whats the deal with powervr?

Ailuros said:
With unified shaders and procedural geometry/HOS, where exactly are they one generation behind?


In two years, which would be the timeframe to design a desktop chip with that IP from the ground up ;)
 
You seem to forget that there was a Series5 to be released last year.

All it takes is to Design an IP at that level make a couple prototypes and show it to the world. Some Company is going to glom onto it get the party started and the rest is history.

Maybe the world just wasn't interested back then? The initial S5 had a triple target namely PC/console and arcade. Apparently it only made it into arcade boards (which many believe to be SEGA's Lindbergh). What exactly makes you so certain that interested parties never looked over PVR's IP for a console in the past?

ATi just designed an *ip only* Chip package for the Xbox360 that is even 90% of a Generation ahead of the generation that isnt even completely *currently* represented.

90% in what?
 
Are you kidding me?

90% of what... ???

Oh, i dont know... Unified shaders, WGF 2.0 like features, memexport..

The current new generation only being half way introduced becuase R520 has not been released yet and only the GTX is out there at this time. When Xenos is about 90% of the features (in some cases past it) what we will expect anther 18 months from now. So clearly someone can design an IP and get it made by someone else that is not behind current technology.

You know whats really funny, is when you guys try to get all disrespectful and mocking when obviously i'm not the who missed their date with captain obvious.
 
there will never ever be a PC gpu card with PVR EVER. Did i say ever? yeah ever. Nobody in there right mind would jump into the vicard biz with flat PC sales and try and compete with ATi and NVDA. Any sign of a threat and you would see a price drop that would kill a start up.
Then let alone compete with intell and Dell and get in the oem biz side...
 
Hellbinder said:
Are you kidding me?

90% of what... ???

Oh, i dont know... Unified shaders, WGF 2.0 like features, memexport..

The current new generation only being half way introduced becuase R520 has not been released yet and only the GTX is out there at this time. When Xenos is about 90% of the features (in some cases past it) what we will expect anther 18 months from now. So clearly someone can design an IP and get it made by someone else that is not behind current technology.

I'm talking about the family SGX belongs to and that has absolutely nothing to do with the PC or console market, since IMG obviously never entered those for eons now.

If you look carefully into the claimed SGX specifications, you'll see that it's a unified shader core that is been claimed to exceed dx9.0 and OGL2.0 specifications.

If we're talking strictly about the technology portofolio here, IMG is not by far one generation behind, not with SEGA Lindbergh neither with SGX for the future.

As for licenses, major semiconductor manufacturers don't license these days anymore when all they can see is a bunch of tech-presentations on paper; they need more than that:

http://www.imgtec.com/News/Release/index.asp?ID=247

You know whats really funny, is when you guys try to get all disrespectful and mocking when obviously i'm not the who missed their date with captain obvious.

I asked a simple question and I got an answer for that. We can have a civilized conversation, can't we?
 
Last edited by a moderator:
karlotta said:
there will never ever be a PC gpu card with PVR EVER. Did i say ever? yeah ever. Nobody in there right mind would jump into the vicard biz with flat PC sales and try and compete with ATi and NVDA. Any sign of a threat and you would see a price drop that would kill a start up.
Then let alone compete with intell and Dell and get in the oem biz side...

I don't believe in a PC GPU anymore either.

The last sentence just made chuckle though, considering the Dell Axim50v has an Intel2700G inside, which is nothing else but PowerVR's MBX.
 
_xxx_ said:
In two years, which would be the timeframe to design a desktop chip with that IP from the ground up ;)

Actually, a chip designed with that i.p has been up and running since summer 2004.
 
Ailuros said:
I don't believe in a PC GPU anymore either.

The last sentence just made chuckle though, considering the Dell Axim50v has an Intel2700G inside, which is nothing else but PowerVR's MBX.
Ok you got me there, and there will more of that kinda stuff. just not a sku# at Bestbuy to upgrade.
 
TEXAN said:
Actually, a chip designed with that i.p has been up and running since summer 2004.

Where? I was talking about a desktop chip and a gfx card designed around that. Surely there is a part for handhelds, but that's nothing you could put into a PC.
 
Hellbinder said:
The current new generation only being half way introduced becuase R520 has not been released yet and only the GTX is out there at this time. When Xenos is about 90% of the features (in some cases past it) what we will expect anther 18 months from now.

What features are you talking about here really ? What major feature will Xenos support that the R520 and GTX's doesn't ?

And i don't count unified shaders as a feature. It's just a different architecture, just like a TBDR.
 
Last edited by a moderator:
The most likely route (doesn't mean it'll ever happen), IMHO, for PowerVR would be to take back some market share through Intel integrated graphics (Intel licencing PowerVR).
Thanks to the benefits they would be able to release an high-end product and fight for the king of the hill crown.

At least it sounds much more probable than PowerVR suddenly trying to crush the competition out of nowhere...
 
Bjorn said:
What features are you talking about here really ? What major feature will Xenos support that the R520 and GTX's doesn't ?

And i don't count unified shaders as a feature. It's just a different architecture, just like a TBDR.

Unified shaders at this point are a feature. It allows you do do things that the current crop of cards cant do. Like do a z pass accross all "pipelines" in one shot. This is not just an architectural change but has a direct effect on the kinds off things you can pull off. There are also improvements in the Shaders themselves and you have memexport and you have on die cache.

How is having some of these things like unified shaders any different that increasng the instruction count by a few hundered and slapping a new number on it and calling it XX+ wazoo feature.

Unified Shaders are not the same kind of thing as TBDR. you could have a unified "TBDR"
 
Much of Eurasia's functionality is probably also in the PowerVR arcade board that Sega Sammy demonstrated months ago since it uses a Series 5 part as well, even if the implementations of the two Series 5s (streamlining, shader unification perhaps) are different.
 
Ingenu said:
The most likely route (doesn't mean it'll ever happen), IMHO, for PowerVR would be to take back some market share through Intel integrated graphics (Intel licencing PowerVR).
Thanks to the benefits they would be able to release an high-end product and fight for the king of the hill crown.

At least it sounds much more probable than PowerVR suddenly trying to crush the competition out of nowhere...

Interesting thought; in such a case it would very well crush the IGP market and literally out of nowhere. Presupposition would be of course (that's a second conditional...) that Intel would actually chose a quite aggressive implementation and not what they usually have released this far.

Jon Peddie mentioned in one of his presentations about the professional market that Intel might enter the ultra-low end professional market with IGPs. I know it sounds weird, but as he notes himself in that slide, never say never.
 
Hellbinder said:
Unified shaders at this point are a feature. It allows you do do things that the current crop of cards cant do. Like do a z pass accross all "pipelines" in one shot. This is not just an architectural change but has a direct effect on the kinds off things you can pull off. There are also improvements in the Shaders themselves and you have memexport and you have on die cache.

I thought that Ati themselves said that the on die cache (EDRAM) of the Xenos wouldn't be a good idea to put in a PC GPU. And i somehow doubt that the improvement in the shaders of the Xenos are something that we have to wait 18 months for afa PC GPU's goes. I leave it to others to describe what additional stuff you can do with a unified shader approach, other then the additonal loadbalancing possibilities.

Unified Shaders are not the same kind of thing as TBDR. you could have a unified "TBDR"

Obviously, since IMG just announced the SGX. What i meant was that it was simply a different architecture, not an extra feature.
 
Last edited by a moderator:
I can't see PowerVR (nor any other new competitor) coming to PC High End any time soon. I'd love to be prooven wrong here, but I just can't see it happen.

Intel integrating PowerVR tech is possibility, but how's the licensing at PowerVR goes? if you have MBX or SGX license, are you allowed to access Series 5 IP as well, or do you need to license that as well?
 
what about this: http://patft.uspto.gov/netacgi/nph-...,924,810.WKU.&OS=PN/6,924,810&RS=PN/6,924,810
Hierarchical texture cache

Abstract

A dynamically configurable portion of a cache shared between central processing and graphics units in a highly integrated multimedia processor is engaged as a secondary level in a hierarchical texture cache architecture. The graphics unit includes a small multi-ported L1 texture cache local to its 2D/3D pipeline that is backed by the relatively large, single ported portion of the shared cache. Leveraging the shared cache as a secondary level texture cache reduces system memory bandwidth and die size without significant sacrifice in performance.

It was already suggested that in 2007 AMD will release CPu with integrated video controller. AMD has 3 choices:
- do it themselves from scratch - do they have the needed resources ?
- ask one of big 2 - ATi/NV to do custom design, but then this won't be too different from what ATi/NV already sell
- ask one of small players wishing to take into desktops - PowerVR or Falanx ?!
 
hi;

I have one question about the SGX:

How many transistors does the SGX have?
AFAIK we only know the die-area @ 90nm which is 2-8mm²

Or can someone translate the die-area - figure into an approximately transistor-count?

Or does anyone know the die-area of the RV350/RV380 @ 130/110nm for a comparison with the SGX?


Why do I ask this question? Well I want to know how many "pipelines" an SGX could have compared with an RS350 for example to approximate the rough performance of such an embedded graphics chip.


Manfred
 
Back
Top