, STA, except for the gameboy developers... > Ah, those were the days. I think I feel a tear coming... > > -- > Tony Whitley > GPLRank +9.99 > "On some laps I'd get as many as fifteen corners really right, but Moss,

he > gets them all right on every lap" - Von Trips after the 1961 German Grand > Prix. > >

Ruud van Gaal wrote:

Quoted text (9 lines)

> > WSC uses a timestep much bigger than that (but not too be afraid that > that does the physics to waste). So it is possible, and oh gosh, will > the physics be light on the CPU then when I get to the same level (in > 2005? ;-). > > Ruud van Gaal, GPL Rank +53.25 > Pencil art : http://www.marketgraph.nl/gallery/ > Car simulation: http://www.marketgraph.nl/gallery/racer/

Hi, which number are you refering to? My guess is that 1/1000 s is what WSC should be using, as GPL already uses 1/288 s. AFAIK WSC doesn't model flexibilty of parts coupled with their inertial modelling (only when the two are modelled together should problems occur, not spearately), so all is well :). -Gregor

On Fri, 02 Mar 2001 15:12:56 +0100, Gregor Veble <[email hidden]> wrote: >Hi, > >which number are you refering to? My guess is that 1/1000 s is what WSC >should be using, as GPL already uses 1/288 s. It's using a mere 100Hz (I talked to Chris yesterday), and according to him, it can go even lower. I'm using 100Hz while developing, because I can then turn on draw_every_integration_step and still get a bit of motion ;-) But other than that, 250Hz is the lowest I go right now. WSC uses some special stuff to be able to get it that low. > AFAIK WSC doesn't model >flexibilty of parts coupled with their inertial modelling (only when the >two are modelled together should problems occur, not spearately), so all >is well :). I have no idea how they do this. :) I only read yesterday about the separation of car parts; seems very cool! Ruud van Gaal, GPL Rank +53.25 Pencil art : http://www.marketgraph.nl/gallery/ Car simulation: http://www.marketgraph.nl/gallery/racer/

"Dave Pollatsek" <[email hidden]> wrote ... > I remember in college they had an old PDP-8 that they used in onecourse--no > terminals or anything, just flipping switches and looking at flashingbinary > lights. You'd write your program on a notepad, hand-assemble it intooctal, > and then spend 5 minutes flicking switches up and down, run your program, > and see the result on the lights, debug, etc. It was great--totally bare > bones, really simple instruction set, you literally knew what was at every > memory location in the thing (only 4K 12-bit words...) Plus, when you > powered up the sucker, the fans would make this cool phase-shiftingturbine > sound... >That was the latest thing when I started out... I remember making "computer music" by setting it going in a loop whose length was defined by those rocker switches and then tuning a radio off-channel to pick up the EMI. We did have a paper tape assembler - "all" you had to do was bootstrap the PDP8 using the rocker switches. That was the first machine I played a game on - Lunar Lander, written in a precursor of BASIC IIRC. My final year project was to program it (in assembler of course) to play Monopoly. The AI turned out pretty crude (I had great plans :-) but still good enough to beat me sometimes. I also did the graphics for someone else's project - Noughts and Crosses (Tic-Tac-Toe in the US) using a "Direct View Storage Tube" - like a 'scope you could draw lines on. He got better marks than me, with a special mention for the display :-( I thought I was an old hand until I met a guy whose first job was tuning the mercury delay lines (on a LEO or something) every morning with a 'scope! -- Tony Whitley GPLRank +9.99 "On some laps I'd get as many as fifteen corners really right, but Moss, he gets them all right on every lap" - Von Trips after the 1961 German Grand Prix.

On Fri, 02 Mar 2001 12:58:16 GMT, "Txl" <[email hidden]> wrote: >I se your point, but the question here at the tart was that F1RC was faster >in FPS than N4 because the physics in N4 were so much better (which remains >to be proved), I never said it was because of that. I suggested it might be.

The real life drivers have reached 1:27 in practice this weekend. Now I'm confused. I know they were slower in 1999, but 10 seconds slower? Actually, when 2001 user-made cars are released for F1RC (soon after shipping I'm sure), then the times should be pretty accurate. David G Fisher "Andre Warringa" <[email hidden]> wrote in message news:[email hidden]... > On Wed, 28 Feb 2001 09:55:19 GMT, "David G Fisher" <[email hidden]> > wrote: > > >Why isn't the physics model as good as N4? Do you have some facts or isthis

Quoted text (18 lines)

> >just an opinion? > > It's just an opinion. What leads me to believe that the physics model > of F1RC is not quite accurate is this: > The tracks of F1RC are very, very close to the real tracks right? > Ok, yesterday I did a 1:26 at Melbourne. That's what, 10 seconds > faster than Schumi? > That doesn't take away the fact that is simply big fun driving in > F1RC. > > Problem is ofcourse that the AI is the worst I've ever seen in any > computer game, unless they made some BIGTIME chances on the last days, > and I heard from a couple of people that the online play via the net > sucks bigtime as well, because there is no prediction model. > That would make F1RC a great hotlapping game, unfortunately. > > >David G Fisher > >"I think Dave Fisher is a true genius, while Andre Warringa really is a

bit

Quoted text (7 lines)

> >of a twat." - Michael Schumacher > > I love you too David :) > > Andre > > "The physics model in the final version of 4x4 Evolution will me much

better than it is in the demo" - David G Fisher

On Sat, 03 Mar 2001 01:04:31 GMT, "David G Fisher" <[email hidden]> wrote:

Quoted text (9 lines)

>The real life drivers have reached 1:27 in practice this weekend. Now I'm >confused. I know they were slower in 1999, but 10 seconds slower? Actually, >when 2001 user-made cars are released for F1RC (soon after shipping I'm >sure), then the times should be pretty accurate. > >David G Fisher >> Ok, yesterday I did a 1:26 at Melbourne. That's what, 10 seconds >> faster than Schumi?

I found the results from '99: 99 Pole : M. Hakkinen - 1'30"462, 210.997 km/h Best Lap : M.Schumacher - 1'32"112 Winner 1999: E. Irvine (Ferrari) - 1h35'01"659 If the tracks of F1RC are realistic AND the physics are realistic, that makes me a damn good driver with my 1:26, on expert level with -default- setup :) Andre

Quoted text (8 lines)

>Hi, > >which number are you refering to? My guess is that 1/1000 s is what WSC >should be using, as GPL already uses 1/288 s. AFAIK WSC doesn't model >flexibilty of parts coupled with their inertial modelling (only when the >two are modelled together should problems occur, not spearately), so all >is well :). >

According to an interview done awhile back, they were using 1/100th second time step. Todd Wasson --- Performance Simulations Drag Racing and Top Speed Prediction Software http://PerformanceSimulations.Com

>I am happy to get a confirmation of what I have been saying for ages, >the physics DON'T take CPU power, the GFX does. I can vouch for that also so far in my work. I run a split frequency system using a couple of different settings simultaneously (300Hz main freq, about what Ruud runs) and the entire thing runs through the loop 20 times or so in less than 1/1000th second on a P-400. That means, if you eliminated the graphics entirely, you'd get about 1000fps with my engine running at 300Hz, with no clever optimizations at all, and no hand written assembly. However, this INCLUDES running intersection and distance calculations with a few hundred triangles per wheel, per iteration (20 times during the 1/1000th second period!!). That's where the vast majority of the 1/1000th second "physics" calculations are, just triangle calculations to get spring and damper forces and what not. This section could be optimized **extensively**, and probably has been far beyond anything I know how to do by the wizards that make the sims on the market today. However, as Ruud pointed out, taking the 2% or so and spending all your time cutting it to 1% won't effect the frame rate much. You'd go from 48 to 50 fps (right?). If I take away the triangle intersection calculations, I can't even tell you how fast it runs because the system clock doesn't have that much precision AFAIK. The rest is graphics, AI, etc.. With 20 AI cars on the field, this percentage would climb noticeably. However, I've yet to see a sim appear to use the "player car" physics to run the AI cars. It's unbelievably complicated to tell the AI cars how much to push the pedals and when with a full blown model like that. Folks that can do it write lap time predictors and sell them to race folks for mega bucks. These probably don't run in real time or with 3-D graphics either :-) That's just my guess, though. This would make the AI car physics even *more* taxing than the player car. I doubt that's the case. Physics ain't too much :-) Todd Wasson --- Performance Simulations Drag Racing and Top Speed Prediction Software http://PerformanceSimulations.Com

Quoted text (7 lines)

>>I don't know if they are going to use the same engine for AI, I too doubt >>it, but even if they use a simplified physics model, it doesn't necessarily >>mean that it is a 'light' model. > >Doesn't N4 and NH use the same physics model for all cars? I don't >like the idea of racing against cars that don't have the same >limitations as you do.

I think the important thing when it comes to AI cars is that they accelerate and turn at the right rates. The subtleties of handling can be taken out and the cars can still be made to perform realistically and get accurate lap times. If an AI car is going 100mph and hits the brakes, it probably doesn't need to handle all the forces from shocks, springs, etc., to get the acceleration right. Just give it an acceleration value that varies with what is typical of the player's car and be done with it. An AI car would lose the accurate transient behavior a player car might have, but I doubt many people would notice. It can be fudged by inputting the "right" accelerations directly. Todd Wasson --- Performance Simulations Drag Racing and Top Speed Prediction Software http://PerformanceSimulations.Com

Well, we already knew that. If you can master GPL than anything else is cake. Right? :-P David G Fisher "Andre Warringa" <[email hidden]> wrote in message news:[email hidden]... > On Sat, 03 Mar 2001 01:04:31 GMT, "David G Fisher" <[email hidden]> > wrote: > > >The real life drivers have reached 1:27 in practice this weekend. Now I'm > >confused. I know they were slower in 1999, but 10 seconds slower?Actually,

Quoted text (19 lines)

> >when 2001 user-made cars are released for F1RC (soon after shipping I'm > >sure), then the times should be pretty accurate. > > > >David G Fisher > > >> Ok, yesterday I did a 1:26 at Melbourne. That's what, 10 seconds > >> faster than Schumi? > > I found the results from '99: > > 99 Pole : M. Hakkinen - 1'30"462, 210.997 km/h > Best Lap : M.Schumacher - 1'32"112 > Winner 1999: E. Irvine (Ferrari) - 1h35'01"659 > > If the tracks of F1RC are realistic AND the physics are realistic, > that makes me a damn good driver with my 1:26, on expert level with > -default- setup :) > > Andre

well, it makes you as good as schumi which made pole in 1.26. apparently they are saying that the tires make tremendous advances this year, bridgestone had to move their asses because michelin was coming and so they finally started to ork on the F1 tires. "Andre Warringa" <[email hidden]> a �crit dans le message news: [email hidden]... > On Sat, 03 Mar 2001 01:04:31 GMT, "David G Fisher" <[email hidden]> > wrote: > > >The real life drivers have reached 1:27 in practice this weekend. Now I'm > >confused. I know they were slower in 1999, but 10 seconds slower?Actually,

Quoted text (19 lines)

> >when 2001 user-made cars are released for F1RC (soon after shipping I'm > >sure), then the times should be pretty accurate. > > > >David G Fisher > > >> Ok, yesterday I did a 1:26 at Melbourne. That's what, 10 seconds > >> faster than Schumi? > > I found the results from '99: > > 99 Pole : M. Hakkinen - 1'30"462, 210.997 km/h > Best Lap : M.Schumacher - 1'32"112 > Winner 1999: E. Irvine (Ferrari) - 1h35'01"659 > > If the tracks of F1RC are realistic AND the physics are realistic, > that makes me a damn good driver with my 1:26, on expert level with > -default- setup :) > > Andre

On Sat, 03 Mar 2001 06:23:59 GMT, "David G Fisher" <[email hidden]> wrote: >Well, we already knew that. If you can master GPL than anything else is >cake. Right? >:-P > >David G Fisher I'm not that good at GPL, I can practice what I want but I will never reach the times the fastest guys are getting. At most tracks they are seconds faster than me. So if I can do a 1:26 at Melbourne, I wonder what the really talented guys will get for laptimes.. So -to get back to the topic dear David- the laptimes in F1RC are totally unrealistic. This year Schumi came damn close to my laptime (yep, Schumi is good) but that's not the point right? F1RC simulates the 99 season, so we should compare laptimes to those from 99. Right? So I'll stick to my opinion, if the tracks are accurate, and laptimes are - a couple of seconds - faster in F1RC, the physics can't be 100% realistic. Untill someone has a better theory :-P Andre

Quoted text (28 lines)

> >"Andre Warringa" <[email hidden]> wrote in message >news:[email hidden]... >> On Sat, 03 Mar 2001 01:04:31 GMT, "David G Fisher" <[email hidden]> >> wrote: >> >> >The real life drivers have reached 1:27 in practice this weekend. Now I'm >> >confused. I know they were slower in 1999, but 10 seconds slower? >Actually, >> >when 2001 user-made cars are released for F1RC (soon after shipping I'm >> >sure), then the times should be pretty accurate. >> > >> >David G Fisher >> >> >> Ok, yesterday I did a 1:26 at Melbourne. That's what, 10 seconds >> >> faster than Schumi? >> >> I found the results from '99: >> >> 99 Pole : M. Hakkinen - 1'30"462, 210.997 km/h >> Best Lap : M.Schumacher - 1'32"112 >> Winner 1999: E. Irvine (Ferrari) - 1h35'01"659 >> >> If the tracks of F1RC are realistic AND the physics are realistic, >> that makes me a damn good driver with my 1:26, on expert level with >> -default- setup :) >> >> Andre

On Sat, 03 Mar 2001 12:11:41 GMT, "Txl" <[email hidden]> wrote: >well, it makes you as good as schumi which made pole in 1.26. apparently >they are saying that the tires make tremendous advances this year, >bridgestone had to move their asses because michelin was coming and so they >finally started to ork on the F1 tires. I'm apparently better than Schumi, remember, in F1RC I'm driving his '99 car :) Andre

Quoted text (28 lines)

>"Andre Warringa" <[email hidden]> a �crit dans le message news: >[email hidden]... >> On Sat, 03 Mar 2001 01:04:31 GMT, "David G Fisher" <[email hidden]> >> wrote: >> >> >The real life drivers have reached 1:27 in practice this weekend. Now I'm >> >confused. I know they were slower in 1999, but 10 seconds slower? >Actually, >> >when 2001 user-made cars are released for F1RC (soon after shipping I'm >> >sure), then the times should be pretty accurate. >> > >> >David G Fisher >> >> >> Ok, yesterday I did a 1:26 at Melbourne. That's what, 10 seconds >> >> faster than Schumi? >> >> I found the results from '99: >> >> 99 Pole : M. Hakkinen - 1'30"462, 210.997 km/h >> Best Lap : M.Schumacher - 1'32"112 >> Winner 1999: E. Irvine (Ferrari) - 1h35'01"659 >> >> If the tracks of F1RC are realistic AND the physics are realistic, >> that makes me a damn good driver with my 1:26, on expert level with >> -default- setup :) >> >> Andre >

On Sat, 03 Mar 2001 12:11:41 GMT, "Txl" <[email hidden]> wrote: >well, it makes you as good as schumi which made pole in 1.26. apparently >they are saying that the tires make tremendous advances this year, >bridgestone had to move their asses because michelin was coming and so they >finally started to ork on the F1 tires. I'm apparently better than Schumi, remember, in F1RC I'm driving his '99 car :) Andre

Quoted text (28 lines)

>"Andre Warringa" <[email hidden]> a �crit dans le message news: >[email hidden]... >> On Sat, 03 Mar 2001 01:04:31 GMT, "David G Fisher" <[email hidden]> >> wrote: >> >> >The real life drivers have reached 1:27 in practice this weekend. Now I'm >> >confused. I know they were slower in 1999, but 10 seconds slower? >Actually, >> >when 2001 user-made cars are released for F1RC (soon after shipping I'm >> >sure), then the times should be pretty accurate. >> > >> >David G Fisher >> >> >> Ok, yesterday I did a 1:26 at Melbourne. That's what, 10 seconds >> >> faster than Schumi? >> >> I found the results from '99: >> >> 99 Pole : M. Hakkinen - 1'30"462, 210.997 km/h >> Best Lap : M.Schumacher - 1'32"112 >> Winner 1999: E. Irvine (Ferrari) - 1h35'01"659 >> >> If the tracks of F1RC are realistic AND the physics are realistic, >> that makes me a damn good driver with my 1:26, on expert level with >> -default- setup :) >> >> Andre >

On Sat, 03 Mar 2001 12:11:41 GMT, "Txl" <[email hidden]> wrote: >well, it makes you as good as schumi which made pole in 1.26. apparently >they are saying that the tires make tremendous advances this year, >bridgestone had to move their asses because michelin was coming and so they >finally started to ork on the F1 tires. I'm apparently better than Schumi, remember, in F1RC I'm driving his '99 car :) Andre

Quoted text (28 lines)

>"Andre Warringa" <[email hidden]> a �crit dans le message news: >[email hidden]... >> On Sat, 03 Mar 2001 01:04:31 GMT, "David G Fisher" <[email hidden]> >> wrote: >> >> >The real life drivers have reached 1:27 in practice this weekend. Now I'm >> >confused. I know they were slower in 1999, but 10 seconds slower? >Actually, >> >when 2001 user-made cars are released for F1RC (soon after shipping I'm >> >sure), then the times should be pretty accurate. >> > >> >David G Fisher >> >> >> Ok, yesterday I did a 1:26 at Melbourne. That's what, 10 seconds >> >> faster than Schumi? >> >> I found the results from '99: >> >> 99 Pole : M. Hakkinen - 1'30"462, 210.997 km/h >> Best Lap : M.Schumacher - 1'32"112 >> Winner 1999: E. Irvine (Ferrari) - 1h35'01"659 >> >> If the tracks of F1RC are realistic AND the physics are realistic, >> that makes me a damn good driver with my 1:26, on expert level with >> -default- setup :) >> >> Andre >

what ? you fuck frentzen's girlfriend too ? inside joke... "Andre Warringa" <[email hidden]> a �crit dans le message news: [email hidden]... > On Sat, 03 Mar 2001 12:11:41 GMT, "Txl" <[email hidden]> wrote: > > >well, it makes you as good as schumi which made pole in 1.26. apparently > >they are saying that the tires make tremendous advances this year, > >bridgestone had to move their asses because michelin was coming and sothey

Quoted text (14 lines)

> >finally started to ork on the F1 tires. > > I'm apparently better than Schumi, remember, in F1RC I'm driving his > '99 car :) > > Andre > > > >"Andre Warringa" <[email hidden]> a �crit dans le message news: > >[email hidden]... > >> On Sat, 03 Mar 2001 01:04:31 GMT, "David G Fisher" <[email hidden]> > >> wrote: > >> > >> >The real life drivers have reached 1:27 in practice this weekend. Now

I'm > >> >confused. I know they were slower in 1999, but 10 seconds slower? > >Actually, > >> >when 2001 user-made cars are released for F1RC (soon after shippingI'm

Quoted text (21 lines)

> >> >sure), then the times should be pretty accurate. > >> > > >> >David G Fisher > >> > >> >> Ok, yesterday I did a 1:26 at Melbourne. That's what, 10 seconds > >> >> faster than Schumi? > >> > >> I found the results from '99: > >> > >> 99 Pole : M. Hakkinen - 1'30"462, 210.997 km/h > >> Best Lap : M.Schumacher - 1'32"112 > >> Winner 1999: E. Irvine (Ferrari) - 1h35'01"659 > >> > >> If the tracks of F1RC are realistic AND the physics are realistic, > >> that makes me a damn good driver with my 1:26, on expert level with > >> -default- setup :) > >> > >> Andre > > > >

> I'm apparently better than Schumi, remember, in F1RC I'm driving his > '99 car :) > > Andre > Schumi is a HACK! What has he ever done on GPL? If you guys insist on comparing yourselves to real world drivers, at least give them the benefit of never having raced on GPL, right David? At least two Winston Cup drivers are regular ONline racers as well. and Juan Monotonal practices with sims...but does he challenge the GODs online? C'mon WE all know Gregor and Wolfy could open up a can of 'Whup Ass' on any of these supposed Super License drivers... -the other dave henrie

HELLOOOOOOOOO??????????? AANNNYYYYBOOODDYYYY HHHOOOOOOMMEE ?????? ed_ (awaiting echo) "Txl" wrote...

Quoted text (8 lines)

> what ? you fuck frentzen's girlfriend too ? > > inside joke... > "Andre Warringa" wrote: > > On Sat, 03 Mar 2001 12:11:41 GMT, "Txl" <[email hidden]> wrote: > > > > >well, it makes you as good as schumi which made pole in 1.26.

apparently

Quoted text (17 lines)

> > >they are saying that the tires make tremendous advances this year, > > >bridgestone had to move their asses because michelin was coming and so > they > > >finally started to ork on the F1 tires. > > > > I'm apparently better than Schumi, remember, in F1RC I'm driving his > > '99 car :) > > > > Andre > > > > > > >"Andre Warringa" <[email hidden]> a �crit dans le message news: > > >[email hidden]... > > >> On Sat, 03 Mar 2001 01:04:31 GMT, "David G Fisher" <[email hidden]> > > >> wrote: > > >> > > >> >The real life drivers have reached 1:27 in practice this weekend.

Now

Quoted text (28 lines)

> I'm > > >> >confused. I know they were slower in 1999, but 10 seconds slower? > > >Actually, > > >> >when 2001 user-made cars are released for F1RC (soon after shipping > I'm > > >> >sure), then the times should be pretty accurate. > > >> > > > >> >David G Fisher > > >> > > >> >> Ok, yesterday I did a 1:26 at Melbourne. That's what, 10 seconds > > >> >> faster than Schumi? > > >> > > >> I found the results from '99: > > >> > > >> 99 Pole : M. Hakkinen - 1'30"462, 210.997 km/h > > >> Best Lap : M.Schumacher - 1'32"112 > > >> Winner 1999: E. Irvine (Ferrari) - 1h35'01"659 > > >> > > >> If the tracks of F1RC are realistic AND the physics are realistic, > > >> that makes me a damn good driver with my 1:26, on expert level with > > >> -default- setup :) > > >> > > >> Andre > > > > > > > > >

ok, my joke went flat... Just in case you didn't know the wife of Schumi (Corinna) used to be the girlfriend of Frentzen, he was bringing her on the racetracks and the other one "stole" her from him, they got mixed into a "verbal fight" (possibly some punches were exchanged) and they have had a very "cold" relationship since then. Ok, F1 gossip column over... "Ed Solheim" <[email hidden]> a �crit dans le message news: 1S9o6.6390$[email hidden]... > HELLOOOOOOOOO??????????? > > AANNNYYYYBOOODDYYYY HHHOOOOOOMMEE ?????? >

Now I like this re-post Dave! You get the YeeHaa post of the week!! <G> "Dave Henrie" <[email hidden]> wrote in message news:I49o6.499614$[email hidden]... <snipped> | challenge the GODs online? C'mon WE all know Gregor and Wolfy couldopen | up a can of 'Whup Ass' on | any of these supposed Super License drivers... | -the other dave henrie

Hsssssssssssss Booooooo Poor joke 2 thumbs & 2 toes down! hahahaha "Txl" <[email hidden]> wrote in message news:Gibo6.2242$[email hidden]... | ok, my joke went flat... | | Just in case you didn't know the wife of Schumi (Corinna) used to be the | girlfriend of Frentzen, he was bringing her on the racetracks and theother

Quoted text (15 lines)

| one "stole" her from him, they got mixed into a "verbal fight" (possibly | some punches were exchanged) and they have had a very "cold" relationship | since then. | | Ok, F1 gossip column over... | | "Ed Solheim" <[email hidden]> a �crit dans le message news: | 1S9o6.6390$[email hidden]... | > HELLOOOOOOOOO??????????? | > | > AANNNYYYYBOOODDYYYY HHHOOOOOOMMEE ?????? | > | | |

Wow--I can't imagine actually writing a real program for the PDP-8! I started with BASIC on a C-64, but was already using C on a PC-XT when I started more serious programming in high school. I recall a program for the C-64 that downloaded some code to the disk drive (which had it's own processor and 2K of RAM IIRC) that would play a song by modulating the drive motor. I think I got into this business just at the tail end of the period of heavy assembly-language usage. I remember seeing pictures of a "tinker toy" computer that played tic-tac-toe; the thing was at least 10 feet high... I think it used some smarts to flip the current game state around to a normalized equivalent, look up the move in a table, and then flip it back; not very interesting AI but a remarkable acheivement nonetheless. I still have dreams of building a stored-program mechanical computer, maybe when I retire or something... "Tony Whitley" <t o n y . w h i t l e y @ nStPlAwMorld.com> wrote in message news:eF1o6.4109$[email hidden]...

Quoted text (8 lines)

> "Dave Pollatsek" <[email hidden]> wrote ... > > I remember in college they had an old PDP-8 that they used in one > course--no > > terminals or anything, just flipping switches and looking at flashing > binary > > lights. You'd write your program on a notepad, hand-assemble it into > octal, > > and then spend 5 minutes flicking switches up and down, run your

program, > > and see the result on the lights, debug, etc. It was great--totallybare > > bones, really simple instruction set, you literally knew what was atevery > > memory location in the thing (only 4K 12-bit words...) Plus, when you > > powered up the sucker, the fans would make this cool phase-shifting > turbine > > sound... > > > That was the latest thing when I started out... I remember making"computer > music" by setting it going in a loop whose length was defined by those > rocker switches and then tuning a radio off-channel to pick up the EMI. We > did have a paper tape assembler - "all" you had to do was bootstrap thePDP8 > using the rocker switches. That was the first machine I played a game on - > Lunar Lander, written in a precursor of BASIC IIRC. My final year project > was to program it (in assembler of course) to play Monopoly. The AI turned > out pretty crude (I had great plans :-) but still good enough to beat me > sometimes. I also did the graphics for someone else's project - Noughtsand > Crosses (Tic-Tac-Toe in the US) using a "Direct View Storage Tube" - likea > 'scope you could draw lines on. He got better marks than me, with aspecial > mention for the display :-( > > I thought I was an old hand until I met a guy whose first job was tuningthe > mercury delay lines (on a LEO or something) every morning with a 'scope! > > -- > Tony Whitley > GPLRank +9.99 > "On some laps I'd get as many as fifteen corners really right, but Moss,he > gets them all right on every lap" - Von Trips after the 1961 German Grand > Prix. > >

The real life times are smashed in every sim. David G Fisher "Andre Warringa" <[email hidden]> wrote in message news:[email hidden]...

Quoted text (32 lines)

> On Sat, 03 Mar 2001 06:23:59 GMT, "David G Fisher" <[email hidden]> > wrote: > > >Well, we already knew that. If you can master GPL than anything else is > >cake. Right? > >:-P > > > >David G Fisher > > I'm not that good at GPL, I can practice what I want but I will never > reach the times the fastest guys are getting. At most tracks they are > seconds faster than me. So if I can do a 1:26 at Melbourne, I wonder > what the really talented guys will get for laptimes.. > > So -to get back to the topic dear David- the laptimes in F1RC are > totally unrealistic. This year Schumi came damn close to my laptime > (yep, Schumi is good) but that's not the point right? F1RC simulates > the 99 season, so we should compare laptimes to those from 99. Right? > > So I'll stick to my opinion, if the tracks are accurate, and laptimes > are - a couple of seconds - faster in F1RC, the physics can't be 100% > realistic. Untill someone has a better theory :-P > > Andre > > > > >"Andre Warringa" <[email hidden]> wrote in message > >news:[email hidden]... > >> On Sat, 03 Mar 2001 01:04:31 GMT, "David G Fisher" <[email hidden]> > >> wrote: > >> > >> >The real life drivers have reached 1:27 in practice this weekend. Now

I'm > >> >confused. I know they were slower in 1999, but 10 seconds slower? > >Actually, > >> >when 2001 user-made cars are released for F1RC (soon after shippingI'm

Quoted text (19 lines)

> >> >sure), then the times should be pretty accurate. > >> > > >> >David G Fisher > >> > >> >> Ok, yesterday I did a 1:26 at Melbourne. That's what, 10 seconds > >> >> faster than Schumi? > >> > >> I found the results from '99: > >> > >> 99 Pole : M. Hakkinen - 1'30"462, 210.997 km/h > >> Best Lap : M.Schumacher - 1'32"112 > >> Winner 1999: E. Irvine (Ferrari) - 1h35'01"659 > >> > >> If the tracks of F1RC are realistic AND the physics are realistic, > >> that makes me a damn good driver with my 1:26, on expert level with > >> -default- setup :) > >> > >> Andre >

On Sat, 03 Mar 2001 12:14:13 GMT, Andre Warringa <[email hidden]> wrote: ... >So I'll stick to my opinion, if the tracks are accurate, and laptimes >are - a couple of seconds - faster in F1RC, the physics can't be 100% >realistic. Untill someone has a better theory :-P Well, the physics *can* be 100% realistic, in terms of its mathematical model, but if you plug in some bad numbers, you can produce wildly varying effects on laptimes. Though ofcourse, things have to be tinkered a *little* bit to get realistic laptimes. It's just things like Pacejka numbers, inertia and such are not really easy to get at all the time. But ofcourse, 'feeling' the car, the way you can correct it, says quite some things about the model. Ruud van Gaal, GPL Rank +53.25 Pencil art : http://www.marketgraph.nl/gallery/ Car simulation: http://www.marketgraph.nl/gallery/racer/

On 03 Mar 2001 05:19:18 GMT, [email hidden] (J. Todd Wasson) wrote: ... >with no clever optimizations at all, and no hand written assembly. However, >this INCLUDES running intersection and distance calculations with a few hundred >triangles per wheel, You may have come across this, but from the book Real-Time Rendering you have one of the many links at: http://www.acm.org/jgt/papers/MollerTrumbore97/ It does tri intersection, and gives you back the distance and u/v coordinates for nearly free. So you immediately have the distance to the triangle if you know it intersects. Might be a good one. Ruud van Gaal, GPL Rank +53.25 Pencil art : http://www.marketgraph.nl/gallery/ Car simulation: http://www.marketgraph.nl/gallery/racer/

On Sun, 04 Mar 2001 22:12:32 GMT, [email hidden] (Ruud van Gaal) wrote: >Well, the physics *can* be 100% realistic, in terms of its >mathematical model, but if you plug in some bad numbers, you can >produce wildly varying effects on laptimes. Like we saw in F1CS :) Funny how fellow-RASsers have improved the game so much by tweaking the values. I wouldn't be surprised if EA uses the chassis-patch values for F1CS2001 :) >Though ofcourse, things have to be tinkered a *little* bit to get >realistic laptimes. >It's just things like Pacejka numbers, inertia and such are not really >easy to get at all the time. Ofcourse. But the huge difference in laptimes of F1RC compared to the real times (99 season) made me suspicious, especially since the tracks seem to be so accurate. We're talking about 5-10 seconds difference with the real lap times here.. And ehh.. Pacejka numbers? You made that word up to impress me didn't you :) Seriously, could you briefly explain what that means? >But ofcourse, 'feeling' the car, the way you can correct it, says >quite some things about the model. And besides the AI and multiplayer problems, this would be my biggest problem with F1RC. The force feedback is lacking, and the sound of the skidding tires seems to be a tad off, making it very hard to reckognize the car starting to loose control when driving on the limit. Contrary to GPL and F1CS, where you -feel- what the car is doing, learning the tracks in F1RC is more a 'trick', where you simply remember in what gear and how fast you can take a corner. Andre >Ruud van Gaal, GPL Rank +53.25 >Pencil art : http://www.marketgraph.nl/gallery/ >Car simulation: http://www.marketgraph.nl/gallery/racer/

Hi, that's interesting news, I didn't know that about WSC! I'd like to ask you, in your model, do you calculate accelerations of the wheels or do you use the 'semi-stationary' approach (neglecting the mass*acceleration part for the wheels, the velocity being determined by spring and damper assembly only)? I just finished some coding yesterday and I use the above approximation (a standard way of doing things in physics which really makes sense in a PC racing sim as there really is not much info in doing it the 'full' way and wasting precious CPU time by doing so), and I was also surprised at how high one can go with the timestep. It really can go much above 1/100s and still remain stable, something I didn't really expect it to. Mind you, I was using the data from Millikens' book for a Champ car, and those beasts have prety stiff springs and dampeners and low inertias, so the frequencies involved are rather high. -Gregor Ruud van Gaal wrote:

Quoted text (26 lines)

> > On Fri, 02 Mar 2001 15:12:56 +0100, Gregor Veble > <[email hidden]> wrote: > > >Hi, > > > >which number are you refering to? My guess is that 1/1000 s is what WSC > >should be using, as GPL already uses 1/288 s. > > It's using a mere 100Hz (I talked to Chris yesterday), and according > to him, it can go even lower. I'm using 100Hz while developing, > because I can then turn on draw_every_integration_step and still get a > bit of motion ;-) But other than that, 250Hz is the lowest I go right > now. WSC uses some special stuff to be able to get it that low. > > > AFAIK WSC doesn't model > >flexibilty of parts coupled with their inertial modelling (only when the > >two are modelled together should problems occur, not spearately), so all > >is well :). > > I have no idea how they do this. :) I only read yesterday about the > separation of car parts; seems very cool! > > Ruud van Gaal, GPL Rank +53.25 > Pencil art : http://www.marketgraph.nl/gallery/ > Car simulation: http://www.marketgraph.nl/gallery/racer/

>You may have come across this, but from the book Real-Time Rendering >you have one of the many links at: > >http://www.acm.org/jgt/papers/MollerTrumbore97/ Thanks! I spied this link previously in one of the programming newsgroups you posted it in.... :-) Will check into it. Todd Wasson --- Performance Simulations Drag Racing and Top Speed Prediction Software http://PerformanceSimulations.Com

On Mon, 05 Mar 2001 10:48:59 +0100, Gregor Veble <[email hidden]> wrote: >Hi, that's interesting news, I didn't know that about WSC! It's a fine piece of engineering, from the looks and features of it. >I'd like to ask you, in your model, do you calculate accelerations of >the wheels or do you use the 'semi-stationary' approach (neglecting the >mass*acceleration part for the wheels, the velocity being determined by >spring and damper assembly only)? Almost all things in my sim are based on 3D forces. In the wheel case, you have a gravity force, the spring force and the damper (stabilizing) force. So the wheel hardly knows there is a car on top; it gets most of its info from the spring and damper. Not too realistic, since it's gyroscopic effects when it's rotating seems to stabilize a real car's heading motion, but well. The damper velocity is actually the only real velocity involved in the whole sim, IIRC. The rest is all accelerations based on F=m*a. So the wheel get's accelerated, both in the spinning direction, and the up/down (suspension direction). >I just finished some coding yesterday and I use the above approximation >(a standard way of doing things in physics which really makes sense in a >PC racing sim as there really is not much info in doing it the 'full' >way and wasting precious CPU time by doing so), and I was also surprised >at how high one can go with the timestep. It really can go much above >1/100s and still remain stable, something I didn't really expect it to. The good thing would be to KEEP it at 100Hz and not need more. >Mind you, I was using the data from Millikens' book for a Champ car, and >those beasts have prety stiff springs and dampeners and low inertias, so >the frequencies involved are rather high. Yes. And Euler in that case can be a nuisance. I'm still using something like: wheelVel+=wheelAcc*timeSpan; wheelPos+=wheelVel*timeSpan; but from a DirectX newsgroup, I saw the simpler version where you do: wheelPos+=wheelVel*timeSpan+0.5*wheelAcc*(timeSpan*timeSpan); Ehm, don't know if it was exactly this, but in any case you take the acceleration into account as well as the current velocity. This way, the Euler tangent of your jumpy curve gets a little more accuracy (damped if you will to the right sinodial things you expect). Which is very important for high frequencies. You could also try Runge-Kutta 4th order integration, which seems to do better. It is more complex though. And realise that WSC 'just' uses Euler (@100Hz). But also does spline interpolation for the road surface. Damn, have to get that in one day; even Monza seemed to give trouble with triangle-based tracks. I have a direction for high frequency springs/dampers which I want to go in, but I'll first have to do some testing to see where I get stuck. ;-) Ruud van Gaal, GPL Rank +53.25 Pencil art : http://www.marketgraph.nl/gallery/ Car simulation: http://www.marketgraph.nl/gallery/racer/

Hi Ruud, about Euler; while it's true that it's stability is lower than, say, with a Runge-Kutta method, it's only a constant factor that separates the two. For example, if RK becomes stable at 0.03s timestep, it may take Euler 0.01 timestep; it's never an order of magnitude difference, just a factor. As you probably know but I'll write it here for anyone who might not have the knowledge, the main difference between the RK method and the Euler one is the order of accuracy in timestep. The error in calculation is of the order of the square of timespan with Euler, while with RK it's the fourth power (or fifth, I keep forgetting). However, this only tells you something about the accuracy of the method once you're well into the regime where stability is achived, and is important for numerical calculations. For PC entertainment simulations, however, the exact solution over a longer stretch time isn't all that important, only the instantaneous proper behaviour is. Another factor to consider is that all the error estimates for Runge Kuttaand other methods rely on an analytical behaviour of all the functions, which is hardly the case in a 6DOF vehicle mechanics where a lot of singularities occur (brakes and friction reversal, loosing contact with the ground etc.). In this case, the accuracy and stabilty of RK methods is nt guaranteed and indeed it may happen that Euler is more stable and accurate in this instance. -Gregor Ruud van Gaal wrote:

Quoted text (25 lines)

> > Yes. And Euler in that case can be a nuisance. I'm still using > something like: > wheelVel+=wheelAcc*timeSpan; > wheelPos+=wheelVel*timeSpan; > but from a DirectX newsgroup, I saw the simpler version where you do: > wheelPos+=wheelVel*timeSpan+0.5*wheelAcc*(timeSpan*timeSpan); > > Ehm, don't know if it was exactly this, but in any case you take the > acceleration into account as well as the current velocity. This way, > the Euler tangent of your jumpy curve gets a little more accuracy > (damped if you will to the right sinodial things you expect). Which is > very important for high frequencies. > You could also try Runge-Kutta 4th order integration, which seems to > do better. It is more complex though. And realise that WSC 'just' uses > Euler (@100Hz). But also does spline interpolation for the road > surface. Damn, have to get that in one day; even Monza seemed to give > trouble with triangle-based tracks. > I have a direction for high frequency springs/dampers which I want to > go in, but I'll first have to do some testing to see where I get > stuck. ;-) > > Ruud van Gaal, GPL Rank +53.25 > Pencil art : http://www.marketgraph.nl/gallery/ > Car simulation: http://www.marketgraph.nl/gallery/racer/

On Tue, 06 Mar 2001 16:57:54 +0100, Gregor Veble <[email hidden]> wrote:

Quoted text (7 lines)

>Hi Ruud, > >about Euler; while it's true that it's stability is lower than, say, >with a Runge-Kutta method, it's only a constant factor that separates >the two. For example, if RK becomes stable at 0.03s timestep, it may >take Euler 0.01 timestep; it's never an order of magnitude difference, >just a factor.

Ah, I didn't know that. Thanks for the explanation. I'll use Euler then, and scrap RK from my wishlist. All the better. :) Ruud van Gaal, GPL Rank +53.25 Pencil art : http://www.marketgraph.nl/gallery/ Car simulation: http://www.marketgraph.nl/gallery/racer/

On Mon, 05 Mar 2001 09:46:26 GMT, [email hidden] (Andre Warringa) wrote: ... >>It's just things like Pacejka numbers, inertia and such are not really >>easy to get at all the time.... >And ehh.. Pacejka numbers? You made that word up to impress me didn't >you :) Seriously, could you briefly explain what that means? Sorry, this one almost slipped by. :) Pacejka is someone from the Delft University (yes, we almost seem all Dutch here lol), who's created some 'magic formula' which gives tire forces based on a lot of parameters. So you get about 8 parameters or so, stuff them in something like: Fx=sin(...sin(...atan(...))) and voila, you get the tire force. Where Fx is the tractive force btw. The same for lateral force (steering) and aligning moment (which I need to implement force feedback). The problem, ofcourse, is getting the right parameters, and they're a bit obscure. But once you have them, you can reasonably accurately simulate a tire. Which is important ofcourse. WSC uses them, and I intend to use them, as soon as I get the chance. Ruud van Gaal, GPL Rank +53.25 Pencil art : http://www.marketgraph.nl/gallery/ Car simulation: http://www.marketgraph.nl/gallery/racer/

On Wed, 07 Mar 2001 19:02:51 GMT, [email hidden] (Ruud van Gaal) wrote: >The problem, ofcourse, is getting the right parameters, and they're a >bit obscure. But once you have them, you can reasonably accurately >simulate a tire. Which is important ofcourse. >WSC uses them, and I intend to use them, as soon as I get the chance. Ok, thanks :) Andre > >Ruud van Gaal, GPL Rank +53.25 >Pencil art : http://www.marketgraph.nl/gallery/ >Car simulation: http://www.marketgraph.nl/gallery/racer/