PowerVR-Based Systems Compare/Contrast

[Megadrive1988]

Megadrive1988:

I've never heard of Aurora until now, interesting. I'd like to more about it.

There's a short rundown of information near the end of the SEGA 2010 thread.

I'd also like to find out about a much older PowerVR based system,

I do remember that time. Namco was even eventually planning for a multiplatform strategy with optimized ports/conversions of their best arcade games, like the Ridge Racer series, to PowerVR equipped PCs. And Kalisto also had the nice looking Ultim@te Race title.

oh yes. I remember that pretty well

there were press releases and other official statements from Videologic (and I think NEC as well, but not from Namco themselves) that Namco was bringing Air Combat 22, Rave Racer and Tekken (or Tekken 2) to the PC platform, via PowerVR cards. there was even a statement by one of the Videologic guys that Rave Racer on PowerVR was better than the arcade version.


http://groups-beta.google.com/group/comp.sys.ibm.pc.games.action/msg/3b9fc4a80e535fbc?dmode=source

Date: 1996/01/31
the PowerVR system will get a similarly powerful ally given
that Namco and NEC have already announced their alliance. Rave Racer,
Tekken and Air Combat 22 are all now complete and ready for release in
May for the PowerVR system. VideoLogic claims that these conversions
are not only faithful to their arcade predecessors, they actually
surpass them. "If you take arcade Rave Racer and make it a little more solid and a little faster the you've got the idea. It's better than
the arcade version."

that was originally in Next Generation (and probably EDGE) magazine

Also, in a later issue of Next Generation (and EDGE i think) there was a 2 page preview on Rave Racer for PowerVR. It was said in that article, that PowerVR Rave Racer was doing 4 times the graphics work that *Playstation* Ridge Racer was doing. now if you realize that Playstation Ridge Racer was, graphically, 1/4 or 1/2 (probably 1/4) that of the arcade Ridge Racer, and arcade Rave Racer was doing significantly more than arcade Ridge Racer, you would come to the conclusion that PowerVR Rave Racer was, although not as impressive as arcade Rave Racer in terms of polygons or framerate, still very impressive.

Arcade Rave was 60fps, PowerVR Rave Racer was 30fps. I believe the eariler comment that PowerVR Rave Racer was more 'solid' and whatnot, comes from the fact that PowerVR series 1 had more rendering features than the more powerful System 22 board.

edit: if you want to see that article, it is in Next Generation number 19, with NiGHTS on the cover. the Next Generation with the other comment, about Rave Racer being better than the arcade, it's in an older issue, and I don't remember which one.


I wish I had some screen shots of PowerVR Rave Racer.... to show you but there are NONE online that I am aware of. you'll have to find the magazines.

 

[Megadrive1988]

a post that I found on the Namco - PowerVR subject via googling

(yes you will notice that System 33 and System 23 are confused and used interchangably)

http://www.system16.com/cgi-bin/www...&view=expanded&mode=threaded&sb=7

Namco also had a Tekken XX running on the PowerVR PC chip. The plan was for Namco to get familar with programming with the PowerVR chip and later on make arcade games on their future System 33 board which was supposed to be PowerVR based. NEC manufactured the PowerVR chips and they wanted Namco to use them in future arcade boards. Same with SEGA.

> I have no idea at all about the accuracy of the information but their arcade
> related info is usually reasonably reliable. Also with the timeframe of sys23
> it'd be original powerVR not Naomi's PowerVR2. One other small thing worth
> mentioning is that it was well known at the time that Namco were planning to
> make a PowerVR rendition of Rave Racer (which got canned) for the PC, which may
> have been a programming exercise. I take it the graphics chips are unlabelled
> on the PCBs?
>
> > > I've no idea of the authenticity of the statement (and I know that the named
> > > leading game is wrong) but as your entry on sys 23 gfx h/ware is empty I
> > thought
> > > I'd mention it, and if anyone can confirm or deny the info then at least
> I've
> > > learned something
> > > Quoted from Edge magazine issue 37 page 119:
> > > "It's expected that Namco will unveil System 23 at the JAMMA show in mid
> > > September. It's based upon Videologic's PowerVR technology and Tekken 3
> will
> > be
> > > the game to showcase the hardware.Just how many PowerVR chips are running in
> > > parallel isn't known but the system's ease of expansion is a feature that
> > could
> > > allow the placement of multiple chips in parallel to deliver in excess of
> > > 2,000,000 polygons per second."
> > >
> > Hmmm, I would be frankly amazed if both Namco and Sega went with the same
> chips,
> > and theres no mention of Namco on the Videologic website, I think they may
> have
> > their facts a bit upside down with the naomi here...and I have no idea where
> > they got Tekken 3 from
> >
> > Six...
> >
> > System16 - The Arcade Museum

it is interesting.... I gleened from this post something that I had not realized. that Namco's efforts to convert System 22 / Super System 22 and System 11 games to PowerVR Series 1 cards (which ive always been aware of) was probably a programming exercise to get them familar with PowerVR architecture, so that they could later make Tekken 3 and other games on more powerful PowerVR based boards.


side note: Sega also worked with PowerVR long before the PowerVR2 based Dreamcast and NAOMI systems. Sega converted its Model 2 arcade game Virtual On to PowerVR series 1. but unlike the Namco efforts, PowerVR Virtual On was actually released. does anyone know how the PowerVR version of Virtual On compared to the Saturn version and the Model 2 arcade version ?


I also remember rumors...and only rumors, nothing close to official.....that there was maybe going to be a PowerVR upgrade card for NEC's PC-FX game console. and it could certainly have used it, since the PC-FX was, from what little I know of the whole thing, a stripped down version of Hudson's Tetsujin / Ironman board, which was more capable of polygon graphics than the released PC-FX, although certainly not to the extent of even the earliest versions of PowerVR1 technology.


It would have been incredibly interesting to see a PowerVR upgraded PC-FX and a Real3D or PowerVR upgraded Saturn go against the Playstation and Nintendo64 which were built from the ground up to handle 3D.

 

[Iron Tiger]

side note: Sega also worked with PowerVR long before the PowerVR2 based Dreamcast and NAOMI systems. Sega converted its Model 2 arcade game Virtual On to PowerVR series 1. but unlike the Namco efforts, PowerVR Virtual On was actually released. does anyone know how the PowerVR version of Virtual On compared to the Saturn version and the Model 2 arcade version ?

I remember seeing this game in an EB shortly after it came out. The screenshots on the box looked like the Model 2 version, and I'd assume it would run closer to that version (with perspective correct textures, no polygon flickering, and higher resolution textures) than the Saturn version, but with a framerate of around 30fps.

 

[PC-Engine]

I have VO for PCX-1 however I haven't seen the Saturn or Arcade version so I don't know how it compares.

 

[Lazy8s]

It was said that the PowerVR-optimized PC port of Virtual ON: Cyber Troopers was by far the best conversion of a game from a traditional Japanese console developer to the PC up to that point. Most of the prior attempts to convert Japanese games to the PC, especially those coming from the arcade, had resulted in disappointing versions which had none of the smoothness/speed of the originals.

Namco's early familiarity and experience with PowerVR hardware helped to explain their spectacularly polished Soul Calibur launch effort for Dreamcast.

 

[PC-Engine]

Something Simon would know is does the CLX in DC still use the Infinite Planes technology? I know the PCX-1/PMX-1 did.

 

[Lazy8s]

Were infinite planes calculations used in hidden surface removal? It's possible that the architecture still operates similarly yet provides a more standard programming interface to developers.

 

[Megadrive1988]

Something Simon would know is does the CLX in DC still use the Infinite Planes technology? I know the PCX-1/PMX-1 did.

from what I remember, PowerVR Series 1 accelerators: PCX1, PCX2 and any Series 1 based chip, used infinite planes.

whereas PowerVR Series 2 used in Dreamcast, NAOMI, Neon250, NAOMI2
(PowerVR2DC / CLX2, Neon250 aka PowerVR250 (sometimes called PowerVR2PC?) used tiling technology and not infinite planes.

Series 3 and MBX use tiling technology also. don't know about the canceled Series 4, or upcoming (?) Series 5, but I assume they use some form of tiling technology also, unless PowerVR guys / Imagination Technologies has moved onto something else for Series 5 (or Series 6)

but I could be wrong.

edit: PCX-1 is not the same as PMX-1 - PCX-1 is Series1, as is PCX-2 and PCX-1 specifially, was the first production / consumer PowerVR chip.

the PMX-1 is actually Series 2 technology.

PMX-1, I believe is what was known as PowerVR Highlander. the Highlander technology, is basicly a super-set of PowerVR2, that encompased multi-media technologies including sound (or maybe not sound) 2D, 3D, MPEG, and a PC interface. and I think that Neon250 *is* PMX-1. aka PowerVR250. The Neon250 of course is Series2, but has less 3D processing power than PowerVR2DC / CLX2, even though the final clockspeed of Neon250 / PMX-1 was higher than PowerVR2DC.



key / legend:

PCX-1, PCX-2 = PC Accelerator 1 and 2 ? [Series 1]

PMX-1 = PC Media Accelerator or Multi Media Accelerator ? [Series 2]

CLX - Console Accelerator ? console and arcade version of Series 2

PCX-3 .... the 3D-only version of PowerVR Series 2 for PCs - possibly not ever released, or, is the 3D portion of Neon250 aka PVR250 aka PMX-1 aka Highlander ?? it is possible that the PCX-3 was never an official designated name by Videologic, but I clearly remember that named used in the magazines and websites. the PCX-3 might have also had stronger 3D capabilities than the PMX-1 even though both are based on Series2 technology. another possibility is that PCX-3 had the same or similar capabilities as the CLX2 / PowerVR2DC, since the PCX-3 was meant to be dedicated to 3D, unlike the PMX-1 which had to do other tasks. but as I said, it is also possible that PCX-3 is the 3D part of PMX-1 / Neon250, that is, if PCX-3 existed at all.

*Neon250 aka PowerVR250 aka PMX-1 aka Highlander was also known as PoweVR2PC ?? [Series 2]


KYRO and KRYO II = Series 3

MBX = Mobile Accelerator ? [Series 3. possibly with some elements of canceled Series 4 ?]









I'm sure Simon F can correct anything I got wrong.


edit: I have no idea if this is accurate (regarding PCX3) but:
http://home.scarlet.be/~pin10741/PVRSGFaq.htm

Q: What will the PCX3 3D only solution offer graphically (speed, visual quality, MPEG2, etc.) over the 2D/3D Highlander.

A: The 3D only version of PowerVR Second Generation will offer the same feature set and similar performance to the PCI version of the combined 2D/3D chip without the 2D and video functionality.

same feature set, but similar (not same) performance. that leads me to believe that PCX3 is not exactly the same chip as PMX-1 / Neon250, and therefore, PCX3 was not released. I don't recall any other Series 2 chips for PC, other than Neon250.

okay, now most followers of PowerVR often say that if PowerVR2 had been released on time for PCs, instead of being a year (or more) late, that it would've swept the industry. I agree. yet imagine if the even faster Series 2 based PCX3 had been released, with more performance than Neon250 / PMX-1 I suspect the difference between Neon250/PMX-1 and PCX3 would've been like that of TNT2 and TNT2 Ultra or GeForce 2 GTS and GeForce 2 Ultra. etc.

 

[PC-Engine]

All of the PowerVR chips released to date used tiling. Infinite Planes have nothing to do with tiling. IP is for creating shadows/light rays. There's also the modifier volumes thing which I don't know if it's the same as IP or not. I always thought DC supported modifier volumes.

 

[Megadrive1988]

btw here are some if not all of the differences between the PC and Dreamcast versions of PowerVR Series 2.

DC = CLX2 (or just CLX) / PowerVR2DC
PC = PMX-1 / Neon250 (sometimes PowerVR2PC)

http://www.beyond3d.com/forum/viewt...ght=&sid=d68dff200eaffcad17204dc1bf58f495

Just to clarify some of the (key) differences between The PC and console versions of the series 2 HW,

CLX tile was 32x32, PC part was 32x16.
CLX support alpha test with HW front->back sorting to deliver massive effective fillrates when it was use, this was not available on the PC part.
CLX tiling was completely handled by HW, the PC part was 50:50 split between HW and SW
CLX include latency buffering that allowed VQ and palettised to run at full rate, this was removed from the PC part resulting in 50% performance for those formats.
CLX support 2 or 3 (think it was effectively 3 wasn't it Simon ?) external 32 bit memory busses (think in terms of much hyped NV/ATi crossbars), The PC part utilised a single 64 bit bus resulting much greate page break impact.

This all adds up to the PC part being between 50-75% of the performance of the console part inspite of the higher clock rate (125 vs 100 MHz)...

...The differences between CLX and PMX1/Neon250 were all to do with minimising chip area...

so Dreamcast's PowerVR2 was significantly more powerful than PC Neon250.

makes me want to know more about the 3D-only PCX3

Simon ?

 

[Lazy8s]

Megadrive1988:

okay, now most followers of PowerVR often say that if PowerVR2 had been released on time for PCs, instead of being a year (or more) late, that it would've swept the industry.

To give an idea, this is what happened the last time PowerVR squared off against the players in the PC market according to Anandtech's tests:

"With its tile based rendering algorithm, the Kyro II provides blazing fast performance considering the price and was actually able to beat products almost $200 more than the cost of a Kyro II based board. Throughout the benchmarks, the Kyro II based 3D Prophet 4500 simply dominated everything else in its price range. The Kyro II was ready and able to tackle any game we sent its way."
http://www.anandtech.com/showdoc.html?i=1435

PC-Engine:

I always thought DC supported modifier volumes.

It does. Later PowerVR chips went with the more generalized stencil support because it wasn't worth trying to get developers to utilize hardware-specific features that weren't being supported in the DirectX API.

 

[Simon F]

Something Simon would know is does the CLX in DC still use the Infinite Planes technology? I know the PCX-1/PMX-1 did.

In the PCX 1 & 2 systems, objects were defined by the "intersections of half-spaces" where you have a plane and one side is defined as "inside" and the other "outside". Convex objects with "any" number of sides could be passed to the HW. Polygons were handled as a subset of these. A cube, for example, was defined with just 6 planes while a triangle required 4.

In Series 2, the hardware only accepted triangles and quads (directly as vertex data) and was optimised to handle these reduced cases.

Series 3 etc, only handled triangles.

Does that answer your questions?

from what I remember, PowerVR Series 1 accelerators: PCX1, PCX2 and any Series 1 based chip, used infinite planes.

whereas PowerVR Series 2 used in Dreamcast, NAOMI, Neon250, NAOMI2
(PowerVR2DC / CLX2, Neon250 aka PowerVR250 (sometimes called PowerVR2PC?) used tiling technology and not infinite planes.

All PowerVR chips use tiling and deferred rendering.

PCX-3 .... the 3D-only version of PowerVR Series 2 for PCs - possibly not ever released,

I don't what the magazines were talking about, but PCX-3 in my documents only went as far as a paper concept, e.g. adding virtual texturing (over PCI).

It would not have lasted because MS changed the programming model. When PCX1/2 were out, it was feasible to have seperate 2D and 3D graphics cards. In Windows 95/98 the 3D driver was able to capture and patch calls to the 2D driver. This was stopped in NT etc which meant it was infeasible to keep the two systems separate.

 

[PC-Engine]

Ok so DC did not use Infinite Planes since it's based on Series 2. Was there any benefit to using Infinite Planes that you could think of if you were to use it with a proprietary API and graphics system? In other words if you were to design a graphics processor that didn't need to be compatible with DirectX, would Infinite Planes have offered any benefits?

 

[Megadrive1988]

I don't what the magazines were talking about, but PCX-3 in my documents only went as far as a paper concept, e.g. adding virtual texturing (over PCI).

It would not have lasted because MS changed the programming model. When PCX1/2 were out, it was feasible to have seperate 2D and 3D graphics cards. In Windows 95/98 the 3D driver was able to capture and patch calls to the 2D driver. This was stopped in NT etc which meant it was infeasible to keep the two systems separate.

so thus, Neon250 / PVR250 / PMX-1 was released, and PCX-3 was not.