ST-Ericsson Nova A9600: dual-core ARM A15, PowerVR Series 6

Looks like ST shared a core expertise of TI: pre-announce a year (or more) in advance but don't deliver. I'm fine with futuristic roadmaps showing some integer number increase in performance: at least it has some inspirational value. But announce a very specific chip so far ahead? What's the upside in that, other than a very short lived spike in publicity? Why even bother?

 

Exactly.

 

Imagination PR has released a Series5XT/Series6 roadmap for the foreseeable future:



https://twitter.com/ImaginationPR/status/279583082955300865/photo/1

The G6200/6400 have a quite weird placement when it comes to timeframe, as the box ranges a tad into 2012 which sounds unlikely IMO if I should interpret it as device availability (I had more or less mid 13' in mind). Then again it's a PR/marketing thingy; it could mean all or nothing.

 

some more than just "unlikely" given we have 16 days left in 2012 !

Interesting that its not an IMG slide (someone else has the copyright).

Finally what do you think the dotted arrows are pointing to, or are they just showing relative performance heights ?

 

It's from PCWatch: http://pc.watch.impress.co.jp/docs/column/kaigai/20121214_578661.html

Use an online translator for japanese->engrisch

Nothing particularly new apart from the 6630 block diagram:

http://pc.watch.impress.co.jp/video/pcw/docs/578/661/p14.pdf

On an unrelated note:

http://www.imgtec.com/corporate/pre..._Half_Year_Results_Presentation_12dec2012.pdf

Page 22: I wonder what the important design win with Samsung stands for.

 

So that's another soc maker that probably won't license high performance powervr cores again.

 

For me it looks more like Jan 2013. So a release at CES 2013 could be possible or maybe at the next mobile world conference 2013 (25. - 28. February 2013)

 

The royalties they would have generated for IMG had to have been a drop in the bucket compared to Apple, Sony, and possibly even Intel. TI is a much bigger loss here, but they weren't exactly using leading edge configurations either..

 

Under that light neither Sony and in the very least Intel have released so far any devices with high end IMG GPU IP and I'd even argue that Intel's market share is smaller than TI by far. How many Clovertrail sales would they exactly need to cover the Amazon Kindle sales volume alone?

Now I have no idea if Ericsson will release the A9600 or not after all, but the target at least according to my undestanding for it were/are windows based tablets and smartphones with possible adopters like NOKIA and/or Sony , who could really use a hw renaissance at least IMHO.

 

SGX543MP4 is not a high end configuration? While PSVita may not be a big success for its markets its volumes are still pretty high compared to what a lot of other SoC vendors get..

I don't argue that Intel's market share is smaller than TI's, what I said is that it's probably bigger (a lot bigger) than what ST-E's would have been with A9600.

You're forgetting Cedartrail. Although the market has dried up Atom netbook sales haven't gone totally bust and probably continue to quietly sell far more than Clovertrail or Medfield at the moment.

 

Apart from apple, mediatek is the big IMG royalty contributor, and is accelerating. 10m chips in 2012, 110m in 2012, and forecast to double in 2013. All the public road mapped smartphone chips from mediatek in 2013 are IMG graphics

IMG conference call did make a remark that intel's royalty contribution should not be viewed as insignificant. I suspect that some of their medfield handset solutions are selling ok volumes in non European/USA markets.

 

No I haven't forgotten about it, it's just that handheld consoles don't necessarily belong into the same bucket as tablets or smartphones. Even if while I don't have a clue what the Vita sales statistically look like up to now, I have severe doubts they're even remotely close to say 10Mio units/year.

How many "ifs" can I fit into one single sentence to reply to that one? If things would had been different and ST wouldn't but out the SFF market, if A9600 would appear early enough, if adopted by big enough customers, if it would turn out a sales success considering its target market and if it can be re-used for N future SoC variants etc. how many unknown variables/constants do I exactly have to answer that question?

We don't even have a clue if it has been cancelled or not, I'm merely assuming that they might have put it on ice. Let's assume it lives and further assume that as an example both SONY and NOKIA use if for future high end products, is your assumption still valid?

Neither/nor are remotely close to being called "high end", nor can I imagine that Cedartrail and Clovertrail combined could give Intel yet a significant enough tablet market share to exceed the statistical "others" category at the end of 2012 when yearly statistics slice up the tablet market according to SoC manufacturers.

The problem with those two is that they don't have anything impressive on the hw side especially compared (in a vacuum) compared to Apple's past and current tablet offerings. It's not particularly surprising that Intel is marketing for its own upcoming GenX based graphics tablets 3x or 10x times performance (I've frankly lost track with those kind of claims) for that one. Would you guess that a hypothetical A9600 would stand above that one or below in terms of graphics performance?

 

Chinese deals will inevitably lower royalty margins IMHO. It might not be directly related to ST-E, however it's one of the side effects IMG probably already knows it'll face. Else higher volumes, lower margins.

I don't take part in those obviously so I can't know all the details. However considering the content of the recent interim results, the few notes regarding that topic I've read in that one sounded to me more like a reminder that Intel is still and will continue to use IMG IP, future tablets excluded.

 

Future tablets may not feature PowerVR graphics but VXD still features in Baytrail for instance.

 

I'm not talking about tablet share here, just what kind of license revenue IMG is receiving. Maybe I took the wrong implication here but the comment "that's another soc maker that probably won't license high performance powervr cores again" made it sound like this is a huge blow to IMG's license revenue. It probably isn't.

And that's why I brought up Cedartrail, not because I think it'll be in tablets now.

Yeah, you could argue A9600 would have done amazingly well under just the right kind of hypothetical circumstances, and you could probably do the same with any product. But in the real world they never had a great shot and looking at ST-E's track record they really never had a chance of releasing in time to make it a very aggressive competitor. They didn't just suddenly get struck with bad luck either, if it's folding now it's because they haven't been slipping behind for a long time. I think people got too excited because they happened to be the first one who announced a Rogue product. That doesn't really actually mean anything about who could deliver first.

If it did similarly to how ST-E's Cortex-A9 SoCs have been doing? Yes, I'd still call it a small player for IMG's revenue.

Future Intel tablets anyway, unless you're making the bold assertion that Intel will take over the tablet market.

 

Even with a late 2013 or early 2014 release for the A9600 they would had been quite a bit ahead in terms of performance compared to what Intel is planning for the same timeframe. When is the halo effect bigger: when you have something mid or low range and more or less comparable hw to the competition or when you have something high end that differentiates itself quite a bit from the vast majority?

The latter is my only other point in terms of hypothetical possibilities and it is my understanding that the 9600 might have been an attempt to aim higher this time.

Again note that I'm using past tense because my gut feeling tells me that we won't see it. In the contrary case I'll be of course pleasantly surprised.

Are any of them material you'd call "high end"?

 

This thread has been used to talk about Rogue in general, and thus this CES IMG PR is relevant.

http://www.imgtec.com/News/Release/index.asp?NewsID=716

"'Rogue’: Imagination’s partners will debut the first PowerVR Series6 products, including TVs, at CES 2013 while Imagination will demonstrate the family’s advanced OpenGL ES 3.0 and OpenCL capabilities."

A surprise to me that the first rogue product implementation (in fact multiple), is not by Apple.

 

Implementation probably; how about device availability in contrast?

 

Novathor L8580 to sample in Q1 2013:

http://www.electronicsweekly.com/Ar...-intros-first-fd-soi-smartphone-processor.htm

http://blog.stericsson.com/blog/201...t-lte-modap-platform-unveiled-by-st-ericsson/

 

I guess that despite using a tech called "eQuad", that's a single-core Cortex A9 at up to 2.5GHz? And a SGX544MP1 at up to 600MHz?

If it's sampling now, who's going to pick these up in some 8 months?