PSX5Central
Playstation/Gaming Discussions => PS3 Discussion => Topic started by: clowd on June 18, 2002, 03:12:41 PM
-
I know the PS2 can generate 66 million polygons, but my question is why does it use only 4,000 for characters?
If it uses only 4,000 for characters why cant there be 100 people onscreen, using only 400,000 polygons of the 66,000,000? (or 10,000,000 if you add all the effects)
Example GT3, there is only 6 cars sporting 5k polygons each, why cant there be 20?
What is using up all the polygons?
-
Originally posted by Clowd
I know the PS2 can generate 66 million polygons, but my question is why does it use only 4,000 for characters?
If it uses only 4,000 for characters why cant there be 100 people onscreen, using only 400,000 polygons of the 66,000,000? (or 10,000,000 if you add all the effects)
Example GT3, there is only 6 cars sporting 5k polygons each, why cant there be 20?
What is using up all the polygons?
PS2 can render 66mpps with no effects.. It can render 25mpps with all effects.
Why the devs only use 5k/car in GT3? Well, Im sure they could use alot more if they wanted to.. But why would they? It would almost be a waste of power (or polys).
Compare a 5 000 polygon GT3 car to a ie. 15 000(?) polygon PGR car.. Which looks better? They look pretty close no?
And with PS2 its not like the developers just can start using 100% of the power from day one.
-
Pretty much on the same lines as what Fatson said.
1. There are a lot of polygons used on other parts of the game, not just the characters and cars. IE: 3d backgrounds. R&C would seem to be a good example of this.
2. There are other processes which take up system power that also run in the background of the system.
3. I doubt there really are that good of developers out there who can write an entire game in ASM or even write a fully efficient C++ compiler which would compile to utilitize 100% of the power.
I\'m not sure if you\'ll understand point 3, but its most likely one of the major reasons. No program out there that takes 2 years to write will fully utilize every aspect of a console. As years pass on, each generation of games will utilize the power more efficiently, but will never completely utilize the entire system\'s power.
-
Those are raw polygons used to show capable specs, but not in a real world or game situation
It takes power to render all those polygons , its taskes power to add textures, lighting and FSAA
It takes more power to ad A.I. physics and other gameplay mechanics
there is alot more to it then just rendering polygons
-
Yeah, those polygons are just estimations, not real world numbers.
And it\'s 66,000,000 polygons/sec, so at 60 frames per second it\'s like 1,100,000 of real polys on screen, because you\'ve to recalculate their position each frame.
So, your example is more like:
6 cars x 5,000 polys x 60 frames = 1,800,000 polys/sec
That\'s just for the cars; then add the polys for the backgrounds (most of the polys in GT3 are being used for the roads for making the physics calculations more realistic), the lightning, the multiple layers of textures, the physics, the AI, the reflections, and so on.
GT3\'s far from maxing out the PS2 though.
PS: Yeah, I\'m bored today.
-
You also have to remember that modern videogames use
to some extent NURBS and Bezier surfaces. Those are a bit
more calculation intensive than "simple" poly meshes. Add
in the power you need for the physics and such ...
-
Does anyone know what does the pixel fill rate has to do with polygons?Cause I rememder a long time ago someone explaining in the forums that the PS2 can do more polygons in real time gameplay scenes than the XBOX because of its higher pixel fill rate.I still dont understand this.(?)
:confused:
The games proove otherwise.
Also if you check the specs you will notice that the EE is capable of handling 66m while the GS 75m.I also dont understand that one.2 separate parts of the hardware each with its own strengths?How does the EE and the GS work together their specs?
BTW Polyphony said that GT3\'s graphics engine can push 20 millions polygons per second although the game is running in much less than that.
-
Originally posted by Unicron!
Also if you check the specs you will notice that the EE is capable of handling 66m while the GS 75m.I also dont understand that one.2 separate parts of the hardware each with its own strengths?How does the EE and the GS work together their specs?
EE can render 66mpps+, GS can draw 75mpps MAX.
EE renders (transforms), GS draws them on the screen.
BTW Polyphony said that GT3\'s graphics engine can push 20 millions polygons per second although the game is running in much less than that.
Yes that is what the limit of the GT3 engine was/is. It does not mean that they must use it to the limits.. GT3 still looks fanstastic.
Its about optimisation.
-
Originally posted by Unicron!
Does anyone know what does the pixel fill rate has to do with polygons?Cause I rememder a long time ago someone explaining in the forums that the PS2 can do more polygons in real time gameplay scenes than the XBOX because of its higher pixel fill rate.I still dont understand this.(?)
:confused:
The games proove otherwise.
Ok, check your pm.
I don\'t want to start one of those "my console renders more polys than yours which makes my thingy WAY bigger than yours too" here.
-
Originally posted by Unicron!
Does anyone know what does the pixel fill rate has to do with polygons?Cause I rememder a long time ago someone explaining in the forums that the PS2 can do more polygons in real time gameplay scenes than the XBOX because of its higher pixel fill rate.I still dont understand this.(?)
:confused:
The games proove otherwise.
Also if you check the specs you will notice that the EE is capable of handling 66m while the GS 75m.I also dont understand that one.2 separate parts of the hardware each with its own strengths?How does the EE and the GS work together their specs?
BTW Polyphony said that GT3\'s graphics engine can push 20 millions polygons per second although the game is running in much less than that.
I\'ll leave out Xbox so it doesn\'t turn nasty... so here it goes:
The Emotion Engine, which is well known the CPU of the PlayStation 2, consists of various different units, all dedicated to different tasks. The more complex and powerful ones are with no doubt the two Vector Units. Then, fairly close to the EE, you have a GPU (Graphics Processing Unit) which can be compared to todays 3d graphicscard. The difference though is, is that GPU and CPU are totally independed from another in the PS2, while in PC\'s they both share certain tasks.
The Emotion Engine (specifically Vector Unit-1) uses its power to perform perspective transformation. That term can be simplified into recalculating those polygons while you move your character in different directions or play your game. This task is very performance consuming and that\'s where the 66 million poly/sec number comes from: The EE (or Vector Unit 1) can calculate 66 million polygons per second.
The GPU (Graphics Synthesizer)\'s primary function is just drawing those polygons on your tv screen while adding some effects such as AA and others. Drawing capabilities is messured in fillrate, which gives you a number of how many pixels a processing unit can draw per second. The GS has a (pixel-)fillrate of 2.4 Gigapixels/sec - which would equal 2400 million pixels per second. This is raw however and does not include texture mapping etc. So, polygons also consist of pixels and if we take 32 pixel polygons and devide that number from the fillrate, you\'ll get 75 million (32 pixel) polygons per second. A game of course doesn\'t only use 32 pixel polygons - but remember these are just estimates and only determin the maximum capabilities of a system under certain circumstances. So, as Fast pointed out:
the EE can perform perspective transformation on 66 million raw polygons per second, while the GS can effectively draw (or render) 75 million (32 pixel) polygons per second.
To also lighten a few things up, the 66 million poly/sec number is only the maximum performance of one vector unit (VU1), while the other vector unit (VU0) is dedicated to different tasks. You could use both for perspectice transformation theoretically, but by doing that, you\'ll neglect other import gameplay elements which also consum performance. While this is a very powerful environment for 3d applications, it is also very hard to implement in the software. Just think about it; you have a CPU that consists of two vector units that have to run in parallel. If it doesn\'t, then you\'re whole program won\'t work, as latency will cause bottlenecks. 1st generation games only made rare use of one of the vector units and efficiancy was also very low. That together with the steap learning curve and we get the result of graphical lackluster games at launch. GT3 perhaps could be also an example, as it only maxes out the PS2 to less than 25% (which is proved by messurement). Also, as others have pointed out - a game does not only consist out of cars, but also a lot of physics, AI, textures, lightning and others which all consum performance. The more efficiant the game is programmed, the more you\'ll have on screen. Hope this clears most questions up in here - if not, feel free to ask. ;)
-
Thanks.
I ve been checking and reading a lot about PS2\'s tech information since its first appearance but I ve never understood completely how everything work(thanks to u now I understand:)) due to my lack of tech knowledge and some of my English.
Sorry if I sounded like starting an "XBOX vs PS2" debate.
Hope this clears most questions up in here - if not, feel free to ask.;)
It surely clears them all.:)
Thanks again.