PowerVR Series 6 now official

[JohnH]

I'm sure there's a witty come back to this thread, but for some reason it escapes me, hmm possible options include,

*Something related to sandwich fillings
*Our plans to start considering Series 5432 at some point or another
*Why/When Rys was last seen wearing bra on his head


*Whistles*

 

[Ailuros]

*Why/When Rys was last seen wearing bra on his head

As I said the codename can hardly be a coincidence harharhar

 

[Megadrive1988]

I wonder if PowerVR Series 6 will get used in PSP2, instead of a high-end implementation of Series 5 / SGX

 

[Exophase]

I wonder if PowerVR Series 6 will get used in PSP2, instead of a high-end implementation of Series 5 / SGX

I'd be pretty surprised, that'd mean Series 6 was available a really long time ago or PSP2 is not coming out for a long time.. Series 5XT can very easily scale well beyond what any competition currently offers. I wonder if Series 6 is even meant for currently available process nodes (or if it really matters)

 

[Ailuros]

I wonder if Series 6 is even meant for currently available process nodes (or if it really matters)

ARM doesn't unfortunately disclose any details about die area under process X and performance characteristics for Mali T604, but I'd expect 2011 to be the 28nm starting point for embedded SoCs in any case. I'd say that under 28nm you could squeeze quite a lot in 5-8mm2 for a future embedded GPU, given that it should target smart-phones.

I wonder if PowerVR Series 6 will get used in PSP2, instead of a high-end implementation of Series 5 / SGX

The feature-set would be most likely too high for the real needs of a next generation hand-held. With a hypothetical SGX543 MP SONY is already ahead the competition; beyond that it would be overkill and think if it would be better to invest X amount of transistors into things a hand-held could never make any use of and or into higher performance?

 

[Ailuros]

I'm too bored to open another thread for that one:

http://www.stockopedia.co.uk/conten...oft-in-move-to-bolster-technology-base-50542/

Sorry for the OT.

 

[Ailuros]

http://www.imgtec.com/News/Release/index.asp?NewsID=608

• New generation POWERVR Series6 (‘Rogue’) graphics technologies, which are already in design with several tier one lead partners, are being presented to a wider customer base

Grrr...............

 

[Simon F]

Grrr...............

????

 

[JohnH]

*Sniggers*

 

[tangey]

????

frustration at the occasional tease but lack of actual info...at a guess

 

[Ailuros]

frustration at the occasional tease but lack of actual info...at a guess

Exactly

 

[cjo]

I guess I need a new sig, the highlighted R was for Rogue!

Is there a coded message in "God I loves me some tuna"? Cause if there is, I'm struggling

 

[roninja]

http://www.pocketgamer.co.uk/r/Various/Imagination+Technologies+news/news.asp?c=27514

[PowerVR Rxx] offers up a speed increase over today's quad-core models by around “20-100 times the power”, yet consumes barely a milliwatt to run.

The Series 6 PowerVR chips (codenamed 'Rogue') are expected to equal today's desktop PC processors – and by extension exceed today's HD consoles – within three years.

 

[MfA]

I'd really want some exact quotes, because I'm pretty sure they didn't say a GPU performing at 20-100X a 5 series quadcore would consume around 1 mW.

PS. I hadn't noticed yet that IMG was pushing Android for Meta (I'm 8 months out of date). Guess they thought that if ARM was going to compete with them they might as well retaliate ... is NDK support on the horizon?

 

[Lazy8s]

That's the marketing part, of course. Still, OpenCL and the like can harness that compute to do all kinds of the advanced recognition, database, editing, etc algorithms that a new generation of mobile apps will need.

 

[Pottsey]

I thought the quote was 20x more than this generation and x100 more than the MX in the Iphone 1.

 

[Lazy8s]

The claim was for Rogue at both points, but something was "lost in translation" as noted considering its total performance at a milliwat wouldn't even be relevant.

20x to 100x is reasonable, though, when ranging across a scalable family across a full generation.

 

[Ailuros]

PS. I hadn't noticed yet that IMG was pushing Android for Meta (I'm 8 months out of date). Guess they thought that if ARM was going to compete with them they might as well retaliate ... is NDK support on the horizon?

No idea if anything has changed but I recall in the past that the META philosophy was to push boundaries up to the point where they wouldn't "annoy" in any way their CPU partners. Since there's a META in video IP for example what would they do if any partner licenses just a video decoder IP for instance for an Android device?

I'd really want some exact quotes, because I'm pretty sure they didn't say a GPU performing at 20-100X a 5 series quadcore would consume around 1 mW.

Warning bad tasted joke: http://www.anandtech.com/show/4177/samsungs-galaxy-s-ii-preliminary-performance-mali400-benchmarked

....depends on the quad-core really *runs for his life meowwww*

 

[MfA]

No idea if anything has changed but I recall in the past that the META philosophy was to push boundaries up to the point where they wouldn't "annoy" in any way their CPU partners.

Why would their CPU partners be annoyed? I doubt say Samsung really minds another CPU architecture being available for licensing, yes they have a lot invested in ARM but even for them a lack of competitive pressure on ARM is not a good thing going forward. ARM, now I imagine they are annoyed at the Android support.

 

[Ailuros]

"Annoyed" in the sense to not create a GPP that's too competitive with a CPU but rather to serve as a general purpose co-processor.