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

Every time you see a chart w/o scale, you should probably assume that you're dealing with a pretty marketing picture.

 

As always sterile numbers and figures are one thing and real time efficiency another chapter. That said in terms of sterile FLOPs (and anyone is free to determine how meaningful or meaningless such a figure can be) at the same frequency there's more than a factor 2x between a G62x0 and a 543MP4; in terms of texel fillrates they'll break even. What then comes into play are any possible architectural advancements for the new generation and the higher frequencies (amongst others) combined.

When it comes to Apple's future Rogue based SoCs and frequencies (which is always a story of its own) one question would be whether they'll use a 4:1 or 3:1 CPU to GPU frequency ratio.

 

Please start putting out architectural disclosures.

 

The bacon butties come with a choice of two sauces?

 

@JH A definite upgrade, then, on the previous IC architecture?

 

There goes the thread ...

 

It's not really a choice - there is only one right answer. Tomato sauce is the way to go!

An IMG thread stayed on topic for almost 6 pages. Pretty good going!

 

Real men use HP.

 

Ahem, we are a unified design you know...

 

Who says IMG don't do great PR

IMG_product_Placement

 

ST going the way of TI it appears

ST to exit ST-Ericsson in 2013

"In the area of application processors ST will abandon the mobile digital sector but continue to make set-top-box chips, cable modems, DisplayPort connection chips, network processors, car infotainment and ASICs for gaming."

"ST said it will continue to support ST-Ericsson with application processors, IP and manufacturing process technology during the disengagement. This will include the fully-depleted silicon-on-insulator (FD-SOI) process which ST has pioneered at 28-nm and with a road-map to 20-nm."

"support" doesn't sound very engaged to me. Adds a lot of doubt to how well the ST-ericsson (which after Q3 next year won't have any "ST" component) progresses with its nova platform.

Unsurprisingly, ST is creating a new "embedded" division, mimicking TI's decision.

http://www.eetimes.com/electronics-news/4403006/ST-to-exit-ST-Ericsson-in-2013

 

And another mobile SoC company bites the dust....

ST already has Cortex-A9 based SoCs they try to sell into automotive etc. They also have a kind of ASIC business, where they configure ARM-based SoCs for specific customers.

So we still have:

Qualcomm (App engine + modem)
Mediatek (App engine + modem)
Apple (in-house, no modem IP)
Nvidia (App engine + modem)
Samsung (App engine + modem)
Renesas (App engine + modem)
Broadcom (App engine + modem)
Intel (App engine + modem)

If I were Apple, I would try to get ST-E's modem IP asap... for the patents and IP. At 14 nm / 20 nm it makes sense to integrate the 3G/LTE modem as well, for cost reasons.

Don't know who else would be interested to buy that part of ST-E. ST-E has Samsung as customer for
feature phones / entry-level smart phones, but who knows for how long.

I think the joint-venture is over and out.

 

At least it finally gets the French government out which is probably a big relief for Ericsson.

 

Believe it or not, but when I visited ST once in Grenoble, people there were joking that they salaries got basically paid for by Sarkozy and they considered themselves civil servants.

Remember that ST-Ericsson started as a JV between NXP and ST, in fact a few months earlier.

You don't want to know how much competencies ST(-Ericsson) already closed and got rid off in its ever on-going effort to cut costs and be competitve.

 

If Apple were to integrate the modem, then they can no longer argue patent exhaustion from buying third-party modems in 3G/LTE patent cases. Although, having their own modem IP strengthens their bargaining position. That might encourage Samsung to buy to prevent Apple from acquiring and leveraging that modem IP. If Microsoft really plans to make first-party smartphones as they have tablets, owning modem IP as a defensive play might also be advisable. I guess AMD really isn't in a good financial position to make a move into smartphone SoC (either ARM or x86) with integrated modem although this would be a good opportunity when the tech/IP is available.

 

Another one bites the dust.

 

Isn't it a bit too early to tell that?

We still don't know how successful the Nova A9600 will be.

 

Pfff hope dies last; it went from delay to delay and I think but am not sure that recent roadmaps don't even mention it anymore.

 

I though ST-Ericsson was being fairly successful with the U8500, seeing how it's featured in all Sony mid-range smartphones (Xperia P, U, Sola, Go) and some Samsung ones (Galaxy S3 Mini, Galaxy Beam, Galaxy Ace 2).

What are all these delays due to?

 

A9600 was supposed to sample in late 2011 which even back then when it was announced sounded like a HUGE joke to me. Frankly I didn't pay too much attention but I think they had some problems finding a foundry (someone correct me if I'm wrong) for which most of us breathed in recently when it had been announced that ST Micro had contracted Samsung for their 32/28nm nodes.

Either way as much as TI is concentrating on embedded stuff as of recently and I don't expect to see an OMAP6 (if ever), the very same pessimism goes also for ST-Ericsson (or should I just say Ericsson?) and NovaThor A9600. What would they need an as ambitious project/SoC exactly for as early if they intend to mostly concentrate on embedded markets?