SCEI & Toshiba unveil 65nm process with eDRAM

[Fafalada]

Obviously not every game does sky the same way, but enough people like to use that sky generating util making it sort of 'standard'.
That said, 'standard' skybox will run you 5(or 6 if you want bottom too)512x512 maps. This should look highres enough for resolutions below 1024. You could break it up into 20 256x256 maps, but that's a lot of fuss for just sky(imo).
256 for entire side of the cube is rather on the low res side, so you'd have to resort to some kind of tiling or overlaying to cover it up. (but if you're not carefull with that you might end up with something like the "3 suns on one screen" effect, seen in F355 DC )

True enough, if you devoted some time to just coding up something for sky rendering, you could probably get away with far less texture data use and still make it look nice, but it's always a question is whether that would be time well spent. Especially since end visual result is unknown.

As for banding, skymaps are typically gradient heavy, which is the worst case scenario for VQ compression schemes.

Until you start getting rather high, a lot of people can get fairly close numbers.

You can get relatively close on some individual objects, yes. But overall is completely random guessing, especially accounting for particle stuff, not to even mention "invisible" geometry that goes into offscreen buffers.

What's VT? I've never seen HOTD for the PC(running).

Virtua Tennis. HOTD you didn't miss much, it's a straight port with nothing added, except the said graphical glitches :\

 

[zidane1strife]

Because they were CPU limited. Throw a title like Giants at the TNT2 Ultra and it gets blown away by the GF(running the same res in both IQ and performance

What's the deal with the cpu limitation anyways... So I'm supposed to believe the Gf1 will continue to go up in perf. with newer and newer cpus... maybe that's because the games are doing part of the T&L on the cpu or something....
The Gf1 clearly has it's limits... and according to the 3dfx guy i quoted it's basically 2 tnt2s put together with some features...
I just don't get it even Nvidia in their Gf3 demos shows few or even one object/face at a time... clearly if the Gf line could've EDIT In the pastEDIT
rendered tons upon tons of high geometry like the consoles do now they'd have done it....Anyways from what the Cage guy in the gspot quote say'd it seems he couldn't get more than 1Mpolys in game out of the Gf1...

"When GeForce256 was released in October 1999 it came with SDR memory at 166 MHz clock. The release of the famous 'GeForceDDR' cards, which were nothing else than the same chip, but with faster memory, showed how much a fast 3D chip can be stalled by slow memory. " hmmm... it seems the DDR ones came later...

Anyways IMHO a few more features here and there aren't enough to outshine the ps2's performance lead in most areas...

found other info outhere...

"In the case of NVIDIA's GeForce, the "GPU" is doing these mathematics. The GPU has special pipelines, which are optimized to execute the T&L mathematics. Basically, the GPU has a piece of hardware on-board specialized and optimized to do this work. Naturally, this hardware has limitations- since it is hardware and is running at a certain clockspeed, it has a certain maximum throughput....
...and later on...

High Polygon Count (1 Light)
4444 KTriangles/s Hardware T&L
6752 KTriangles/s Intel Pentium III 742Mhz
3969 KTriangles/s AMD Athlon 700MHz (*)
High Polygon Count (4 Lights)
3164 KTriangles/s Hardware T&L
5453 KTriangles/s Intel Pentium III 742Mhz
3317 KTriangles/s AMD Athlon 700MHz (*)
High Polygon Count (8 Lights)
1711 KTriangles/s Hardware T&L
4300 KTriangles/s Intel Pentium III 742Mhz
2653 KTriangles/s AMD Athlon 700MHz (*)

....Yes NVIDIA says that they don't agree, and there is a good reason for not agreeing. Want to know what the reason is? Well as the numbers at the start indicate, both Pentium III's and AMD Athlon's software T&L is faster than NVIDIAs special hardware T&L. If software is beating your hardware you are in big trouble, and you'll do anything to convince the public that something is not right. It's called damage control people!....
...The reason why NVIDIA wants us to use the Microsoft implementation is because this is the least optimized one available. Top games, like Quake3: Arena, use their own custom optimized engines (light is done in software), and Messiah uses software T&L for most major parts of the geometry (characters). There are many more that also fall under this category.
....GeForce is slow because it hits its hardware limit. Hardware always has a limit, a limit introduced by the implementation (the actual hardware pipeline) and its clock frequency. There is only so much that hardware can do.
...The thing that amazes me most is that, to the public, NVIDIA says that developers should use as many lights as possible, while their optimization FAQ says this:
Q. How many lights should I use?
A. In general, as few as possible…

Anyone who doubts that marketing crap exists today, here is some solid proof!
...NVIDIA's defensive arguments are pushing it a bit - call it fanatical. We found out that by using more lights, especially non-directional lights, is very expensive (performance drops more than a whopping 50%!). Also, using a lot of point lights and spotlights is probably a big no-no on the GeForce, unless you would like to see a slideshow using Hardware T&L."

hmm.m.m.

 

[Lazy8s]

chap:

IGN do not see any drastically improved special effects

I'm just stating what the developers of the game said on their forum. If you go over to the Black Isles message board, you could probably have them address any inquiries directly.

Also, I never once said the effects were supposed to be "drastically" improved.

and BGDA is rendered at high res internally on the PS2.

Um, ok. So? It downsamples from that for AA purposes which is nice, but it sounds like you're using that to put it on equal footing with the Xbox port which also has the addition of progressive scan output. Rendering internally at a high res on the PS2 doesn't stop it from being field rendered, or give it the benefits of non-interlaced output, or help it in any way with regards to that. Like anything else, output is the final limitation.

Since both are 1999 tech

Not sure about the GeForce (I think that was holiday 1999), but the PS2 didn't reach the market until March of 2000. All technology is of course finalized at least a few months before it ships, but basing it off the standard of when its finalized means you'd have to make the same readjustment for all hardware.

 

[zidane1strife]

Not sure about the GeForce (I think that was holiday 1999), but the PS2 didn't reach the market until March of 2000. All technology is of course finalized at least a few months before it ships, but basing it off the standard of when its finalized means you'd have to make the same readjustment for all hardware.

Well that's just a few months difference(not unlike what nvidia has done in the past a few very rare times...releasing a few months later than competitors...)... and i think it's still the same fiscal yr... not sure.

 

[marconelly!]

Um, ok. So? It downsamples from that for AA purposes which is nice, but it sounds like you're using that to put it on equal footing with the Xbox port which also has the addition of progressive scan output

PS2 version doesn't have progressive support, but rendering at higher resolution internally has resulted in excelent AA. You can see on many of the framebuffer shots that Xbox version seemingly doesn't AA everything, like on this pic:

I can double check, but I'm pretty sure, in the same scene, PS2 version has none of the aliasing you can see here.

 

[Phil]

I can actually confirm that, as I have the game running right this second and I can't see any aliasing what so ever. I wish I could grab a screen...

BTW; nice info Zidane. I actually wrote you a PM but I don't think you saw that one yet, so I'll just ask again in here: what's that MGS2 doc you guys are talking about? Any link to where I can get a hold of it, anyone?

 

[zidane1strife]

MGS2 doc is a special MGs2 documentary that was released a while ago...

It is available in most gamestores i guess... It's valued at 20$, i think...
At least for gamestop u can check local availability of their products through their site.


It contains all Mgs2 cutscenes with no sound... but u can rotate the camera..

There are also stage, and char. models for u'r viewing...

5 vr missions...

Some notes on dev.(Not many though...)

Some vid.s, promo material photos...

and a few other stuff..

 

[marconelly!]

Also the original artwork all the trailers ever released and many other things like that. You can order it from EBworld site

http://www.ebgames.com/ebx/categories/products/product.asp?pf_id=228917

 

[randycat99]

I can actually confirm that, as I have the game running right this second and I can't see any aliasing what so ever. I wish I could grab a screen...

How about the subtle jiggly breasts? I wonder if that "feature" made it through on the port?

 

[randycat99]

You can see on many of the framebuffer shots that Xbox version seemingly doesn't AA everything, like on this pic:

Is it AA'ing anything in that shot? It looks like the multisampling is boned and the anisotropic filtering (the textures?) is the only smoothing going on. Did I get the terminology correct?

 

[BenSkywalker]

Cybermac-

Just a bunch of polygons with alpha textures on them.

It appears to be using vertex shaders based on the movement of the hair(not nearly as clumpy as they usual alpha texture trick). If that's all they are doing, they are devoting a hell of a lot more polys to it then the other titles using the technique.

Fair enough... I want to know about an Xbox title that competes with SH3 in all aspects.

Splinter Cell.

Not more so than in DOA3. There's a shot where the girl picks herself up off the floor and the hair moves quite a bit. Not really saying it's super impressive but I fail to see how DOA3 does it better.

Rewatching it and looking for it, it appears to move in a couple of large clumps, very similar to Links hair in Celda. DoA3 looks like there is more then three thick strands of hair.

Faf-

You can get relatively close on some individual objects, yes. But overall is completely random guessing, especially accounting for particle stuff, not to even mention "invisible" geometry that goes into offscreen buffers.

On the PC side of things, people regularly nail down within 10% the amount of polys in use(some applications include counters, and developers will regularly state poly complexity if pressed). As far as texture sizes, on the PC we can look at the individual textures in PS or the like making it pretty easy.

Virtua Tennis. HOTD you didn't miss much, it's a straight port with nothing added, except the said graphical glitches :\

Didn't even know VirtuaTennis came out for the PC, not that I would even consider picking it up anyway(only tennis game that ever kept my interest for more then ten minutes was Mario tennis for the VB)

Zidane-

What's the deal with the cpu limitation anyways... So I'm supposed to believe the Gf1 will continue to go up in perf. with newer and newer cpus... maybe that's because the games are doing part of the T&L on the cpu or something....
The Gf1 clearly has it's limits... and according to the 3dfx guy i quoted it's basically 2 tnt2s put together with some features...

And you could say the GS is a pretty much a Voodoo1 with some features too

Anyways from what the Cage guy in the gspot quote say'd it seems he couldn't get more than 1Mpolys in game out of the Gf1...

Which game was it?

hmmm... it seems the DDR ones came later...

In terms of actual availability I think there was about five weeks between them.

High Polygon Count (1 Light)
4444 KTriangles/s Hardware T&L
6752 KTriangles/s Intel Pentium III 742Mhz
3969 KTriangles/s AMD Athlon 700MHz (*)

Looks like numbers from Kyle(under five million with one light is a give a way). Besides his numbers being lower then what anyone else could manage with their GF1(had a lengthy discussion about this when the article was first posted- people even tried to underclock the boards to hit his numbers), he also used a benchmark that was using two different types of lighting engines. He had another one you might be able to dig up. He used TestDrive6 which has a built in bench, a game with extreme levels of overdraw, and then fill limited the board and showed soft T&L being faster then hard. Those of us following his crusade against T&L ran the benches at the lowest resolution allowed and saw framerates four times faster and it was still CPU limited(while using hard T&L there was a noticeable bandwith hit with early drivers, one that was rectified with the Det2s IIRC although simply lowering the resolution was enough to disprove his assertions).

Well that's just a few months difference(not unlike what nvidia has done in the past a few very rare times...releasing a few months later than competitors...)... and i think it's still the same fiscal yr... not sure.

Don't know why, but I have been thinking this whole time that the PS2 launched in 1999. March of 2K? That was a month prior to the GF2, not that it changes my stance.

Marco-

I can double check, but I'm pretty sure, in the same scene, PS2 version has none of the aliasing you can see here.

I have numerous duplicate screenshots one with extreme aliasing and one with almost none for each. All you have to do is adjust the contrast on your capture device.

Randy-

Is it AA'ing anything in that shot? It looks like the multisampling is boned and the anisotropic filtering (the textures?) is the only smoothing going on. Did I get the terminology correct?

I don't recall any XBox games that use MSAA off the top of my head. Can't tell what texture filtering is being utilized, without more angular textures or knowing what type of LOD bias they are using it won't show up in a screen capture.

 

[zidane1strife]

Splinter Cell.

Characters there are kinda blocky... probably because of the use of the unreal engine...

(I know a bit repetitive but...)Anyways as for Gf1, Gf2 both lack the pixel shader effects that xbox titles have, and as we've seen geometry and vertex lighting the ps2 is not too behind the xbox a system that's basically Gf4.... so u'r only argument is in favor of texture rez, and aniso, tri, is not too good...

Either u'r implying that nvidia has done minor improvements in the vertex, fillrate area since the ps2 is not to behind an xbox in such areas and u claim gf1 can compete, or u'r implying the xbox is capable of several fold what's been shown...

 

[BenSkywalker]

or u'r implying the xbox is capable of several fold what's been shown...

Without a doubt. Have you seen Halo2 yet?

 

[cybamerc]

BenSkywalker:

> It appears to be using vertex shaders based on the movement of the
> hair

I think the word shader is being used just a tad too liberally these days (damned be Nvidia PR).

That's not to say it isn't done with VS but it might as well be done on a CPU. Either way, in the classical meaning the word shader hardly applies to some basic animation and physics routines (again, damn Nvidia).

> Splinter Cell.

Splinter Cell doesn't have full scene shadowing. I can't tell if SH3 does either but the use of dynamic shadowing is certainly far more extensive. Also, while Splinter Cell employs a mish-mash of various techniques SH3 seems to stick with volumes and vertex lighting (or per-pixel... hard to tell).

Also, the dynamic shadows in Splinter Cell are hard edged and unlike the volumes in SH3 the shadow map method comes with visible aliasing.

 

[zidane1strife]

Without a doubt. Have you seen Halo2 yet?

Well yeah, but that's basically due to use of pixel shaders effects... i meant a several fold increase in the vertex perf...

 

[BenSkywalker]

Cybermac-

That's not to say it isn't done with VS but it might as well be done on a CPU.

You can do all vertex and pixel shader effects on a CPU, vertex shader ops in particular are actually quite well suited to decent CPUs.

Either way, in the classical meaning the word shader hardly applies to some basic animation and physics routines

Classical meaning of vertex shaders? I have to say I've never seen vertex shaders discussed in any meaningful way prior to the GF3. Spent a few years working with 3D viz and everything was done per pixel when talking about shaders.

Splinter Cell doesn't have full scene shadowing.

It doesn't? Playing through the demo it appeared to.

Also, while Splinter Cell employs a mish-mash of various techniques SH3 seems to stick with volumes and vertex lighting (or per-pixel... hard to tell).

A mish mash is better then vertex for everything(if it is vertex).

Also, the dynamic shadows in Splinter Cell are hard edged and unlike the volumes in SH3 the shadow map method comes with visible aliasing.

Didn't notice any aliasing in particular while playing through the demo. Not saying it wasn't there, but in the half hour or so I played the game it didn't stick out. True it didn't have soft shadows, that was easily noticeable.

Zidane-

Well yeah, but that's basically due to use of pixel shaders effects... i meant a several fold increase in the vertex perf...

Why would I argue vertex performance? I have never argued against the PS2 in that aspect, certainly no reason to start now. Even if the XBox was capable of tripple the geometric complexity of the PS2 at the level we are already at it isn't going to make a huge difference. We are at the point now where hard edges are increasingly rare. Telling the difference between a 100poly sphere and a 500poly is easier then comparing a 500 with a 1,000,000. There is certainly still room for improvement, but it will take a generational gap to make it much of a factor.

 

[Guest]

A bit OT, but i hear the only PS2 game to feature bumpmapping is called Stretch Panic from Treasure.
Does it look good?
Does it kills PS2 peformance?

 

[cybamerc]

BenSkywalker:

> Classical meaning of vertex shaders?

Classical meaning of shader. Vertex Shader is merely a buzz word invented by Nvidia PR.

When ppl talk about shaders I think of something that describes the properties of an object... not geometry transformation and whatnot.

> It doesn't? Playing through the demo it appeared to.

I haven't played it but there are plenty of screens on the net proving that is not the case.

> A mish mash is better then vertex for everything(if it is vertex).

Well... you seem to like DOA3's lighting and that is done per-vertex save for a very few effects. I think a valid criticism of SH3 would be its simple background but you can't fault the lighting and shadowing in the game.

> Didn't notice any aliasing in particular while playing through the demo.

Like RL you won't notice it most of the time but there are times when the maps are stretched across the screen that it becomes very apparent.

Apparently SH3 uses soft edged stencil volumes (I still would like to know how that is done without using several jittered volumes) so aliasing never becomes a problem.

 

[Fafalada]

On the PC side of things, people regularly nail down within 10% the amount of polys in use(some applications include counters, and developers will regularly state poly complexity if pressed). As far as texture sizes, on the PC we can look at the individual textures in PS or the like making it pretty easy.

Eh... would those people be referring to PC users - same PC users perhaps of which 90% thinks Doom3 is high poly?
Anyway, from what I've seen of people's estimates on console games, I don't recall anyone coming within 50% of actual numbers(except maybe by pure luck), though granted usually they were mislead by developer PR releases. (this in reference to the few games where I was privy to actual info - I don't think myself capable of telling what pp/s a game is running off hand)

As for textures, there's no way to know when a single large map is used in multiple parts just by observation. This before we mention the obvious stuff with multipass for surface detail. Although granted, it's somewhat more quantifiable with textures.

Didn't even know VirtuaTennis came out for the PC

I'm something of an avid tennis player so video games in the genre are also of interest to me. VT isn't my favourite tennis game, but it's still one of the better ones.

 

[Guest]

Hey Faf! Any insider info about Fafracer2k? ?