PowerVR Serie 5 is a DX9 chip?

[Tagrineth]

We can say at least one thing: if the chip's in time and performs sufficiently, it'll be most interesting to watch how Series 5 will fare against ATi's RV350 and nVidia's NV31. As was to be expected, PowerVR is again headed for the sub-200€ market.

Dammit! Well let's hope they pull good performance out of sub-200 Euro... :\

 

[PC-Engine]

As long as they can recoup the R&D costs through good sales, I'm happy as this would definitely mean Series 6 is not far behind

I hope everything goes well for the PowerVR group

 

[Joe DeFuria]

...this would definitely mean Series 6 is not far behind...

DAMNIT! You probably just jinxed Series5 for even THINKING about Series 6....

 

[Ailuros]

ROFL and 3DCenter's sources are?

If (insert as many conditionals as the past track record could suggest) Series5 will make as a PC card on shelves and to that ON TIME, then I'd rather think that some should start saving already

 

[Magnum PI]

ROFL and 3DCenter's sources are?

If (insert as many conditionals as the past track record could suggest) Series5 will make as a PC card on shelves and to that ON TIME, then I'd rather think that some should start saving already

i am saving for the DX9 powervr card.

a lot better than wasting my money on a ATI or nvidia that will soon be obsoleted by the said powervr card.

 

[Ailuros]

a lot better than wasting my money on a ATI or nvidia that will soon be obsoleted by the said powervr card.

That's a very bold prediction and I personally won't go as far, at least until there is something sitting on shelves and can even come close to that.

Also I don't consider an ATI or NV purchase a waste of money; rather the contrary. I want to finally see a high end TBDR; whereby I couldn't care less who would produce it. Difference being that PVR is the only remaining IHV that uses currently the approach.

 

[darkblu]

i'm particularly glad that IMG didn't go for a high-end part (at least not as the series' introductory card) - sub-$200 is the range i consider a videocard to be a viable purchase. as i've been holding long enough for a dx9-level part, IMG has got my money guaranteed if they deliver this year. btw, a dx9-compliant deferred renderer will only gain relative performance with the increase in pixel-shader complexity during the lifespan of dx9.

 

[LeStoffer]

btw, a dx9-compliant deferred renderer will only gain relative performance with the increase in pixel-shader complexity during the lifespan of dx9.

It might well be a bit give and take relative IMHO. You will of course win on not having to render all those hidden, complex pixels, but you may not have much of an advantage in terms of memory bandwidth anymore as the complex shader could mainly be about ALU ops without a lot of data/texel fetch.

But I hope that we will all see some sunny day. 8)

 

[Ailuros]

i'm particularly glad that IMG didn't go for a high-end part (at least not as the series' introductory card) - sub-$200 is the range i consider a videocard to be a viable purchase. as i've been holding long enough for a dx9-level part, IMG has got my money guaranteed if they deliver this year. btw, a dx9-compliant deferred renderer will only gain relative performance with the increase in pixel-shader complexity during the lifespan of dx9.

Could anyone please deliver some reliable proof for a change that it's actually about a mainstream product only?

As for performance what anything concerns let's see the thing and it's specs announced first, shall we?

 

[Ailuros]

but you may not have much of an advantage in terms of memory bandwidth anymore

With how many bits per channel exactly?

 

[Nappe1]

ROFL and 3DCenter's sources are?

If (insert as many conditionals as the past track record could suggest) Series5 will make as a PC card on shelves and to that ON TIME, then I'd rather think that some should start saving already

i am saving for the DX9 powervr card.

a lot better than wasting my money on a ATI or nvidia that will soon be obsoleted by the said powervr card.

uh, oh...
there's too much hope on this post...

I was saving for Axe. Then, I was saving for Parhelia... and year and half after I started to save for new card, I was back to ATI again with more than 3 failed cards...

just wanted to say, don't burn your fingers with "trusted" news. History teached that shit truly happens...

or then it is just me.

 

[Ty]

i am saving for the DX9 powervr card.

a lot better than wasting my money on a ATI or nvidia that will soon be obsoleted by the said powervr card.

Good luck to you. How many times have I heard this from PVR fans before? The Neon250 wasn't the next coming. Neither was the Kyro I, Kyro II, or Kyro II SE. In the end many of these same fans ended up buying NVidia or ATI.

But I wish you luck nonetheless. More competition is a good thing!

 

[KimB]

btw, a dx9-compliant deferred renderer will only gain relative performance with the increase in pixel-shader complexity during the lifespan of dx9.

Not necessarily. It depends on the rendering algorithm used.

First of all, if the rendering algorithm uses a shadowing technique that uses an initial z-pass, then there will be no overdraw when rendering the more advanced pixels. If this sort of rendering becomes common, we will likely see acceleration of z-only rendering, eliminating any benefit a deferred renderer would get here.

Secondly, where are the bottlenecks in rendering? If the bottlenecks are computational, and not related to memory bandwidth, then there will be no benefit from going for deferred rendering.

In the end, I still think that full-frame deferred rendering is not as good for the future of desktop 3D graphics as immediate-mode rendering.

 

[Althornin]

Secondly, where are the bottlenecks in rendering? If the bottlenecks are computational, and not related to memory bandwidth, then there will be no benefit from going for deferred rendering.

Not true.
If the card has to render less pixels, and is computationally (shader) limited, then rendering less pixels will result in a speed up.

standard IMRs must do shading work for pixels that never get displayed.

 

[LeStoffer]

but you may not have much of an advantage in terms of memory bandwidth anymore

With how many bits per channel exactly?

Just because you're rendering at 16 or 24 or 32 bits FPU precision doesn't mean that you need texel data of the same precision.

 

[KimB]

Not true.
If the card has to render less pixels, and is computationally (shader) limited, then rendering less pixels will result in a speed up.

standard IMRs must do shading work for pixels that never get displayed.

Did you notice the part about the initial z-pass?

 

[Ailuros]

Did you notice the part about the initial z-pass?

I bet he did as he did of course notice the conditionals involved and the sentence that followed it:

if the rendering algorithm uses a shadowing technique that uses an initial z-pass, then there will be no overdraw when rendering the more advanced pixels.

Followed by:

If this sort of rendering becomes common, we will likely see acceleration of z-only rendering, eliminating any benefit a deferred renderer would get here.

 

[Althornin]

Did you notice the part about the initial z-pass?

Um, that was in the first IF-->THEN

re-read your post.
If you say "First of all" if blah blah blah, then blah blah blah.

Then you say "Secondly, blah blah blah blah",
the IF is not part of the "secondly" statement. That constitutes an entirely new statement, without the first statements conditional.

However, if that was what you meant, then of course, but thats not what you wrote.

 

[ActionNews]

i am saving for the DX9 powervr card.

a lot better than wasting my money on a ATI or nvidia that will soon be obsoleted by the said powervr card.

Good luck to you. How many times have I heard this from PVR fans before? The Neon250 wasn't the next coming. Neither was the Kyro I, Kyro II, or Kyro II SE. In the end many of these same fans ended up buying NVidia or ATI.

But I wish you luck nonetheless. More competition is a good thing!

Hmm..right, but this was more or less the licencees fault!
NEC wanted to have a chip for their DreamCast first, then they concentrated on Neon250. I don't think that NEC ever was very interested in PC market!
STMicro redesigned KyroII and KyroIISE to get higher clockrates, so this isn't PowerVRs fault either!
Now at the moment PowerVR has no licensee (at least not that i knew), so i hope PowerVR has a "Plan-B" if none is willing to produce Series 5. Or perhaps ImgTec finally realised, that in PC space it's better to produce directly at TSMC or UMC and not to rely on a licensee !

I hope PowerVR developes Series 5 as a Highend part, because if you have the fastest card on market, mainstream and lowcoast cards sell better, too. It's a matter of image! And if it's right what John Metcalfe says in the EE Times interview, that Series 5 has a very flexible design, then ist should be easy to serve the mainstrem and lowcost market, too. So i hope we will not only see one Serie 5 card, but a range of cards heading for all different markets !

CU ActionNews

 

[KimB]

Or perhaps ImgTec finally realised, that in PC space it's better to produce directly at TSMC or UMC and not to rely on a licensee !

There's only one problem with that. Money.

I hope PowerVR developes Series 5 as a Highend part, because if you have the fastest card on market, mainstream and lowcoast cards sell better, too. It's a matter of image! And if it's right what John Metcalfe says in the EE Times interview, that Series 5 has a very flexible design, then ist should be easy to serve the mainstrem and lowcost market, too. So i hope we will not only see one Serie 5 card, but a range of cards heading for all different markets !

CU ActionNews

I hope not. As I've said many times in the past, I feel that fully-deferred rendering is the wrong way to go. Deferred rendering will have problems as polycounts increase in the next few years.