PowerVR Series5 still alive and it's "high end"

[Joe DeFuria]

I think part of the problem with PowerVR always being behind the curve is their business model (selling IP).

Yup, as I've stated counteless times before.

Their IP model works well with console and/or arcade or other types of business where the turnover is measured in several years, not several months. If PowerVR is plans to be a serious competitor in the PC space (not just a 'one hit wonder', if that), they are going to have to, IMO, go with a more traditional fabless semiconductor model. (Again, at least for the PC space.)

Hopefully, something is wrong with what I've said, or they've found some brave souls to step up to the plate without the goods being finished.

Either even better, perhaps ImgTech has decided to build and sell actual chips like we're all hoping

 

[jjayb]

That would be great to have 3 major players. But do you really think it's going to happen or is everyone just smoking something hallucigenic?

 

[RussSchultz]

Either even better, perhaps ImgTech has decided to build and sell actual chips like we're all hoping

*cough* www.puredigital.com *cough*

Of course, no proof. But I think that's what that arm of the company is all about. (productizing their IP in markets where they're competitive in hope of OEM sales "after the fact")

 

[Joe DeFuria]

*cough* www.puredigital.com *cough*

Of course, no proof. But I think that's what that arm of the company is all about. (productizing their IP in markets where they're competitive in hope of OEM sales "after the fact")

Sort of...Pure Digital (formerly known as "VideoLogic Systems") does productizes their IP, but in the form of consumer level products, not OEM components. They once sold Kyros, Neons, and Apocalypse 3Ds for example. However, they don't sell (at least in the past) "Imagination Technology" chips, or even consumer products WITH Imagination Technology chips, because Imagination Technology chips never existed.

Thier PowerVR Video products, for example, had NEC or STM chips in them.

So, essentially, for example, NEC bought ImgTech IP, which NEC used to fab a chip. Then NEC turned around and sold the finished chip product to VL Systems / Pure Digital. Pure Digital takes the chip and makes the consumer level product.

 

[Rob Evans]

Either even better, perhaps ImgTech has decided to build and sell actual chips like we're all hoping

*cough* www.puredigital.com *cough*

Of course, no proof. But I think that's what that arm of the company is all about. (productizing their IP in markets where they're competitive in hope of OEM sales "after the fact")

*cough* www.frontier-silicon.com *cough*

 

[Rodéric]

*cough* www.frontier-silicon.com *cough*

"Frontier Silicon is a fabless semiconductor business", whose fabs are used then ?

 

[Kristof]

*cough* www.frontier-silicon.com *cough*

"Frontier Silicon is a fabless semiconductor business", whose fabs are used then ?

Depends on the product. IIRC the new DTV chip is being made by Samsung... then again I could be wrong since I am not really up to speed. Remember that NVIDIA is also a fabless semiconductor...

 

[Grall]

Hellbinder:

What makes you think PVR is implementing external this and external that in series 5? I don't make that interpretation at all. All these devices would logically sit on the same die but rather use clock/power gating instead to make for a cooler running chip. For example.

Not knowing exactly what the guy meant makes it difficult to guess, but I am pretty certain it will not feature external chips, since that - especially from a cost standpoint - really is crossing the stream to get water. Ie, not much of a saving at all.

I also don't see why geometry has "always" been considered any kind of limiting factor. Tim Sweeney - the guy who can't program an engine worth a damn you know - made some confused statements regarding TBRs (amongst other things they couldn't use hardware geometry acceleration), one of which sparked off this debate. Others have theorized it could be a limiting factor, etc. NEVER was it shown to actually be the case in anything BUT theory, and Kristof or whomever guy at PvR have stated here in fact they had ways for future products to circumvent that issue.

Of course they do, they invented the darned thing!

Still it seems this stupid rumor refuses to die. I'd prefer you not continue to spread it around as it amounts to FUD at best. Thank you.

Worry more about your own credibility instead, your assurances R350 would have distinct new features/tech etc didn't exactly pay off...


*G*

 

[megadrive0088]

Oh one more thing for Simon and Kristof.

I really hope your company takes the Handcuffs off your hardware this time around.. and *Seriously* address the High-End. Dont just go with barely enough Core/Ram/Bandwidth to eek out a few wins here and there.

There is no doubt in my mind that you guys could pull a 9700pro on the entire industry if you wanted to. It is obvious that every time you release a new product.. *if* it had been clocked the same as the competition... it would have completely Demilished even the closest competitor.

For once in your companies life.. Make a damn statement for a change. As personally, i think that an 8 piped TBR card clocked at the same MHZ as the Nv35 would be MANY % faster in most benchmarks. i could wager based on past experience 200-250% faster just like the 9700pro was over the GF4.

Like PowerVR2 PC over TNT1 and the rest of the gaming chips in 2H 1998, IF it had been released then....

 

[Entropy]

Metcalfe added that the Series-5 architecture would debut in 2003 in a 0.13-micron process technology. He said that it was not yet decided whether Imagination would get first silicon implementations or test chips made at a licensee's wafer fab or at a foundry wafer fab in Taiwan.

Now this was in October 2002.
It would seem that at that point in time, they hadn't struck a deal with a particular liscensee, but were still talking.
"..at a licensee's wafer fab..." is interesting though. Not a whole lot of fab owners around. Out of the ones who do own fabs, the only one I can think of that might be interested in liscensing would be Intel (particularly since they are already doing business with ImgTech). I can't see STM doing a flip-flop, particularly since they dropped the Series4, and dumped the portable gaming platform, a niche a number of other players are attacking with gusto. OTOH, Intel has been content to eat up the gfx-market from the bottom up, would they really go into making discrete chips again?

Hmm.

Entropy

 

[Hellbinder]

Grall..

First of all, if you dont think that PowerVR designs are Geometry limited, and in turn Bandwidth limited in those cases.. You may want to go read a few more articles. Secondly, it was speculated for a long time that PowerVr might overcome some of its Future Geometry issues by splitting the Geometry workload to a seperate processor. In fact. They did just that with the Naomi 2 Arcade board.

Still it seems this stupid rumor refuses to die. I'd prefer you not continue to spread it around as it amounts to FUD at best. Thank you.

Worry more about your own credibility instead, your assurances R350 would have distinct new features/tech etc didn't exactly pay off...

The R350 DOES have significant hardware differences. Like the Fbuffer, increased Vertex cababilites, improved memory controler etc. How do you not consider the F-Buffer a significant new feature??? You are really trying to stretch something i said into something it was not.. clearly for your own personal motivations.

Personally, i think my credability is just fine thank you.

 

[Brimstone]

I'd like to see Intel fab Power VR VPU's and Microsoft include one in the XB 2. That would be a great combination if Power VR still has the juice to hang with Nvidia and ATI these days.

 

[megadrive0088]

I'd like to see Intel fab Power VR VPU's and Microsoft include one in the XB 2. That would be a great combination if Power VR still has the juice to hang with Nvidia and ATI these days.

agree!

 

[Joe DeFuria]

I'd like to see Intel fab Power VR VPU's and Microsoft include one in the XB 2. That would be a great combination if Power VR still has the juice to hang with Nvidia and ATI these days.

That would be interesting indeed, and I can see something like that happening. (Even if it's not Intel fabbing the chip.) I do believe that a console chip is something PowerVR can pull off. (Duh...see Dreamcast...) The console is a "closed" system, so devs can cater to PowerVR tech strengths without having to worry about other architectures.

Though a possible major challenge could be maintaining full backward compatibility with XBox-1, not only because of the different architecture, but a different manufacturer as well.

 

[Tahir2]

This is where DX compatibility can help, no?

 

[Joe DeFuria]

This is where DX compatibility can help, no?

Well, it helps immensely, but it's not a guarantee.

Look at the different "compatibility" levels among different hardware within a single DX revision on the PC. Every architecture has its nuances. Different IHVs can "interpret" the spec differently. And if if they are interpreted the same way, there are different performance characteristics for using different hardware in different ways.

What works on one piece of hardware, might not work right on some other hardware, or have unacceptable performance on some other hardware.

This isn't too much of a problem when devs are coding for multiple platforms simulatenously, as they can be addressed readily. But right now, the X-Box 2 chip doesn't exist.

If ATI wins the contract, they'll have similar challenges.

 

[Tagrineth]

Well... as long as devs stick to DX interfaces (more or less)...

The XBN's chip could support DX despite any performance caveats.

Ideally, the XBN's graphics chip will very happily and decidedly outclass NV2A, not being slower than it at any one thing... but it could still conceivably dive to similar performance levels, for some mean frame rate fluctuation.

But of course... that's where Vertical Sync comes into play 8)

 

[PC-Engine]

Secondly, it was speculated for a long time that PowerVr might overcome some of its Future Geometry issues by splitting the Geometry workload to a seperate processor. In fact. They did just that with the Naomi 2 Arcade board.

Geometry issues??? NAOMI 2 used a separate TnL processor because it didn't need to be integrated into a single chip. It's an arcade board and the majority of arcade boards that existed used multiple DSP and coprocessors. The fact that the TnL chip was separate from the PowerVR chip has nothing to do with the PowerVR architecture.

 

[asicnewbie]

I'd like to see Intel fab Power VR VPU's and Microsoft include one in the XB 2. That would be a great combination if Power VR still has the juice to hang with Nvidia and ATI these days.

If Intel doesn't license the core, then I think the odds are very low. From a business standpoint, if IMGTEC is paying for the silicon validation costs, the foundry with the largest potential customer base (TSMC) is the logical choice. Then depending on market response, IMGTEC may choose to validate its IP on additional foundry lines.

http://www.siliconstrategies.com/story/OEG20030505S0064 - IBM breaks through in Semico 2002 foundry rankings

This article shows the sales rankings (by revenue) of merchant foundry industry. TSMC holds 1st place and commands a huge lead over the 2nd-place runner-up (UMC.)

...

On a lesser note, I don't think Intel has ever produced peripheral logic (chipsets) on its cutting-edge fabs. (On the other hand, the 0.13u process node is 'old' for Intel.)

 

[Rodéric]

Would prefer PowerVR in a NINTENDO console or SEGA console over any other, since I'll never buy a MS console, and I'm not likely to buy a SONY console either...