View Full Version : Is Flipper's geometry engine more powerful than Elan?
Luminescent
16-Dec-2002, 13:55
Overall, how does the Flipper's geometry engine compare to Elan, geforce 2, and other hardware containing hardwired T&L (not sure if Elan is a hardwired implementation)?
I think the elan can do 10 millon pollygon per sec with 8 light sources sustained . I have no clue how that compares to anything .
Tagrineth
16-Dec-2002, 15:27
Overall, how does the Flipper's geometry engine compare to Elan, geforce 2, and other hardware containing hardwired T&L (not sure if Elan is a hardwired implementation)?
Well I can say this much, Flipper beats the hell out of a GeForce2... :lol: :roll:
Elan can do 10 million polygons per second with 6 light sources:
http://www.segatech.com/arcade/naomi2/index.html
Not sure about GameCube:
http://www.segatech.com/gamecube/overview/index.html
But Virtua Striker and Beach Spikers were Naomi2 ports and they seem to run fine on GameCube.
CaptainHowdy
16-Dec-2002, 17:17
Elan can do 10 million polygons per second with 6 light sources:
http://www.segatech.com/arcade/naomi2/index.html
Not sure about GameCube:
http://www.segatech.com/gamecube/overview/index.html
But Virtua Striker and Beach Spikers were Naomi2 ports and they seem to run fine on GameCube.
GC is supposed to be 10-12 mil with 8 light sources
I could be off,its been a while since I have looked it up..
I THINK thats what it was, so its pretty close.
202.5MHz Flipper:
http://cubemedia.ign.com/media/news/image/gamecubefaq/polygon-perform-table.jpg
Overall, how does the Flipper's geometry engine compare to Elan, geforce 2, and other hardware containing hardwired T&L (not sure if Elan is a hardwired implementation)?
Well I can say this much, Flipper beats the hell out of a GeForce2... :lol: :roll:
I'm not so sure about that.
Simon F
16-Dec-2002, 18:03
That figure of 6 lights for Elan is for "full complexity" lights (i.e. position, focus direction, distance atten etc etc).
You will often find marketing figures for some chip will quote a directional light which amounts to a dot product and a scale, so make sure you're comparing like with like.
I'm not so sure about that.
Show me a game with anywhere near the polygon counts and complex lighting of Rogue Leader running on a Geforce 2 at a decent framerate.
Rogue Leader must have around 5+ lights with around 250,000 polys per frame at 60fps.
Yeah you have to take into account that this is a closed system vs a PC, but still I think if a theoretical test on Geforce 2 can't even come close to a real game on Flipper its obviously Flipper has faster T&L.
EDIT: Sorry zurich, I edited this before I realised that you'd already replied to me.
I'm not so sure about that.
Just look at Geforce 2 T&L tests, like for instance 3dmark2001. Geforce 2 gets 2.9 million pps with one directional light and 7 point lights with no textures. I bet Rogue Leader, a first gen game, is pushing 15 million pps with 4-5 lights and upto 6 (or even more) textures. It looks to me like Flipper is a hell of allot more powerful then Geforce 2.
I'm not disagreeing, but you have to take into account the OS, HAL, DirectX, etc. overhead involved with benchmarking a video card.
"full complexity" lights
And this is really what makes comparison impossible.
Flipper has a somewhat "simplified" lighting model, you can't really do an apples to apples comparison with either Elan or GF2 for that matter.
overclocked
16-Dec-2002, 18:39
Teasy Posted: Mon Dec 16, 2002 6:17 pm Post subject:
--------------------------------------------------------------------------------
Quote:
I'm not so sure about that.
Show me a game with anywhere near the polygon counts and complex lighting of Rogue Leader running on a Geforce 2 at a decent framerate.
Rogue Leader must have around 5+ lights with around 250,000 polys per frame at 60fps.
Yeah you have to take into account that this is a closed system vs a PC, but still I think if a theoretical test on Geforce 2 can't even come close to a real game on Flipper its obviously Flipper has faster T&L.
EDIT: Sorry zurich, I edited this before I realised that you'd already replied to me
When will Factor5 reveal their new game... :o
Show me a game with anywhere near the polygon counts and complex lighting of Rogue Leader running on a Geforce 2 at a decent framerate.
Ok. How are the two comparable? How do you compare what any PC games designed around what the Geforce 2 can directly compare to the features seen in RL? Is RL really that complex? How many polygon's is it pushing?
Rogue Leader must have around 5+ lights with around 250,000 polys per frame at 60fps.
Really? Did you count the polygons Teasy? Did you count the lights? That number for polygons sounds absolutely ridiculous. 250,000 perframe? Show me a resource that suggest this. I am willing to wager this game only runs around 60,000 - 100,000 ppf. 250,000 is a totally outlandish number.
Yeah you have to take into account that this is a closed system vs a PC, but still I think if a theoretical test on Geforce 2 can't even come close to a real game on Flipper its obviously Flipper has faster T&L.
??? I don't see how you can make such an assumption. I haven't seen a trustworthy source for GC stats. Most of what i have seen are nothing more than bs figures presented by factor 5.
Legion,
Don't forget that Flipper's T&L is fixed function, meaning it lends itself quite nicely to geometric shapes and such (ie: Star Wars is a nice showcase for this). If it is approaching 250k per frame, then I wouldn't really be surprised.. hardwired T&L is built for speed (lower latencies, predictable data), but has its limitations. IMO, Rogue Leader was tailored around these limitations.
Legion,
Don't forget that Flipper's T&L is fixed function, meaning it lends itself quite nicely to geometric shapes and such (ie: Star Wars is a nice showcase for this). If it is approaching 250k per frame, then I wouldn't really be surprised.. hardwired T&L is built for speed (lower latencies, predictable data), but has its limitations. IMO, Rogue Leader was tailored around these limitations.
whatever. How does he know the polycount? Who ever claimed that many polygons were on the screen? it sounds like nonsense. Furthermore how do you compare fixed function T&L lighting to programmable high complexity lights?
Steve Dave Part Deux
16-Dec-2002, 19:16
Aren't the lighting models pretty much fixed for DX8 class hardware as well, as far as what can and can't be practically used with a particular shading model?
How do you compare what any PC games designed around what the Geforce 2 can directly compare to the features seen in RL? Is RL really that complex? How many polygon's is it pushing?
I don't really understand that question. But Rogue Leader runs at between 12-15 million pps.
Really? Did you count the polygons Teasy? Did you count the lights? That number for polygons sounds absolutely ridiculous. 250,000 perframe?
No I don't need to, I've seen numbers from the developer, which is backed up by the incredible complexity of the game. I also remember Factor 5 mentioning the number of lights but can't remember exaclty what they said. But then just look at the game, the lighting is spectacular.
Show me a resource that suggest this. I am willing to wager this game only runs around 60,000 - 100,000 ppf. 250,000 is a totally outlandish number.
Yeah right, Rogue Leader pushes only 3.6m - 6m pps? That's outlandish mate.
I could have a look for some interviews for poly pushing info, but as I remember it was at least 12 million pps. Also remember back when EA measured T&L performance with one of its engines? GC was around the same performance as a Athlon 1.4ghz and Radeon 8500.. which is a hell of allot faster then a Geforce 2.
whatever. How does he know the polycount? Who ever claimed that many polygons were on the screen? it sounds like nonsense. Furthermore how do you compare fixed function T&L lighting to programmable high complexity lights?
In loads of threads here people have mentioned that Rogue Leader is pushing upto 12 million pps, its from Factor 5 themselves. How did you miss that? Also why would anyone need to compare fixed function to programmable?.. we're talking about Flipper vs Geforce 2, both fixed function.
CaptainHowdy
16-Dec-2002, 19:22
So, if Gamecube is a GF2, what does that make PS2? a voodoo 2?
more realistically
PS2=GF2(except the GF2 was capaple of bumpmapping, and anisotropic filtering)
Gamecube=Radeon 8500
Xbox=GF4 TI
Aren't the lighting models pretty much fixed for DX8 class hardware as well, as far as what can and can't be practically used with a particular shading model?
I would imagine what most hardware does is comparable to what the GF2 does.
However Flippers T&L is very basic, the lighting model in particular, is just enough to get the job done. The difference between what GF2 considers a local light and what Flipper considers a local light are significant. You could probably turn enough stuff off on the GF2 light to make them comparable but I don't know of any benchmark in such a situation.
ERP
Nothing I've seen Geforce 2 do, even in tech demo's, comes close to what Flipper does in Rogue Leader. I really don't know what you mean about its lighting being simple compared to Geforce 2. The lighting in Rogue Leader is fantastic, that certainly doesn't look simple to me. Everything I've seen, from benchmarks to just looking at the games tells me that Geforce 2 cannot stand up to Flipper in T&L performance. So you'll forgive me if I don't just take your word on this.
BTW, what a few people here are forgetting is that we shouldn't just be taking raw T&L power into account here, but also the efficiency. Lets see what Geforce 2 is pushing with 4 textures (having to do two passes) vs Flipper, or even 6-8 textures (after Geforce 2 has taken 3-4 passes and Flipper only one).
I don't really understand that question. But Rogue Leader runs at between 12-15 million pps. Most of the time it runs at 60fps.
Really? Did you count the polygons? Who claimed this?
No I don't need to, I've seen numbers from the developer, which is backed up by the incredible complexity of the game. I also remember Factor 5 mentioning the number of lights. But then just look at the game, the lighting is spectacular.
Yeah we all know that factor5 is a credible source :lol:. Seriously, show me something that indicates there are that many polygons on screen.
LOL.. yeah right, Rogue Leader pushes only 3.6m - 6m pps? That's outlandish mate. I could have a look for some interviews for poly pushing info, but as I remember it was at least 12 million pps at around 60fps. Also remember back when EA measured T&L performance with one of its engines? GC was around the same performance as a Athlon 1.4ghz and Radeon 8500.. which is a hell of allot faster then a Geforce 2.
Really? Its outlandish? and 12-15 million isn't? Why, because you want to be that it is 15 million/second out there teasy? How do you know there are that many out there? Do you always believe what you are told?
EA's benchmark :lol:. The one people here keep saying is inaccurate? Teasy you and i don't even know what was happening in that benchmark. Again i can't imagine the two are directly comparable. If you really want to go that way the Xbox was blowing away the athlon 1.4 GHZ/Radeon 8500. It makes me wonder if they had tested the xbox verse that same computer with a geforce 4 if it still wouldn't have beaten it. I have a hard time believing that teasy.
I want to see some real figures.
Teasy how do you know what you are seeing is or is not as complex as you think?
Back to top
Nothing I've seen Geforce 2 do, even in tech demo's, comes close to what Flipper does in Rogue Leader. The lighting in Rogue Leader is fantastic, that certainly doesn't look simple to me. I really don't know what you mean about its lighting being simple compared to Geforce 2. So you'll forgive me if I don't just take your word on this.
I'm not talking about application here, just implementation.
Basically Flipper isn't doing the same Math as GF2 when it's doing a "full featured light". A light on GF2 is not comparable to a light on Flipper. Flipper's lighting model is grossly simplified.
I'm not comparing software on the platforms and I'm not comparing performance I'm just saying that I know of no apples to apples comparisons.
Yeah we all know that factor5 is a credible source . Seriously, show me something that indicates there are that many polygons on screen.
Your asking me to show you how Rogue Leader is pushing that many polys but you refuse to believe Factor 5's numbers.. the people who actually made the game! So then who's numbers are reliable for any game? Every developer has a reason to make their game sound good, so we shouldn't believe any numbers from any developers and instead just make the numbers up ourselves as you just did? It seems that your just setting me a impossible task just so you can still believe that your right here.
BTW you keep asking me if I counted the polys on screen, yet you then come up with numbers that have came from no source at all (60,000-100,000 per frame). So did you count all those polys then? At least I get my number from a real source.
Teasy you and i don't even know what was happening in that benchmark. Again i can't imagine the two are directly comparable.
So now EA are bias towards GameCube are they? They compare their engine in T&L power on different systems and you say the benchmarks aren't comparable?? They may not be exact and optimised as much as possible for each system, but they are some indication, and they're are th best indication we have since you refuse to believe numbers from Factor 5.
I want to see some real figures.
Do you only count real figures as ones you either make up or that prove your point? I've already shown you numbers as real as they get, now you show me some sort of source to disprove them.
There is an interview at gamespot where they say exactly the interval (polygon number fluctuate did you know ?) and they said that the game was most of the times at 60fps.
I think people got the 15Mpol by multiplying both numbers, which I think is not fair.
I'm too lazy to get the link :(
RL has more that 200k pol by frame, be it a 30 or 60fps I do not know.
Oh my that IS rediculous. Your asking me to show you how Rogue Leader is pushing that many polys but you refuse to believe Factor 5's numbers..
If i can remember they have made many BS statements in the past. That effects their credibility Teasy.
the people who actually made the game!
Did you know on sega's first page for VF4 they claimed the game was running 60,000,000+ polygons/second? How do you know that is BS teasy? I mean afterall they are the game's developer.
So then who's numbers are reliable for any game?
Teasy because they make claims about their games you are willing to believe them? You have to first substantiate their claims to be reliable before you can compare them to anyone else. If you want to do then you probably need to find other sources to do that. This forum has a number of developers and people with incite. Perhaps we can ask them what they know about RL. Simply trusting one source doesn't cut it for me when that source has checkered history.
BTW you keep asking me if I counted the polys on screen, yet you then come up with numbers that have came from no source at all (60,000-100,000 per frame). So did you count all those polys then? At least I get my number from a real source.
Teasy don't be ridiculous. 60,000 - 100,000 is probably about average for games. What was stating was nothimg more than that - that i have a hard time believing the polycounts are dramatically higher then the average. I have seen no evidence for this and find your claims hard to swallow without further evidence.
Your really in no position to be talking like that. Do you only believe what you want to believe? Common, at least I got my numbers from the devs themselves, were as you pulled your numbers out of your arse.
Really? I am not? Why is that? A developer makes a claim so it must be true? Come on Teasy. I am not trying to fight with you here. I am asking you to do something simple - back up your claims. How do you know Factor 5's assertions are what you think they are? How do you know they are honest? They have bsed in the past numerous times. In fact you can scout around these forums and find people discussing what they have said.
Again sega's webpage claimed their game was running 60,000,000 polygons/second. This must be true then?
So now EA are bias towards GameCube are they?
WTF? What does this have to do about bias? I mentioned their outlandish figures for the xbox as well. I have a hardtime believing both. Those same benchmarks had the PS2 beating the GC in effective texturing with 8 layer textures as well. Do you believe this?
They compare their engine in T&L power on different systems and you say the benchmarks aren't comparable?? They may not be exact and optimised as much as possible for each system, but they are a good indication.
That sounds damn arbitrary to me. How do you know their engines properly utilize each system? How do you know the behavior of these engines are indicative of the system's performance as a whole?
I mean common, you don't believe Factor 5 on their own game, you don't believe EA, it seems you just don't believe anyone who shows that Flipper's T&L power is above and beyond Geforce 2.
Back to sega. They made some goofy claims on their webpage about VF4. Should we believe them on face value?
Do you only count real figures as ones you either make up or that prove your point? I've already shown you numbers as real as they get, now you show me some sort of source to disprove them.
Its a bit odd that you are accusing me of this when it seems you were the one who was making claims.
DeathKnight
16-Dec-2002, 20:35
Flipper's raw T&L power is pretty evenly matched with a GF2. However, Flipper's internal workings are more than likely much more efficient than the GF2. Couple that with a closed architecture and you can certainly push more polygons in a game scenario (littered with effects).
I love how everyone is ignoring ERP's comments. If the math in the lighting implementation is different between cards, how are you supposed to compare.
Arguing by anecdote is hardly worth much weight, especially when there are proffessionals present who have actually worked with the console.
::Mutters:: Religion!
Exactly fred. of course, some people will believe what they want to, no matter what's been said.
I love how everyone is ignoring ERP's comments. If the math in the lighting implementation is different between cards, how are you supposed to compare.
Arguing by anecdote is hardly worth much weight, especially when there are proffessionals present who have actually worked with the console.
::Mutters:: Religion!
I agree. This is not the first time he has said something like this either. He is usually ignored.
I don't know if this matters at all, but Factor 5 claimed that the Death Star in RS2 was made up of at least 100,000 polygons.
DeathKnight
16-Dec-2002, 21:03
Not all being drawn at once. The highest polycounts on screen for RL reach around 250,000. Don't know if it still keeps a steady 60fps in those peak situations though.
Legion, there are more than likely a ton of games above the "average" polycount. Azurik on the Xbox *peaks* at 500,000 polys a frame (at 30fps). It's not uncommon to have higher polycounts in this generation of consoles.
Legion
Can you show me an interview were Factor 5 lied?, or any links at all where they lied? I'm serious, I'd like to actually see quotes and links so I can find out once and for all what happened there.
Teasy because they make claims about their games you are willing to believe them?
Many times in the past it has been said on this forum that Rogue Leader pushes 250,000+ per frame, nobody else argued with that, not even any of the devs on this forum.
Those same benchmarks had the PS2 beating the GC in effective texturing with 8 layer textures as well. Do you believe this?
Hmm, where did you see this? Could you post it, because I didn't see that.
Back to sega. They made some goofy claims on their webpage about VF4. Should we believe them on face value?
Did Sega ever claim in an interview that they were actually pushing that number? Did you go into specifics on how many polys were onscreen per frame at what framerate ect? Or was this a silly off hand, and probably confused, claim from a PR guy who wouldn't know a polygon if it poked his eye out?
Its a bit odd that you are accusing me of this when it seems you were the one who was making claims.
Well so far you've came up with some numbers that you think sound reasonable, and refused to believe any numbers I've given you. What I'd like to see is you show me some numbers from a few sources to show the opposite. Show me something anyway, so far you seem to be laying all the burden of proof on me.
Luminescent
16-Dec-2002, 21:25
I guess everyone had beef with the Flipper's T&L; many responses in such a short time. We should listen to ERP on the fact that the Flipper implements a simpler model for vertex lighting, and thus does less math per lighting computation. The Flipper seems to be a little faster at transformation though (I believe it could transform around 30-some million, while the Gefroce 2 achieves 25 million at 200MHz). This generation, with the exception of various PS2 titles, the consoles will used per-pixel lighting of some sort with little use for the L part of T&L. Whether it be procedural per-pixel lighting (lighting math is done in the combiners) or per-pixel lightmaps, vertex lighting will barely be used. Maybe PS2 can afford it because of its ability to pump out high rates of polygons (like the other consoles), which would allow for more vertex normals, and less interpolation artifacts. With more polygons, the benefits of complete per-pixel lighting are a little less, unless there are regular bump-maps. Just my 2 cents.
DeathKnight
16-Dec-2002, 21:29
while the Gefroce 2 achieves 25 million at 200MHz
Use the Ti or Ultra and the poly figures are nearly identical (31 for GF2 and 32.5 for Flipper). The original Radeon also pushes ~30 or so million.
I hadn't thought of the TI or Ultra, I was just thinking of the normal Geforce 2.
Can you show me an interview were they lied? All I've heard from people on this is complaints that Julian of Factor 5 compared an early XBox to a finished GameCube and said GameCube was faster. Even then Julian admited that this was an early XBox vs a finished GameCube, and was that even an official comment from Factor 5? Was it not just Julian talking in a forum? I'd really love to have links to all these 'lies' from Factor 5 just to know exactly what went on once and for all.
Can you give me a link that substantiate your claims? Again there is doubt in their honesty. I would be enterested in seeing if you can find some information that backs these claims up. They when can discuss whether they lied or not.
I can't remember that, but if it happened then fine, you can question what Sega say in the future about their games not Factor 5. As for how I would know that's BS, well because 60,000,000 is quite obviously impossible on PS2, that's just a technical fact.
If a developer makes certain claims they should back them up. The same should hold true to RL. How can we say these figures are accurate? How do you know these figures they are proposing aren't as ridiculous as factor5's assertions?
Many times in the past it has been said on this forum that Rogue Leader pushes 12,000,000+ polys per second, nobody else argued with that, not even any of the devs on this forum.
OK. SO because no one caught it and denied it must be true? I have heard people say those high polygon models were for cut scenes while lower polygon count models were used in game. I have heard that ingame models were around 2,000 - 4,000 polygons.
What's rediculous about it?
For one they sounds extremely high. Two there is no evidence to back up these claims. The same could be said about VF4. They can claim what they want but without actual figures they're assertions are meaningless.
You kept asking, did I count up the 12-15,000,000 per second number. I was saying that at least my numbers come from a source and if anyone's counting polys its you, your numbers come from just looking at the game and guessing.
come from a source? So if i turned to sega's page for the vf4 polycount i'd be justified in believing the game was rendering 60,000,000 polygons/second because i have a source that says so?
As for Rogue Leader having average polygon counts.. its poly counts and lighting quality is quite obviously way above average.
Did you see what ERP said? The lighting methods the gc uses are "grossly simplified." Above average Teasy? ERP doesn't think so.
If you don't want a fight then we don't have to have one. I don't want to have a slanging match, but I will retaliate if someone tries to insult me.
no one is insulting you teasy.
That back up your claims thing is a nasty one,
Why because i am asking for evidence? Why must you interprit this as i am calling you a liar?
it can come back and bite you. For instance back up your claims that Factor 5 have lied about Rogue Leaders polycounts, because nobody else seems to think so.
A lot of people here seem to think so Teasy. I am not the first person here that disagrees with the RL figures. Check the forum history. Second it doesn't make a difference what people here in this thread believe. If Factor5 misrepresented these figures then they're claims are null and void regardless of what people believe.
Or you could back up your opinion on Flipper not having more powerful T&L then Geforce 2.
Did you read what ERP said?
What i said exactly was that i wasn't so sure about Tangrineth's assertions. I didn't say i believed the Geforce 2 was more powerful then the Flipper. I just wan't sure about her exagerated claims.
Afterall we are talking about a much newer chip here (Flipper) when compared to Geforce 2, and with games like Rogue Leader and Metroid Prime on GameCube I would imagine its sensible to assume that Flipper is faster until its proven otherwise.
Newer doesn't beget better teasy in all cases. This chip is newer then the Geforce 3 - do you think it is more powerful then the geforce 3?
Where did you see this? Could you post it, because I didn't see that.
Sure i can i have the benchmarks right here (PDF file). If you want i can email them to you or upload them to you via messenger service or ftp.
Did Sega ever claim in an interview that they were actually pushing that number?
Would this make their webpage figure more accurate?
Did you go into specifics on how many polys were onscreen per frame at what framerate ect? Or was this a silly off hand, and probably confused, claim from a PR guy who wouldn't know a polygon if it poked his eye out?
would this make their webpage figure more accurate?
Well so far you've came up with some numbers that you think sound reasonable, and refused to believe any numbers I've given you no matter what source they're from.
No Teasy that isn't the case. I stated numbers i have been lead to believe are the norm. I have a hard time believing RL's polycount is dramatically higher then the norm without evidence for this. I don't believe developers should be trust on face value. A polycount figure is meaningless if they don't tell you how they got it, why, and what effects were being used or what conditions the tests were done under.
What I'd like to see is you show me some numbers from a few sources to show the opposite.
Hmmm i have had similiar arguments with christians. Teasy prove to me that the God of the universe isn't a 26th dimensional cream puff. Prove to me that VF4 isn't rendering 60,000,000 polygons/second. Do you see the similarity? You have made an assumption those Factor 5 figures are accurate without looking deeper into them. Teasy lets be reasonable, ok? You made a claim, you should back it up. Numbers from a webpage don't constitute evidence. They are meaningless figures if they have no value applied to them or conditions.
Rather then asking me over and over to prove it, then saying "I don't believe that source prove it another way". Your trying to make me jump through hoops here man :)
Teasy i am asking you to provide evidence for your claims.
Luminescent
16-Dec-2002, 21:58
Rogue leader mainly uses vertex lighting with some per-pixel lighting here and there. I guess the fact that its models are high poly mask the fact that they are, at least partially, vertex-lit.
Not all being drawn at once. The highest polycounts on screen for RL reach around 250,000. Don't know if it still keeps a steady 60fps in those peak situations though.
Legion, there are more than likely a ton of games above the "average" polycount. Azurik on the Xbox *peaks* at 500,000 polys a frame (at 30fps). It's not uncommon to have higher polycounts in this generation of consoles.
How do you know this?
Not all being drawn at once. The highest polycounts on screen for RL reach around 250,000. Don't know if it still keeps a steady 60fps in those peak situations though.
Legion, there are more than likely a ton of games above the "average" polycount. Azurik on the Xbox *peaks* at 500,000 polys a frame (at 30fps). It's not uncommon to have higher polycounts in this generation of consoles.
How do you know this?
Both are coming from dev mouths, something you are not likely to trust. I do not think Deathknight will bother to give you a link because
- anything older than one month is almost impossible to retrieve on the net
- you would not trust it anyway
FYI, many UPCOMING games on the xbox are clocking higher than 300k, for example Racing Apex is at 500k by frame (I do not know the framerate), so 250k for a GC game is completely realistic (and many recent ps2 games are at 150k+ at 60fps, WRC2 for example).
Fafalada
16-Dec-2002, 22:19
the consoles will used per-pixel lighting of some sort with little use for the L part of T&L.
That's simply not true.
Per pixel math still uses per vertex setup, which is just as much or even 'more' intensive then doing traditional vertex lighting.
In case of Flipper specifically, the Lighting part of the transform unit is used to perform the setup math for per pixel effects such as bumpmapping.
There are situations where the fixed math available on Flipper will not suffice to perform the operations needed, in which case you would need to fall back to CPU for help.
Did you see what ERP said? The lighting methods the gc uses are "grossly simplified." Above average Teasy? ERP doesn't think so.
Oh dear... -__-
well what are your thoughts faf?
Fafalada
16-Dec-2002, 22:32
My "thoughts" happen to be in agreement with what ERP said - but you're basically putting words into his mouth - whether by misunderstanding or not.
I'd rather not say anything furtther on the matter to avoid even more confusion.
I was just refering the lighting. I just find it hard to compare flipper T&L to Geforce 2 t&L.
Can you give me a link that substantiate your claims? Again there is doubt in their honesty. I would be enterested in seeing if you can find some information that backs these claims up. They when can discuss whether they lied or not.
Well this is incredible. From the beginning you've done nothing but ask me to to "prove this" or "prove that". Now you say Factor 5 lied and when I ask you to show me where you even ask me to prove they haven't lied. This is simple, if you claim something you back it up, or at least try before asking someone else to prove you wrong.
If a developer makes certain claims they should back them up. The same should hold true to RL.
How exactly would you have them back them up then? Apparently you don't believe that so they can't back up what they said.
How can we say these figures are accurate?
How can we say they aren't? Lets see, we have a quote from Factor 5 to say that Rogue Leader pushes 12-15 million pps at times, and against we have.. err.. nothing. All we have against is your opinion that they can't be trusted, and when I ask you to show why they can't be trusted you can't show why.
OK. SO because no one caught it and denied it must be true?
My point is we as you said, we have developers here, and many other people, these numbers must have been mentioned over 10 times in this forum and no'one has ever said they were false. Even now nobody says they are false but you, and you have nothing to show that they are false.. just guess work.
For one they sounds extremely high. Two there is no evidence to back up these claims. The same could be said about VF4. They can claim what they want but without actual figures they're assertions are meaningless.
Figures?.. what figures are you looking for. Factor 5 have given figures.
come from a source? So if i turned to sega's page for the vf4 polycount i'd be justified in believing the game was rendering 60,000,000 polygons/second because i have a source that says so?
Are you reading my comment at all before responding? I was saying that at least I have some sources, while you have nothing at all but guess work and stuborness. Show me this comment on Sega's site, I want to see who said it, and how it was said. My info comes from a interview from a highly skilled group of programmers, not just Factor 5 but also Lucas Arts themselves. Does that 60,000,000 info come from a real programmer or a PR guy?
Also no of course you couldn't believe it pushes 60,000,000 polys per second, because as everyone knows its totally impossible to do that with the PS2. Its very obviously possible to push 12,000,000+ polys on GameCube.
Did you see what ERP said? The lighting methods the gc uses are "grossly simplified." Above average Teasy? ERP doesn't think so.
Yeah I read what ERP said.. but you didn't obviously. ERP was not talking about software. Your taking something ERP said and changing it for your own use. If you really need to mis-read ERP's comments to get an opinion on wether or not Rogue Leader's poly counts and lighting is above average then you either haven't played Rogue Leader or your just in denial. I mean really, nobody here is going to agree with you that Rogue Leader's polys counts and lighting is not above an average game. Not even any PS2 or XBox fan-boy, and that's saying something!
Why because i am asking for evidence? Why must you interprit this as i am calling you a liar?
The question is why did you interpret my comment like that? Because I did not say you were calling me a liar. I was explaining that when you constently ask people to back everything up its going to come back and bite you when someone then says to you "why don't you back the opposite up instead?". You base too much of your argument on "I don't believe them" and "back that up then". Apparently you don't see any reason for you to back anything up yourself.
A lot of people here seem to think so Teasy. I am not the first person here that disagrees with the RL figures. Check the forum history. Second it doesn't make a difference what people here in this thread believe. If Factor5 misrepresented these figures then they're claims are null and void regardless of what people believe.
No, AFAICS people do not disagree with these polycount numbers from Factor 5. I'd ask you to post some threads with people disagreeing but I fear you'd just ask me to prove that these threads don't excist, since you don't seem to think you carry the burden of proving anything, everyone else must prove you wrong no matter the situation.
Did you read what ERP said?
Yes, once agiain I did, you didn't. ERP said he was not talking about performance comparisons. So where exactly did ERP say that Flipper is not faster at T&L then Geforce 2?
didn't say i believed the Geforce 2 was more powerful then the Flipper.
I didn't say you did, I said you claimed Flipper wasn't faster.
Also, yeah you didn't say that to Tangrineth, but apparently you think so because you've quite obviously argued with me saying Flipper is faster.
Newer doesn't beget better teasy in all cases. This chip is newer then the Geforce 3 - do you think it is more powerful then the geforce 3?
No newer doesn't always mean better, but usually it does. What I was saying is you have no reason to think it isn't faster. Flipper is ALLOT newer then Geforce 2, we've seen games with much better T&L work on Flipper, Flipper's raw poly numbers are higher then that of Geforce 2.. and to counter you have.. nothing AFAICS, other then "I don't believe that source" and "back that up".
Sure i can i have the benchmarks right here (PDF file). If you want i can email them to you or upload them to you via messenger service or ftp.
I can't get them through email. Could you just post the relivant part.
Would this make their webpage figure more accurate?
No, that wasn't my point. I was trying to distinguis between something said by programmers, technical people who know what they're talking about, in an interview and what a PR guy puts on the website.
No Teasy that isn't the case. I stated numbers i have been lead to believe are the norm. I have a hard time believing RL's polycount is dramatically higher then the norm without evidence for this.
Heh, could you give me a source saying that's the norm ;)
Seriously, if you actually played Rogue Leader you'd change your mind on it being around average polygon counts. Although I don't you'd admit it here.
Hmmm i have had similiar arguments with christians. Teasy prove to me that the God of the universe isn't a 26th dimensional cream puff. Prove to me that VF4 isn't rendering 60,000,000 polygons/second. Do you see the similarity? You have made an assumption those Factor 5 figures are accurate without looking deeper into them. Teasy lets be reasonable, ok? You made a claim, you should back it up. Numbers from a webpage don't constitute evidence. They are meaningless figures if they have no value applied to them or conditions.
You also made a claim, that Factor 5's numbers are false, that they lied. What you don't seem to realise is I made a claim and backed it up, then you made a counter claim and its in your court now, back it up. Just because you don't believe my source that does not automatically disqualify it, not until you back up your claim that my source is lying.
Just to lay it out for you. I said Rogue Leader is pushing upwards of 12,000,000 polys per second. You then asked me to back that up, I did so by telling you that the games developers said exacly that. You then claimed that the developers were not to be trusted because they're liars, so now its your turn to back something up.
megadrive0088
16-Dec-2002, 23:26
IMO, GF2 GTS isn't even comparable to Flipper since Flipper has 3.12 MB of on-chip 1T-SRAM while GF2 has a tiny cache in comparasion.
legion. Just ignore what teasy is arguing about and go with what Faf and ERP agree on. They are the onyl ones here that really knows what they are talking about.
Well this is incredible. From the beginning you've done nothing but ask me to to "prove this" or "prove that". Now you say Factor 5 lied and when I ask you to show me where you even ask me to prove they haven't lied. This is simple, if you claim something you back it up, or at least try before asking someone else to prove you wrong.
You have an uncanny nack for drama Teasy. By your own reasoning why should i prove anything? You aren't bothering to back up anything you are saying so why should i be held to a higher standard? If you want to make random unverifiable statements why can't anyone else? This is just mere digression on your behalf as it doesn't justitify your claims.
How exactly would you have them back them up then? Apparently you don't believe that so they can't back up what they said.
If you CAN'T back them up why do you continue to assume they are right? Perhaps you can ask some one here with knowledge.
How can we say they aren't?
How can we say God isn't a 26th dimensional cream puff? How do you know that VF4 isn't rendering 60,000,000 polygons/second? If you make a claim teasy you ought to have evidence to back it up. You ought to have evidence before believing in it. If you can't verify it shouldn't assume it is true.
All we have against is your opinion that they can't be trusted, and when I ask you to show why they can't be trusted you can't show why.
Apparently you don't see the reasoning. I did show you why their word shouldn't be taken at face value without evidence. The same reason why we shouldn't believe in God because he MAY exist.
My point is we as you said, we have developers here, and many other people, these numbers must have been mentioned over 10 times in this forum and no'one has ever said they were false. Even now nobody says they are false but you, and you have nothing to show that they are false.. just guess work.
No one has ever said they were false? Not a single person? Are you sure of this Teasy? I never said they were without a doubt false. Did i say they were false teasy? I remember stating others have opposed those figures.
Figures?.. what figures are you looking for. Factor 5 have given figures.
Anything that can be used to varify the polycount.
Are you reading my comment at all before responding? I was saying that at least I have some sources, while you have nothing at all but guess work and stuborness.
And i replied to you with the same asinine reasoning you are using. I can find a source that says Jesus was a alien if i wanted Teasy. I can find a source that says VF4 is rendering 63,000,000 polygons/second. Does that make their claims or your claims true? No. None of them can be substantiated.
Show me this comment on Sega's site, I want to see who said it, and how it was said. My info comes from a interview from a highly skilled group of programmers, not just Factor 5 but also Lucas Arts themselves. Does that 60,000,000 info come from a real programmer or a PR guy?
How do you know it didn't come from a programmer to a PR guy? It doesn't make a difference Teasy. If you can't substantiate their claims their figures are meaningless.
here are some links:
http://www.outpost.com/product/3259641/
http://learn.senecac.on.ca/~cnikolak/bcs/project/
Sega's original page is down concerning the information. If you search google for combinations of vf4 and 63 million you will find numerous mentions throughout web boards and forums
here is the original link:
http://www.sega.com/games/ps2/post_ps21look.jhtml?article=1look_ vf4
Also no of course you couldn't believe it pushes 60,000,000 polys per second, because as everyone knows its totally impossible to do that with the PS2. Its very obviously possible to push 12,000,000+ polys on GameCube.
Teasy you are missing the point. The fact is i could toss Factor 5's figures into the realm of hype just as easily as i could Sega's claims PR or not. I don't believe that figure in anyway represents the true polycount of VF4. My reasoning for that is knowing a bit about the hardware and its theoretical capabilities of the PS2. These claims exceed the capabilities of what the PS2 can handle even if it were 100% efficient. My problem with the Factor 5's figure is that neither of us can substantiate it. Should we really take these figures on face value? If so why should anyone who isn't familiar with the PS2 in anyway assume the VF4 claims are bs? By your reasoning we should take them on face value.
Did you see what ERP said? The lighting methods the gc uses are "grossly simplified." Above average Teasy? ERP doesn't think so.
read what ERP said.. but you didn't obviously. ERP was not talking about software. Your taking something ERP said and changing it for your own use.
No actually i wasn't. He specifically said there isn't an easy way to compare the two as "apples to apples".
really need to mis-read ERP's comments to get an opinion on wether or not Rogue Leader's poly counts and lighting is above average then you either haven't played Rogue Leader or your just in denial.
He specifically said there was no way to compare the lighting in its complexity as "apples to apples". How can you then say RL features more complex lighting then ANY Geforce 2 game? Because it APPEARS that way to you? I did i no way associate polygon counts and ERPs statements.
I mean really, nobody here is going to agree with you that Rogue Leader's polys counts and lighting is not above an average game. Not even any PS2 or XBox fan-boy, and that's saying something!
Teasy what they think is irrelevant. What ERP stated was you can't easily compare the two (Geforce 2 to Flipper) as far as their T&L complexity is concerned. You were the one who stated RL is superior to ALL games built around or run on Geforce 2 hardware. How can you make that statement? Where is your evidence for this? If i recall correctly i said around the average game. I was specifcally questioning you on how you know RL to feature vastly superior polycounts and lighting to any or all games that run on Geforce 2 hardware.
Honestly i don't give a damn if they agree with me or not. I am asking you HOW you know what the RL polycounts are and why that is so. I have a hard time believing the game is as complex as you say on face value Teasy. There is nothing wrong with being scepticle of figures that lack substance.
The question is why did you interpret my comment like that?
You assertion about a personal attack seem to suggest this.
Because I did not say you were calling me a liar. I was explaining that when you constently ask people to back everything up its going to come back and bite you when someone then says to you "why don't you back the opposite up instead?".
Ok prove there is no God. Prove there is a God. its all the same reasoning. If you make a claim why shouldn't you provide evidence to back yourself up?
You base too much of your argument on "I don't believe them" and "back that up then". Apparently you don't see any reason for you to back anything up yourself.
andy ou base your entire argument on "i believe". If there is no evidence to substantiate the claim you shouldn't assume it is correct.
No, AFAICS people do not disagree with these polycount numbers from Factor 5. I'd ask you to post some threads with people disagreeing but I fear you'd just ask me to prove that these threads don't excist, since you don't seem to think you carry the burden of proving anything, everyone else must prove you wrong no matter the situation.
Really? So you are telling me if hunt through these forums i won't find one person who disagrees with the framerate, or polycount, or lighting? Are you so sure of this Teasy? I didn't ask you to pove anything. I asked you how do you know SOME one out there doesn't agree with the polycounts with good reason? Does it matter if they agree or not? You were the one who through the random statement out there Teasy not I. To say everyone is like minded on this issue is a bit absurd.
Yes, once agiain I did, you didn't. ERP said he was not talking about performance comparisons. So where exactly did ERP say that Flipper is not faster at T&L then Geforce 2?
:lol: he didn't say the geforce 2 was faster. He also didn't say the Flipper was faster. Thats my point. Its difficult to compare the two. That was what he was suggesting. That on top of the fact the difference in the ways the T&L is implemented in games is different. Why on earth do you assume the Flipper's T&L is faster? How do you know? How can you compare what the Geforce 2 is doing in X game to what the flipper is doing is Y game?
I didn't say you did, I said you claimed Flipper wasn't faster.
Thats essentially the same thing. Either they are equal or one is faster then the other by that reasoning. In said case you make it sound as if i am asserting the Geforce 2 is faster which i am not. I originally disagreed with tangrineth who said the flipper's T&L was a HELL of a lot faster.
Also, yeah you didn't say that to Tangrineth, but apparently you think so because you've quite obviously argued with me saying Flipper is faster.
I quoted here in my response. Check it out for yourself. If i were responding to lumi would have quoted him.
What we have argued about is your reply to me. I want to know how you know what you are asserting. You claimed the flipper was faster then the geforce 2. How do you know this? How?
No newer doesn't always mean better, but usually it does.
Not in this case.
What I was saying is you have no reason to think it isn't faster.
Whoa goofy logic. There is no reason to assume it isn't faster likewise there isn't a reason to assume it is faster. Would we assume the Flipper is faster then the Geforce 3 because it happens to be newer?
Flipper is ALLOT newer then Geforce 2, we've seen games with much better T&L work on Flipper, Flipper's raw poly numbers are higher then that of Geforce 2..
What? The Geforce 2's theoretical pixel fillrate was 800 megapixels/second. The FLipper's is 648 megapixels/second. How do you figure the raw poly counts would be larger for flipper then the Geforce 2 if they rendered polygons containing the same number of pixels?
and to counter you have.. nothing AFAICS, other then "I don't believe that source" and "back that up".
http://nvidia.com/view.asp?PAGE=geforce2pro
GeForce2 GTS Pixels Pixels per Second: 800 Million
I can't get them through email. Could you just post the relivant part.
THen why can't i send it through a messenger? that would be easier.
Heh, could you give me a source saying that's the norm ;)
you could give sourcers that substantiate your claims ;).
Seriously, if you actually played Rogue Leader you'd change your mind on it being around average polygon counts. Although I don't you'd admit it here.
I have played RL. I can tell you that the models have a good number of polygons in them. Can i tell you there are 12,000 or 3,000 or 5,000 in each model by looking at them? No i can't. Can you?
You also made a claim, that Factor 5's numbers are false, that they lied.
No i stated they have misrepresented facts in the past and that their figures are meaningless by themselves.
What you don't seem to realise is I made a claim and backed it up,
No you didn't. You made a claim that no one disagreed with it.
then you made a counter claim and its in your court now, back it up. Just because you don't believe my source that does not automatically disqualify it, not until you back up your claim that my source is lying.
No, i didn't. I stated there is no reason to believe the source without evidence for it.
Just to lay it out for you. I said Rogue Leader is pushing upwards of 12,000,000 polys per second. You then asked me to back that up, I did so by telling you that the games developers said exacly that. You then claimed that the developers were not to be trusted because they're liars, so now its your turn to back something up.
I did back myself up by saying they are no trust worthy just as you backed yourself up with some one else's claim.
legion. Just ignore what teasy is arguing about and go with what Faf and ERP agree on. They are the onyl ones here that really knows what they are talking about.
I agree. What Teasy is arguing is nothing more than his personal opinion. I am not going to see these real figures. I believe ERP and FAF know more then they are sharing. However i doubt they will relate what they know to me.
I have no interest in arguing this point further with Teasy.
DeathKnight
17-Dec-2002, 03:17
For Azurik I have a link:
http://forums.teamxbox.com/showthread.php?s=&threadid=99899
I don't doubt the figure either since I've played the entire game. There are some pretty jaw dropping moments in the game as far as complexity and draw distance goes.
Plasmatics
17-Dec-2002, 09:19
People here take Legion seriously?
I haven't seen this level of stupidity since ArbitraryName. People say things like that a lot on message boards, and I usually think they're exaggerating. . . but this guy can't be serious.
Joke character.
Right?
PC-Engine
17-Dec-2002, 09:45
F5 did say RL pushes up to around 12 million polys/s in an IGN interview. I'm too lazy to go dig up the info so take it anyway you wish.
Magnum PI
17-Dec-2002, 09:51
ERP
Nothing I've seen Geforce 2 do, even in tech demo's, comes close to what Flipper does in Rogue Leader. I really don't know what you mean about its lighting being simple compared to Geforce 2. The lighting in Rogue Leader is fantastic, that certainly doesn't look simple to me.
but how do you know if the gekko didn't take part in the T&L calculation of SW:RL ?
anyway this has nothing scientific. this is only your subjective perception.
this is only your opinion, not a fact..
Everything I've seen, from benchmarks
i'm very curious to know which benchmarks did you run on the GC and on a PC with a geforce ?
to just looking at the games
very scientific...
tells me that Geforce 2 cannot stand up to Flipper in T&L performance. So you'll forgive me if I don't just take your word on this.
don't you want to understand ? what this guy is telling is that the lighting made in hardware by the flipper is more basic compared to the lighting of the geforce 2.
so you can't just simply compare the T&L performance of these two cards, because that the lighting they calculate isn't the same.
it would be like saying that a vespa is better than a tank because it is faster.. you can't compare because these two vehicle doesn't have the same functionnality.
BTW i read the ERP message in this thread and it didn't seem to me he was implying the geforce 2 T&L was superior to flipper.. just that the lighting GF2 did was different.
BTW, what a few people here are forgetting is that we shouldn't just be taking raw T&L power into account here, but also the efficiency. Lets see what Geforce 2 is pushing with 4 textures (having to do two passes) vs Flipper, or even 6-8 textures (after Geforce 2 has taken 3-4 passes and Flipper only one).
what does it have to do with this topic ?
"Is Flipper's geometry engine more powerful than Elan?"
i guess we shouldn't create silly topic like that. topic containing the words "more powerful" can only degenerate..
Luminescent
17-Dec-2002, 12:22
By more powerful I meant more capable.
I remember that comment from F5 about Rogue leader. They said that throughout the game RL pushes 12-15 mil. pps with 5-8 layers of effects. The problem is that after that "GC can do everything the Xbox can" nonsense, they kinda lost some of their credibility as far as I'm concerned. They even said that those figures were guesses and that they "didn't have time to count all the polygons, or else we wouldn't have time to complete the game in time for launch". So who knows, those numbers might have been inflated. There's no way to know. Like it was mentionned in another thread, they had a part in creating the GC hardware so there's a reason for them to speak highly of it.
Now, this has nothing to do with their accomplishements when it comes to RL and their new game. RL is still one of the best looking game on the GC, some would argue still the best looking, and I can't wait to see their new stuff. I have a feeling it will outdo anything we've seen so far on the platform. You just have to learn to seperate that from BS PR statements.
PC-Engine
17-Dec-2002, 19:39
F5 also said they were doing only basic stuff in RL and mentioned that their next game will used advanced stuff done on the TEV unit.
Magnum PI
17-Dec-2002, 20:06
perhaps the next factor5 game is the megaton !! :o
I actually find it intriguing as to why factor5 has such a reputation from some members on these forums. What bullshit have they said? They only controversial claims they've made in the past were that flipper had a better pixel shader (TEV) than nv2a, and had neater vertex indexing methods.
And guess what, Julian was pretty damn right with his opinions, back then it was controversial but now its pretty common knowledge that TEV can hold up to nv2a's combiner for most situations, and that the vertex indexing on gamecube is pretty unique, well from what i've heard, it was a request for the openGL forum from developers but never implemented until flipper (not entirely sure).
The rest was mostly about sound, and while they've made a number of spins about their 5.1 sound, technicly, they were still holding up to their claim, though misleading on some points, but heck, even dolby played the game.
12-15M polys for RL is believable anyway, just need to play ison corridor or the death star battle and bam, if you think there's not enough objects on screen then you're crazzeeeee (well there's no such things as too much ;) )
The problem is that after that "GC can do everything the Xbox can" nonsense, they kinda lost some of their credibility as far as I'm concerned.
That was when they had xbox alpha or beta kits that didnt have nv2a in it.
Its useless to take very old quotes out of context and put them back in an argument about current knowledge.
The last comments factor5 did about gamecube's graphical capabilities were with its pixel shaders and vertex indexing, nothing else.
Fafalada
18-Dec-2002, 00:49
nv2a, and had neater vertex indexing methods
To be specific, Flipper allows indexing per vertex component, while NV2a indexes per vertex.
This potentially allows for smaller vertices. We use this in our app as well, savings aren't huge, but it's just one of the small things that's nice to have.
If I am not mistaken, NV2a also requires memory alignment on per vertex basis, which can further increase their size (maybe ERP can clarify more), but with 64mb available, I know I wouldn't be too upset over it. :P
On the flipside, Flipper doesn't have post transform cache, which NV2a uses with indexing to speed up vertex processing. This is also something that comes rather handy, so I wouldn't necesserily say one feature is always better then the other.
It entirely depends whether you end up memory or speed limited which one you'd like to have more IMO - which in turn usually means you always want the one you don't have ;).
Plasmatics
18-Dec-2002, 10:58
Since no one else wanted to find them:
Factor 5 PGC Interview (http://www.planetgamecube.com/specials.cfm?action=profile&id=203)
IGN's Interviews (http://cube.ign.com/features/537.html) Includes a number of Factor 5 interviews as well as IBM talking about Gekko, ATI talking about Flipper and even talk about DPLII. Too many to individually link.
Mariner
18-Dec-2002, 11:56
Most of the reason for the argument between Teasy and Legion is that Teasy is arguing using the modern day mores of international law i.e. If F5 say they are getting 12-15mps and you disagree then the onus is on you to prove they are lying.
Legion, on the other hand, seems to be using the Napoleonic code i.e. If F5 say they are getting 12-15mps and you disagree then the onus is on them to prove they are telling the truth.
Both are valid viewpoints although virtually the only time we encounter the Napoleonic code in these days is in dodgy episodes of Star Trek!
And Quebec's 'Code Civil' :lol:
j/k
vBulletin® v3.8.6, Copyright ©2000-2013, Jelsoft Enterprises Ltd.