Nokia's Present & Future

[Exophase]

I don't think you understand what I'm saying. Yes, Flash supports 3D and 2D acceleration, but that doesn't mean that that's all that's necessary to run it. That's basically akin to saying that GPUs should be running PC games without any involvement from the CPU entirely, and that we should all use Atoms because GPUs are better than CPUs at everything. Flash applications can have complex scripting on the level of any other program - if they weren't complex then no one would have ever done JITs for it.

And JIT has nothing to do with video acceleration. You benefit from both, not one or the other; they're complementary technologies that address two different parts of the Flash workload. Actually, in light of your misunderstanding of what Flash is, I'm now doubting that the Broadcom chip really interprets Flash Lite all by itself without CPU involvement.

So I hope you understand why Adobe considers a strong CPU necessary for flash, regardless of how great the GPU is.

 

[roninja]

Don't rule out Renesas supplying into Nokia and I'm not just talking about the baseband business they acquired but getting the apps processors as well.

 

[wco81]

Which factors lead to sales or provides momentum for a platform?

a. HW specs -- CPU/GPU, camera capabilities, etc.
b. Industrial design/build quality
c. Software -- UI and apps.

For a time, it seemed Nokia lived on a) and a bit of b) for some time.

Now it's trying to catch up on a) and c).

But is it easy for any platform to get an advantage in a) any more? SOC vendors are competitive so everyone will get the fastest A8 or A9 silicon at the same time.

Consumers seem to be willing to compromise on b) in favor of a).

So c) seems to be more elusive yet if all the platforms get parity on the high-profile UI features like pinch-zooming (subject to resolution of litigation of course) and each platform gains a critical mass of apps. in various categories, software differentiation becomes less important.

Of course, all these factors assume that smart phone purchasing decisions are always more rational than say purchasing decisions for fashion.

 

[wishiknew]

b) brings so many bad memories of the N97 design/quality. The camera lens 'protector' that scratches the lens. So very loose lock button that falls right off. Unshielded gps antenna thats susceptible to interference from pcb. That gps antenna again which is just a piece of tape and somehow pressing down on it helps with gps signals.

 

[codedivine]

b) brings so many bad memories of the N97 design/quality. The camera lens 'protector' that scratches the lens. So very loose lock button that falls right off. Unshielded gps antenna thats susceptible to interference from pcb. That gps antenna again which is just a piece of tape and somehow pressing down on it helps with gps signals.

On behalf of a friend who owned a N97, I can attest to it being a crappy product and a self-goal for Nokia. I think the N97 did more harm to Nokia than the iPhone ever did

 

[frogblast]

On behalf of a friend who owned a N97, I can attest to it being a crappy product and a self-goal for Nokia. I think the N97 did more harm to Nokia than the iPhone ever did

It didn't help that it was right about the time the iPhone hit the market.

I believe that for most handset buyers, specs are far down the list of things they care about, and I don't agree that Nokia ever really pushed high-spec products. For example, the N95 was the premier handset with a good GPU, and the N97 regressed hugely there (which then went up against the iPhone, its UI fluidity, and the games push). Nokia has always been a volume manufacturer: production cost mattered more to them then delivered specs. Even their recent announcements are specced much lower than the competition, likely to keep production costs in check.

I would like to think that perceived performance matters much more to buyers than spec sheets (as it should), but Nokia isn't producing Symbian products with this quality. This is something Apple has wrapped up tight right now (still significantly nicer than the top-end android devices), and Apple's retail presence is a powerful way to demonstrate perceived performance to buyers.

I haven't used a meego device, but somehow they have to deliver an iPhone-level UI experience, and they have to demonstrate that experience to buyers. Failing on either count will doom them in the smartphone race. I don't have any personal experience with the meego technology yet. On the retail side, Nokia has no way to achieve the necessary exposure in the US, and so I cannot imagine them gaining any ground here even if the meego products are fantastic. I don't know what the Nokia retail situation looks like in Europe (can someone fill me in here, on a scale of 'Apple Retail' to 'T-Mobile Franchise Store' to 'Best Buy'?).

 

[codedivine]

*Removed*

 

[Deleted member 13524]

Has anyone ever seen an explanation from someone at Nokia for removing the GPUs when going for touch devices?

Battery life? Price reduction?
They thought UI experience, gaming and video wasn't important at all? Just a massive ammount of flash memory to listen to 10 000 hours of mp3 songs?

The N95 was excellent, we can still find some pretty good looking 3D games for it, why downgrade everything in their following generation?

 

[Arun]

Has anyone ever seen an explanation from someone at Nokia for removing the GPUs when going for touch devices?

Platform reuse - they wanted to keep using the same TI chips for the entire line-up (I know this isn't used in the N97 but as I said the 600MHz+ ARM11 chip was one of the the world's first 45nm handheld products and quite cost-effective) and they screwed up massively when they decided not to include any GPU in them. The world might be different if they had decided to integrate a MBX Lite like Apple did with the original iPhone. Or if they had decided to use an OMAP3 in the high-end rather than stupidly assume they only needed that on the ultra-high-end Maemo.

 

[Deleted member 13524]

Platform reuse - they wanted to keep using the same TI chips for the entire line-up (I know this isn't used in the N97 but as I said the 600MHz+ ARM11 chip was one of the the world's first 45nm handheld products and quite cost-effective) and they screwed up massively when they decided not to include any GPU in them. The world might be different if they had decided to integrate a MBX Lite like Apple did with the original iPhone. Or if they had decided to use an OMAP3 in the high-end rather than stupidly assume they only needed that on the ultra-high-end Maemo.

Well, I understand that they wanted to ramp up clock speeds of the ARM11 for the touch models (Q4 2008), but even if there were only low clocked OMAP2 and OMAP3 was still half a year away, they had already dozens of MBX-equipped ARM11 solutions at +400MHz. Even the Adreno 200 was available at the time with the MSM7225 in the Touch 3G shipping approx. at the same time as the 5800XM.

Maybe they were too afraid of the public backlash they'd have if they lowered the standby longevity?

Maybe it was purely a question of mass availability? They couldn't find a single supplier that could provide the enormous ammount of GPU-equiped SOCs they would eventually need for their S60v5 products?
They were, in fact, selling ~250 000 ARM11 smartphones a day during 2009 -> mid-2010.
Even if only a fifth of that was touch-based devices, what chipset manufacturer could withstand that kind of supply of high-end ARM chipsets?
Look at all the other smartphone manufacturers. Combined, they sell less phones than nokia and even Samsung, which makes its own ARM chipsets, "has" to buy CPUs from other vendors for some of their phones.

If what they wanted was an unified architecture for all touch-based devices, they'd probably need a deal for fab exclusivity for high-end ARM11 chipsets, and that could cost tons of money and little to no insurance for availability.


In the end, the N95 itself was kind of an exception (along with N82 and N93, I think). Every other S60v3 device continued to ship with regular ARM11 CPUs. So the N900 is actually the true successor to N95, and not N97 as the branding implies.

Just some more thoughts. I'd really like someone from Nokia to step up and give us some explanation, though.

 

[Arun]

Oh, I think you're misunderstanding. Nokia uses *custom* TI SoCs for Symbian - not the standard OMAPs you'll find on TI's website. Nokia is solely responsible for the decision of not integrating a GPU in these custom SoCs. The decision for the 45nm TI SoC used in 2009/2010 products was nearly certainly made before anyone had ever heard of the iPhone!

My assumption is that these contracts include a legal obligation for Nokia to buy these SoCs in very high quantities or pay a penalty - so I'm very confident Nokia couldn't have decided to use only chips with 3D GPUs even if they wanted to. Unless, of course, they were standard parts from TI - I'm sure they would have been very happy to renegotiate the contract if it was to their advantage.

This still does not excuse reusing the same chips for their ultra-high-end - they certainly could have used a different platform for that. Besides being greedy/lazy/stupid, I can think of only one real excuse: they did not have any better platform with existing drivers for Symbian and thought it was better to release a subpar product than add a 6+ months gap in their line-up. A very flawed decision in retrospect certainly, but still slightly understandable.

 

[Deleted member 13524]

Oh, I think you're misunderstanding. Nokia uses *custom* TI SoCs for Symbian - not the standard OMAPs you'll find on TI's website. Nokia is solely responsible for the decision of not integrating a GPU in these custom SoCs. The decision for the 45nm TI SoC used in 2009/2010 products was nearly certainly made before anyone had ever heard of the iPhone!

Yes, but what if the negotiation went something like this:
- Without the GPU, the die size is X, so we can fit Y in a waffer, which means we can provide Z chips per month
- With the GPU, the die size is 2*X, so we can fit Y/2 in a waffer so we can only provide Z/2 chips per month.

If Nokia needed an absolute minimum of 2*(Z/3) chips per month, they could never go with the second option.


Just a very basic example.

 

[Arun]

Heh, a MBX Lite is so utterly negligible in terms of die size that I don't think it's worth even calculating that stuff. Seriously it's what, 50x smaller than the 10.2Mbps 3G baseband part alone? Certainly a SGX530 would have been much bigger, but I don't think anyone's considering that throughout the line-up here.

I'm sure Nokia considered adding a MBX/MBX Lite to that SoC when they had to make the decision many years ago. And I'm sure the people responsible felt very stupid after the iPhone came out. But that's excusable - what isn't so excusable is not having a higher-end backup plan including 3D graphics for 2008/2009. There are mitigating circumstances as I explained, but they're not good enough. So stupidity and/or arrogance should be seriously considered.

 

[Tchock]

how about battery life?
afaik none of the first cortex-based solutions were frugal in terms of power consumption...

2009 was forgivable for them given that it was too late to tailor a holistic solution that suited the nokia "audience" anyway; what's really damnable however are their platforms for 2010- late and slow.

tangent: was iOS 1 really that far away from the other RTOS based code of yore? It had fixed functionality and optimization really could be case/usage based, whilst symbian in general offered less of such luxuries. It really took apple till a8 to shed most performance (outside of ux) concerns, and android's still struggling with simple things like sms databases! A gpu can mask, but still can't resolve underlying performance issues once they become that apparent

 

[Deleted member 13524]

Heh, a MBX Lite is so utterly negligible in terms of die size that I don't think it's worth even calculating that stuff. Seriously it's what, 50x smaller than the 10.2Mbps 3G baseband part alone? Certainly a SGX530 would have been much bigger, but I don't think anyone's considering that throughout the line-up here.

Ok, you got me there, I have no idea about the proportions between the parts.
The only referrence I had are the proportions in x86 solutions.
Seeing how the 80 shader igp in Zacate takes about 60% of the total die size of the APU, I figured an integrated OpenGL ES 1.1 GPU would take a substantial space from an ARM11 soc.

I'm sure Nokia considered adding a MBX/MBX Lite to that SoC when they had to make the decision many years ago. And I'm sure the people responsible felt very stupid after the iPhone came out. But that's excusable - what isn't so excusable is not having a higher-end backup plan including 3D graphics for 2008/2009. There are mitigating circumstances as I explained, but they're not good enough. So stupidity and/or arrogance should be seriously considered.

I remember reading somewhere that before all the management changes during 2010, people at nokia complained that the company based all its decisions in the hardware division. The hardware team didn't answer to demands from the software team. The hardware team simply said "the hardware will be this" and the software team would have to find a way to code all the content based on the imposed hardware.
Maybe it was simply the hardware team telling the top management "Hey ma, look how cheap I got this!" and the software team had nothing to say about it, so they got screwed.

 

[Ailuros]

Maybe it was simply the hardware team telling the top management "Hey ma, look how cheap I got this!" and the software team had nothing to say about it, so they got screwed.

I don't think anyone but true insiders could know if that's true or not, but however absurd it may sound to some it doesn't surprise me in the least considering other nonsense I occasionally hear left and right.

I sure hope things will change at NOKIA; if not it'll be there loss. Besides I can see other vendors currently selling extremely well in the smart-phone arena possibly repeating the very same mistake. If true then Nokia isn't alone.

The only vendor out there that seems to stay focused (despite the criticism they receive here and there, some justified some not) seems to be Apple for the time being. And I'm only mentioning it here because you made a very legitimate point about the importance of software in any SoC. For a balanced SoC a fine balance between hardware and software is an absolute necessity. Wherever either/or lacklusters things can get tricky.

 

[Entropy]

Which factors lead to sales or provides momentum for a platform?

a. HW specs -- CPU/GPU, camera capabilities, etc.
b. Industrial design/build quality
c. Software -- UI and apps.

For a time, it seemed Nokia lived on a) and a bit of b) for some time.

Now it's trying to catch up on a) and c).

But is it easy for any platform to get an advantage in a) any more? SOC vendors are competitive so everyone will get the fastest A8 or A9 silicon at the same time.

Consumers seem to be willing to compromise on b) in favor of a).

Actually, I'd turn that around and say that consumers seem to be willing to compromise on a) in favor of b). Of course there are categories of people for either direction, but I'd argue that the momentum is in the a) -> b) direction.
This goes both for phones, and computing devices sold to private individuals in general. The reason being twofold - one is that as people gain experience with a type of device they get a clearer picture of their usage patterns, and can make selections based on their experiences. (Screen and keyboard gaining in priority relative to the processor on a laptop for instance.) The other being that once a category is established, so is a technological baseline, and all it takes is a lithographic process node or two and the silicon side of the device is largely irrelevant as a differentiator. Smartphones are well on their way of being feature defined, a few additions to be made, but then? Their customers are even less tech oriented than the PC market, and technology has been weakening as a sales driver for PCs for many years now. I can't see any powerful mechanisms pushing momentum from b) -> a).

It's hilarious to see tech sites originating in the heydays of PC overclocking reporting on and reviewing the new mobile devices pretty much as if they were really small desktop PCs.

 

[Arun]

Smartphones are well on their way of being feature defined, a few additions to be made, but then? Their customers are even less tech oriented than the PC market, and technology has been weakening as a sales driver for PCs for many years now. I can't see any powerful mechanisms pushing momentum from b) -> a).

Ah, well I'm sure Apple will be happy to know they're wasting money investing in a high-end SoC division. And I'm sure they would be even more happy if every other OEM agreed with you so that they could beat them ridiculously on HW specs without investing too much in them. A 50mm² application processor on 40nm costs about $15 - and in Apple's case it probably only costs about $8 given that they don't have to pay for anyone else's gross margins. Do you really expect them to try saving $5 per phone on a $500 device after investing this much in their SoC division? Or do you expect everyone else not to try competing with them on SoC specs despite it remaining only a small part of the BoM?

I think a much better way to think about this is to realise that as smartphones take over the world, the classical ULC/Basic/Feature/Smart market delimitations will no longer make sense. By 20nm we'll have good 3G "smartphones" with fast web browsing at the prices of EDGE basic phones in developing countries today. As feature phones die a slow death and ULCs become more capable, the way I like to think about it is to divide the market in four new parts: voicephones, webphones, multimedia webphones, and superphones. I think you're expecting webphones to massively outsell multimedia webphones (which I'd define as anything with 1080p recording) and superphones (much higher CPU & 3D performance).

My expectation is that 2015 volumes will look roughly like this: 10% Superphone/20% Multimedia Webphones/35% Webphones/35% Voicephones. But that's worldwide. In the USA and Western Europe, I expect the numbers to be closer to 30% Superphone/30% Multimedia Webphones/20% Webphones/20% Voicephones. There are very subjective estimates of course, but personally I think you'd be crazy to expect superphones to shrink into irrelevance. They're certainly not the only exciting part of the market though.

 

[Entropy]

Ah, well I'm sure Apple will be happy to know they're wasting money investing in a high-end SoC division. And I'm sure they would be even more happy if every other OEM agreed with you so that they could beat them ridiculously on HW specs without investing too much in them. A 50mm² application processor on 40nm costs about $15 - and in Apple's case it probably only costs about $8 given that they don't have to pay for anyone else's gross margins. Do you really expect them to try saving $5 per phone on a $500 device after investing this much in their SoC division? Or do you expect everyone else not to try competing with them on SoC specs despite it remaining only a small part of the BoM?

It seems you partially misunderstand (on purpose?) or maybe I was unclear.
Of course there will be advantages to efficient integration - I wouldn't underestimate those $5 saved on 100 million devices. You know as well as I do that those $500 are not profit. Another $5 on their margin per device from every iPod/iPhone/iPad is significant. Additionally, there will be advantages in terms of power control = battery life = customer satisfaction. AND they keep control and give themselves options for differentiation in the future. Of course it makes sense for a player of Apples size and ambitions to invest in their own SoC know-how.

So - I gave the mechanisms I see for why hardware specs will diminish in market importance.
Where are your arguments for why they would increase?

Btw, there are other reasons I didn't bring up for assuming that additional hardware capabilities won't get increasingly important in the marketplace. Most important that there are no applications that require much more than we already have, and we know that, because there are no such applications in widespread use even on PCs! (There are low volume exceptions, sure, but even of those, few would transfer to a 4 inch diagonal screen.)

As happens so often on technology enthusiast sites I'm left with the question WHY?
WHY should 30% of USA phones in 2015 be devices that are beyond multimedia capable webphones?
(Hell, in this case I don't even understand what such phones are supposed to do.)
What would sell them? Why would the general public give a damn, much less fork over real money? If you give them an octa A15 processor with SGX565 parallelized to the gills, what do you expect them to do with it except bitch about the battery life?
I'm serious. These are the real questions. Not "what is possible".

PS. I can see only one trajectory I believe in (indeed, I even hope for it to happen) that would justify a bit more than a consolidation of current capabilities and some additional output, and this is where the phones replace the PC completely for 99% of the population. Even this would require little more in the way of hardware than output to high resolution screens. Mini displayport can already serve this purpose, but a better connector for phone -> peripherals might be Light Peak. One connector daisy chaining to devices, and input via bluetooth that is already available. But even in this scenario where the phone is the pc, we still have to recognize that even in the PC market we can't sell high performance PCs to the general public and that the general public quite sensibly favors other values.
And unfortunately I believe we are a long way from replacing PCs completely with our phones, for software reasons, not hardware.

 

[Deleted member 13524]

The only vendor out there that seems to stay focused (despite the criticism they receive here and there, some justified some not) seems to be Apple for the time being. And I'm only mentioning it here because you made a very legitimate point about the importance of software in any SoC. For a balanced SoC a fine balance between hardware and software is an absolute necessity. Wherever either/or lacklusters things can get tricky.

About keeping focused for the last 2-3 years, that's right.
About being focused right now, I'd throw in Windows Phone 7 and Symbian^3.

WP7 has some very tight hardware requirements (only 1st Snapdragons are allowed, not even the models with Adreno 205 are permited!!) and the UI is very fast and responsive, with its biggest problem being so dumbed down it almost resembles a feature phone.
Microsoft claims it will add functionality with firmware updates (Apple-style) but they set the starting bar way too low, IMHO.


Symbian^3 models have a very decent set of hardware, which has now become unified across all models.
Thanks to being the only Symbian^3 vendor, Nokia stated that there won't be firmware upgrades to model X or Y anymore, there will only be upgrades to Symbian^3 (which, by the time all the S60v5 models are dead and buried, will be called just "Symbian").
It'll take the same unified upgrade system from Apple but now with the advantage of offering various models across a wide set of price ranges and tastes (smaller:C6, good camera:N8, keyboard:E7, etc), which is something that Apple doesn't offer.

The only problem is that Nokia's taking the complete opposite approach from WP7. They took an OS with a fantastic set of functionality (IMO, the closest there is to a desktop OS, next to WM6.x) and they'll be gradually changing the UI to fit the consumer's current tastes..






Now going a bit more ontopic:
http://www.inderes.fi/news/22/15/No...joonaa-N8-a-vuoden-viimeisellae-kvartaalilla/

http://www.gsmarena.com/nokia_n8_sales_nearing_4_million_units-news-2180.php

Nokia N8 going near 4 million units sold, which is more than 1.5 million per month. Not bad of a start, at all.
I wonder if the C6-01 + C7 numbers are equally substantial.