ps2 and ngc technical details

[kopio0]

when both systems were announced, there were many talks about their
capabilities. sure, years later the ps2 proved its critics wrong. raw poly and smart artistic textures/effects are the winning formulas. but, all of sudden the debates and questions of each systems' hardware just disappeared.
does the ps2 offers texuture compression (with it mpeg decoder?) or its majority pallette based?
does flipper offers any custom features along with its fixed functions?
how efficient is gecko compared with sh3, ee, p3?
yeah i know most can be observed with the games, but i'm just curiously
in need of a good technical analysis of these technologies. i seriously think it is more interesting than any of its games...

 

[London Geezer]

Not trying to spoil the party, but this has been debated for death, maybe you want to search old threads? Much easier than having the same 10-page discussion all over again for the Nth time...

EDIT: Oh, and i think u meant SH4 if u were talking about DC. I believe the SH3 was in Saturn (or in some other old Sega machine).

Also, like we said a long time ago, it's a bit tricky to campare EE to anything the other consoles have, since T&L is done on the EE itself while it's done on the GPUs on GC and Xbox. The EE core (without counting the VUs) might be the only thing you can compare to the XCPU and the Gekko, but even then i'm not sure of the validity of any comparisons.

The best thing would be to compare EE+GS to Gekko+Flipper and XPU+NV2A. And we can all see the differences in both the focus and the limitations of each architecture.

(Just thought i'd give something to the thread)

 

[Nick Laslett]

I would perform a search for Performance Analyser and my name.

There is a thread which dicusses a SCEE presentation about the findings from a lot of Performance Analyser reseach. This includes games up to Return of The King, Jak2 and ZOE2. The actual author of the report posts in the thread.

The results are quite shocking really, but you can extrapolate them to the other consoles as well. Making the assumption that the Xbox and GC are being held back by constraints of the PS2.

The outcome is that dev's have hardly began to scratch the surface of maxing this generation of consoles out, if their original performance specs are to be used as a benchmark.

Highlights:

Rendering Analysis:

Average: 52,000 polys per frame
Maximum: 145,000 polys per frame

That equals 8.7 million polys per sec with a title running at 30fps
The best performance was on a title that pushed 7.5 million polys per sec running at 60fps

(These numbers are averages for a game not peaks. The PA counts the polygons rendered on screen, not how many were sent.)

Framerate: 60% were running at 25/30 or less

95% were using full height buffers. (This means most PS2 games could run in prog scan.)

VU0 Usage:
Average: 2%
Best performing games: 8%

VU1 Usage:
Average: 56%

These are the spec sheet performance numbers for the 3 consoles as I remember them. (edit 11th May)

GC - 12million
PS2 - 22 million
Xbox - 30 million

The other point that has come across strongly in my time on this board is that all three consoles are constrained by the CPU, which hampers physics, game logic and AI.

It is clear to me that there are still significant advances to be had on the current systems and it would be better if the next gen was delayed. (The development tool and programming tricks are know for the current machines. Another software iteration after the games being shown at this E3, I'm sure would look impressive.)

 

[Teasy]

My reference numbers for achievable Polygons per sec based on various technicals discussions here:

GC - 12million
PS2 - 22 million
Xbox - 30 million

Which discussions around here bring you to those conslusions? Your GC estimate is far to low.

 

[London Geezer]

God.... Not again!!!!!!

 

[Teasy]

God.... Not again!!!!!!

Oh stop being such a drama queen

 

[Qroach]

Haha lol, I noticed you changed your response...

Your GC estimate is far to low.

Anyway, what brings you to that conclusion? Before you edited your posts, you seemd to think neither xbox nor PS2 could read the numbers he quoted. I'm not sure what gives you the idea that the gamecube numbers are wrong... but it's a known fact that flipper with it's fixed funtion T&L is the weakest of the three consoles when it comes to pushing poly's.

If I recall correctly, nintendo quoted 12 million fully featured poly's in the original documentation that shipped to developers before gamecube was released.

 

[Teasy]

 

[Panajev2001a]

Qroach, that T&L unit in Flipper is also quite good at Lighting as the number of polygons per seconds drops by a much slower rate than it does on other consoles if you read those performance figures.

 

[Teasy]

GC has already surpassed 12 million pps in game and reached 12 million pps in a first generation title. That, and knowledge of GC from past discussions ect, is what brings me to that conclusion.

Before you edited your posts, you seemd to think neither xbox nor PS2 could read the numbers he quoted.

I thought some might see it like this, which is why I edited. All I was trying to say is that they haven't come close to those numbers in game at this time. While GC has surpassed the number mentioned. Which makes the estimated numbers a strange conclusion to come to IMO.

it's a known fact that flipper with it's fixed funtion T&L is the weakest of the three consoles when it comes to pushing poly's

Its weakest in terms of T&L flexibility, not neccesarily speed. Plus its the system as a whole that matters here not a single component.

P.S. Yes we all know about Nintendo's 12 million number, as you said, its fully featured though, not the most acheivable in game.

 

[a4164]

Nintendo also had this in their official GameCube Hardware Overview documentation (Flipper@200mhz):

 

[Teasy]

About Flipper's lighting engine. From a GC Gamasutra article:

In addition, one directional and one ambient light on the Nintendo Gamecube are guaranteed to be computationally for free. Therefore, that decision does not impose a performance penalty (strictly speaking, as soon as one starts to use more complex shader setups, even more hardware lights come at no performance penalty, because the graphics processor computes light values in parallel to other things).

 

[Qroach]

Panjev,

Qroach, that T&L unit in Flipper is also quite good at Lighting as the number of polygons per seconds drops by a much slower rate than it does on other consoles if you read those performance figures.

Well that's true, but that doesn't mean it still isn't the weakest of the three regardless of how fast performance drops... it still doesn't output as many polygons as PS2 or xbox.

The best comparrison of all three hardware platforms was provided by EA. If someone could find that article cmoparing all three, I think it would shed some light on the question.

 

[Phil]

Meh. What's the point? There are realtime demo's that have PS2's number higher than what it has achieved in games so far. Why even compare x console's with y's perforamance figures to y console running game z?

There's no frickin point. GameCube can theretically do that many polygons, PS2 that many [insert number] and Xbox so many. Draw your conslusions from there. There really isn't any point arguing what developers achieved as they all like to emphasize on different aspects which nullfies any comparasment on this subject.


Quincy:

EDIT: Ugh, please leave that article out of this. It was utter rubbish and besides, since when is EA's developments a benchmark on anything?

 

[Qroach]

Well I partially agree with you. There really isn't any point in arguing. At least when one company does proper benchamarking for all three consoles, there really shouldn't be any reason to continute arguing.

Yet, somehow it continues to happen, over and over again.

 

[Qroach]

Um, what was rubbish abotu it again? As I saw it, it was faily comprehensive. Far better than going off numbers provided by Nintendo MS or Sony, and far better than the eyeball judges around here.

Not only that, they benchmarked it using game code for the most part, so I really don't see how it could be rubbish

 

[Phil]

well, the article or the benchmark itself perhaps wasn't rubbish, but more the comparasment itself or the result it implies. Basically, why even take a benchmark of one developer for granted? We don't know how efficiantly coded that benchmark was to beginn with, so why even bother mentioning it?

 

[Teasy]

Well that's true, but that doesn't mean it still isn't the weakest of the three regardless of how fast performance drops... it still doesn't output as many polygons as PS2 or xbox.

It doesn't output as many polygons in what scenario though? Yeah the fact that GC doesn't drop poly performance near as much as the others when effects are added doesn't mean its not the weakest poly pusher. But at the same time the fact that the other consoles output more polys at peak with few to no effects doesn't mean GC is the weakest. In the end its actual acheivable output in games that matters, not theoretical peak performance.

As has been said there's not much point in arguing which console is the strongest and weakest at pushing polys. Especially since each one has its strenghts in different situations. However I think its pretty obvious that, in game, PS2 cannot push twice as many polys as GC. To suggest otherwise is crazy AFAICS. And frankly AFAIR even the EA benchmark backed this up.

 

[Qroach]

Yeah the fact that GC doesn't drop poly performance near as much as the others when effects are added doesn't mean its not the weakest poly pusher.

Well I'm glad we can agree on that at least.

But at the same time the fact that the other consoles output more polys at peak with few to no effects doesn't mean GC is the weakest.

AFAIK the game cube outputs the least amount of polygons when it's stressed. I said before Gamecube is teh wekest poly pusher of the three from what I recall. I was talking strictly about polygons.

However I think its pretty obvious that, in game, PS2 cannot push twice as many polys as GC. To suggest otherwise is crazy AFAICS. And frankly AFAIR even the EA benchmark backed this up.

I agree again. I didn't claim it could push twice as many polys as the gamecube, but from what I've seen PS2 can beat the gamecube in raw poly throughput. Although raw poly's don't make a complete game, but I was just stating that one point. That's all I was saying...

really here the question is where you want to get your information from. If you ge tthe marketing numbers it will say the same thing, and if you look at what developers are saying about the hardware, you'll get the same answer I'm faily certain.

- edit -
Anyway, does anyone have a link to that EA benchmark? I'm curious what it said inregards to this question again. I'm only 50% sure that it follows what I've been saying. It's been a while since I read it....

 

[Phil]

However I think its pretty obvious that, in game, PS2 cannot push twice as many polys as GC. To suggest otherwise is crazy AFAICS.

Why crazy? Didn't we just basically agree that it's up to the developer to make use of each console's strength? Given the theoretical performances of each console, it isn't hard fetched to see how PS2 could infact push double the polygons. In-game after all is a very subjective term.. in the end, a demo is "in-game" too and just as valid as the raw-specs of any game really.