SEGA - PowerVR - May 20th [ oh please be Series 5!]

[SuperCow]

Then again we haven't seen any of the techniques used yet in games that are included in the 3 DX9.0 demos (closest chance could be if rumours are true that STALKER uses a very similar deferred shading technique to the DS demo).

Deferred Shading
STALKER does use it, there is a lengthy article about the way they've used it in one of the nVidia GPU programming book.
Ensemble Studios is using it, or is planning to use it, they gave a GDC presentation about it last year: http://www.gdconf.com/archives/2004/pritchard_matt.ppt
According to Deano some PC and XBox game are already using this technique (ref. to his ShaderX3 article).
Rendering of height maps and sample them in vertex shader (cloth demo)
Wasn't there a game that used this technique for a water effect that was discussed here a few months ago? I remember this being fairly controversial because the technique was costly (and would obviously be only supported on nVidia PS3.0-capable cards), but didn't seem to make such an impact on the visual quality? I do see this technique being used much more as PS3.0-hardware becomes more prevalent.
Voxel rendering with PS3.0 ray trace operations
Yeah, this one might take some time before it's used in games due to the excessive rendering cost associated with the technique.

 

[amk]

B3D posted a safe assumption based on the demos on it's frontpage

There were also the SM3.0 articles written by PowerVR people in ShaderX.

 

[Ailuros]

I'm not saying that it isn't at least SM3.0 compliant, there just wasn't anything definite announced. A serious website like B3D would naturally just report the appearance of the demos with a minor question mark at the end.

There were also the SM3.0 articles written by PowerVR people in ShaderX.

The demos were initially coded for those articles anyway. If S5 ever surfaced in the PC space there would had been definitely more demos than just those.

 

[Ailuros]

Then again we haven't seen any of the techniques used yet in games that are included in the 3 DX9.0 demos (closest chance could be if rumours are true that STALKER uses a very similar deferred shading technique to the DS demo).

Deferred Shading
STALKER does use it, there is a lengthy article about the way they've used it in one of the nVidia GPU programming book.
Ensemble Studios is using it, or is planning to use it, they gave a GDC presentation about it last year: http://www.gdconf.com/archives/2004/pritchard_matt.ppt
According to Deano some PC and XBox game are already using this technique (ref. to his ShaderX3 article).
Rendering of height maps and sample them in vertex shader (cloth demo)
Wasn't there a game that used this technique for a water effect that was discussed here a few months ago? I remember this being fairly controversial because the technique was costly (and would obviously be only supported on nVidia PS3.0-capable cards), but didn't seem to make such an impact on the visual quality? I do see this technique being used much more as PS3.0-hardware becomes more prevalent.
Voxel rendering with PS3.0 ray trace operations
Yeah, this one might take some time before it's used in games due to the excessive rendering cost associated with the technique.

Apparently I did mention STALKER as you can see; as for the cloth demo that's a very simplistic illustration up there:

http://www.pvrdev.com/pub/PC/eg/h/Cloth.htm

I might have missed something but I don't recall any game out there to support geometry textures and vertex texturing at the same time.

 

[KimB]

I'm not saying that it isn't at least SM3.0 compliant, there just wasn't anything definite announced. A serious website like B3D would naturally just report the appearance of the demos with a minor question mark at the end.

Seems a bit more certain than that to me. With SM3 demos, they must have had SM3 hardware in the pipeline. Since it's been quite some time since those demos have appeared, this next part being SM3 (or something very similar: it's not meant for a Windows system) is only logical.

 

[Lazy8s]

The two companies are working with NEC again, I'm thinking.

 

[Ailuros]

I'm not saying that it isn't at least SM3.0 compliant, there just wasn't anything definite announced. A serious website like B3D would naturally just report the appearance of the demos with a minor question mark at the end.

Seems a bit more certain than that to me. With SM3 demos, they must have had SM3 hardware in the pipeline. Since it's been quite some time since those demos have appeared, this next part being SM3 (or something very similar: it's not meant for a Windows system) is only logical.

Of course do they have SM3.0 capable hardware in the pipeline; I can´t imagine SEGA´s Arcade board to be based on anything else than Series5 and I don´t figure there are any differences between that one and what was planned for the PC. Especially since the concept had a triple initial target, namely PC, console and arcade.

I´ve no idea what the future target markets for PowerVR/IMG are (apart from PDA/mobile and amusement devices), but by now they should be working already on Series6.

Could Series 5 (assuming a timely introduction - often an issue with PowerVR parts!) be one of the first PS/VS3.0 capable pieces of hardware?

May 27th, 2003

http://www.beyond3d.com/forum/viewtopic.php?t=6104

...which may give us some clues as to what shader functionality will be available in PowerVR's Series 5 architecture.

The timing of the demo release is likely due to the release of the ShaderX2 book more than anything else, however it would be insteresting to know if these were developed purely with the DirectX reference rasteriser - after all, even a hardware simulator should be faster!

November 2nd, 2003

http://www.beyond3d.com/forum/viewtopic.php?p=186468&highlight=series#186468

It widely expected, from the technology demonstrations that PowerVR released for the ShaderX2 book, that PowerVR series 5 will be a DirectX9 Shader 3.0 capable part.

March 18th, 2004

http://www.beyond3d.com/forum/viewtopic.php?p=239212&highlight=series#239212

Of course are you right in what your saying and I am aware that I´m pulling straws with this one and yes it´s only a minor technicality, but B3D or any other serious site wouldn´t had reported anything in definite terms.

 

[euan]

well???

 

[Lazy8s]

Sega Sammy have been running a demonstration at E3 called Next Level that showcases footage from some of their next generation software development. The platform(s) that the footage was taken from has not been disclosed, but X360 or their new high-end PowerVR board seem like strong candidates.

 

[Reverend]

I get the distinct feeling that there are quite a number of PowerVR fans at this site.... is that true?

 

[Panajev2001a]

I get the distinct feeling that there are quite a number of PowerVR fans at this site.... is that true?

Look, all I know is that these checks signed Simon Fenney keep coming in the mail: you have all my support .

 

[Reverend]

Seems a bit more certain than that to me. With SM3 demos, they must have had SM3 hardware in the pipeline. Since it's been quite some time since those demos have appeared, this next part being SM3 (or something very similar: it's not meant for a Windows system) is only logical.

Hey, anyone can buy a GeForce6 y'know!

 

[zsouthboy]

I get the distinct feeling that there are quite a number of PowerVR fans at this site.... is that true?

I've never even had a powerVR board, ever.

I just like the possiblity* of someone coming in and kicking nVidia and ATi around, and the technology associated with said kicking.


*faint

 

[BrynS]

...I've never even had a powerVR board, ever...

That's a court marshal offence in my view!

Cheers,


BrynS

 

[Ailuros]

I get the distinct feeling that there are quite a number of PowerVR fans at this site.... is that true?

I claim to be an antialiasing freak; I guess you can figure the rest out.

Hey, anyone can buy a GeForce6 y'know!

Have one since late spring 2004 and am quite satisfied with it. 8)

 

[Dave B(TotalVR)]

I hadn't heard that they were developing explicit support for voxels in hardware. Maybe it was just rumor.

Its not, its just SM3 allows you to accelerate voxel calculations because of the flexibility of the instructions set. This has a use in graphics so a dewmo of it was made by the PowerVR team. With voxels, perhaps, they may be acceleratable sufficiently fast enough to dispel the matra that they are all blocky and horrible because the last time they were used was software rendered. I mean how good did software rendered polygon scene look compared to the Voodoo 1's rendering?


On top of that, which i think would be better, it may be possible to have a kind of LOD arrangement where as you go further out the card starts rendering the scene with voxels instead of polygons. It sounds highly likely to me that polygon rendered geometry with a nice pixel shader on it is going to look better up close than voxels but voxels will look just as good, if not better in the distance and on top of that it would be cheaper (I think) to render massive draw distances fro terrain using voxels.

I say this without knowing how fast they can be rendered, im not sure if we could even refer to a cards fillrate as a meaningful number when rendering voxels with regards to their speed.

Off the top of my head I don't know how they are calculated but I suspect somebody on this board does, care to elaborate?

Hey Kristof, you *can* comment on how voxels are rendered in SM3, we'll just think you're talking about ATi's next SM3 card or something

 

[Dave B(TotalVR)]

I get the distinct feeling that there are quite a number of PowerVR fans at this site.... is that true?

What on earth gave you that idea? 8)

 

[Dave B(TotalVR)]

Seriously, if you havent watched the voxel demo at the powervr site
http://www.pvrdev.com/pub/PC/eg/h/Voxel.htm

watch it, u need a 6800 but it looks amazing.

 

[Rob Evans]

There is a preview of next gen Sega games running on "next-gen, undisclosed arcade hardware" at http://news.spong.com/x?art=8760

 

[amk]

http://www.beyond3d.com/forum/viewtopic.php?t=23184