We all know that to hold more detailed textures you need more video ram, thus a platform with more video ram can have better textures… Also, to hold very detailed models you need to have alot of system RAM, to hold all the vertex data, so a platform with more system RAM will be able to have better and more detailed models… Now, can someone please tell me how consoles can manage such decent looking graphics on such low video and system RAM?? Dont tell me about the “optimisation” om consoles, this has nothing to do with optimisation, ram on a pc and ram on a console hold exactly the same amount of data. So how does a ps2 with like 12MB of RAM display near PC quality textures when the PC has like 64 megs of video RAM, same with models…
Its…just…optimized. It doesn’t have to worry about the os or other backround tasks so everything gets pumped to the graphics.
[This message has been edited by iNsaNEiVaN (edited 09-28-2001).]
i don’t know exactly, but i guess there is some specialized hardware the reason for this. for example, i guess there’s some kind of “direct” connection between gpu and cpu and ram (without that damn pci/agp concept).
also, as iNsaNEiVaN said, the os is optimized as well (no background tasks, no or very little “multitasking”).
the other way round, all this means, you could tune your home pc to a gaming machine from hell (eg booting a special kernel for every game, or
hum, does anyone know anything specific about the ps2-gpu?
Because consoles only have to render at 640x480, not the 1024x768 minimum (like most games are shooting for). You don’t need 512x512 textures for everything at 640x480. You can get by with lower polygon models because your resolution is lower and your display device (TV) is lower quality that that of a PC.
And yes, optimizations.
Tolga, the PS2 has 2 very powerful vector units, that are fully programmable. This helps alot with geometry throughput and stuff. You can even create proceduaral stuff on the fly with these.
I dont know much about the actual rasterizer though.
As for consoles in general, you really do have to optimize quite alot in order to squeeze stuff down. Using bit fields etc… which are probably rarely seen in a modern PC game.
Alot of textures on consoles are still 4 bit, 8 bit, or probably 16 bit at most. 32 bit textures on a console really do have to be justified, to the limited amount of VRAM consoles have.
You’ll also find that most PC coders are just lazy, and the code is often not particularly memory efficient.
P.S. Dont flame me about the PC coders bit, this just seems to be common knowledge in the PC to console conversion buisness. It does take a slightly different mental attitude to console programming than PC programing. With it being fixed spec, virtually no OS, no other apps at the same time etc…
Also theres the situation with patches. (sorry for going on a rant here) but how many PC games get followed with patches nowadays… With console games, the game has to be crash proof, and virtually bug free (no gameplay issue bugs) at the point of submission, else you will just fail!
It was quite funny noting M$ docs on the difference in standards between consoles and PC games…
Thanks for all your input… Several of you said optimisation but what does optimisation have to do with ram?? no matter how much something is optimised you will still need raw memory to hold vertex and texture data, something consoles lack… I suppose using lower polygon counts and less detailed textures explain it…
Also, some PC coders arse lazy, but lookj at people like Carmack (hes my hero)… also, concentrating less on code but more on design produces a better game in my opinion…
A model, or it’s textures, doesn’t have to be high-res to look good. A wellmade model along with a good texture will result in very good looking graphics, even though it’s low-res.
If you scale your textures upwards with a factor 2, 4 or even 8, does it make it look better? No, it makes it consume more memory, thats all. See my point?
tnx for that info nutty…
but one thing to add: we all shouldn’t be too lazy with our resources. we’d rather use them for better (nicer or faster) scenes…
nutty, why don’t you write a tutuorial about optimizing opengl programs???
i’m sure, a lot of people would benefit from your knowledge…
Well, you see , you need to look at the difrence between console machines and pc’s.
There is quite allot.
The console was built for graphics and gaming.
It’s possible that the video memory of the graphics card is probably shared with system ram.
It’s not just optimization software (code) wize. It’s hardware wize too. The XBox is made for graphics, not the pc.
So there is allot of things they can take advantage of on consoles.
Just my 1c, not exact or anything
Also, note that console processors and data buses are much more powerful.
example: the PS2’s core CPU is 256 bit, while the PIV is effectively 32 bit with 64 bit extensions, and the G4 is also 32 bit with a few SIMD 128 bit instructions.
the PS2’s main data bus that connects VRAM and other units is effectively 2560 bits wide, where usual PC data paths are 32 or 64 bits wide.
thus, more geometry can be calculated, instead of stored statically in RAM.