Thursday, 21 June 2012
An ecosystem for microsoft
I think you need to ask what it means for consumers to have freedom. There are necessary freedoms - the ability to choose from a series of applications, the ability to choose where and when you get your computer out and use it - and then there are frivolous or unnecessary freedoms - the ability to hack the registry, the ability to download illegally pirated material etc, the ability to put a port of another OS onto your hardware. Most of these latter so-called "freedoms" are the domain of people like you and me, who have the skill to perform these changes. We would buy a product that allows us to do this if we are hell-bent on doing so, although you'd have to question the necessity to have this capability - I think a lot of people do this just because they can, rather than because they have to.
The vast majority of consumers - people who don't read tech sites in detail - want a product that works, that gives them ecosystem components they can understand and value, and that sews all of this up in a cohesive package that can pervade their lives. These ecosystems within themselves provide competition within the important things - there will be a dozen apps that do the same thing in the app store, so pick the one you like, for example. There is also competition around which ecosystem a consumer should adopt. Once ecosystem is pitted against ecosystem, they then need to innovate and compete with each other on what makes an ecosystem great, rather than just what makes the OS, or the hardware, great.
Microsoft can see that having an ecosystem is what makes Apple the largest company on Earth. But it isn't just ecosystem. Apple have made friends with their customers, by having exclusive retail and advertising environments that avoid mixing their products up with those of others, and that clearly communicate the benefits, as opposed to just the implementation: e.g. a tablet isn't just 9mm thin machined aluminium and glass; it's thin and light so you can carry it with you and hold it in your hand.
I'm impressed that this part of Microsoft is moving with the times. No one is saying that ecosystem is the way of the future forever, but in this wave of computing ecosystem is king, and to compete that's where you need to be. Microsoft are going there, and if it's at the expense of a few broken noses in Taiwan, I'm not sure they really care.
Saturday, 4 December 2010
Can Intel keep up in mobile?
So Intel bought the wireless baseband unit of Infineon, which may go some way to explaining why Nokia sold their baseband business to Renesas. I guess Intel could have acquired Nokia's baseband business just as easily as Renesas, but Intel already has relations with Nokia on various fronts, and the one thing that Nokia's baseband didn't provide is relationships inside the hottest tech company on the planet - Apple. Additionally, if Nokia went into the relationship with Intel hoping it would sell it 3G technology as its part of the bargain, then they must have been sorely disappointed when Infineon went and put its baseband up for sale; Intel may have given Nokia the shove out of lust to go after Apple. Infineon's biggest customer for baseband is Apple, and the purchase now gives Intel a foot in the door of Apple's iPhone laboratories.
What would happen, though, if Apple chose to switch to Qualcomm, the undisputed heavyweight of baseband? The answer is quite clear - it would be another blow to Intel's stillborn mobile strategy. Sure, they can sell baseband now alongside the Atom processor, but if their atom mobile strategy falls short of expectations, then alongside the Mcafee purchase, it makes Intel look more like IBM - punching as hard as it can at multiple targets, trying to leave a bruise.
And of their mobile hardware and software? Atom may be going through a lengthy refurbishment inside their offices - see this LinkedIn profile - but will it be good enough? Intel are trying hard to make a dent with Meego but even that won't be a safe haven from ARM since Linaro appeared on the scene.
The first problem Intel may have is they may be trying to aim too low with their next atom, while ARM is very aggressively and deliberately moving up into Intel's coveted server space. ARM already have products that satisfy the niche the current atom is trying to enter better than atom ever can. Can Intel really tune their performance/power to the levels that ARM can, talking as they do to so many partners who use their products every day?
The real clincher is that Intel don't have the upper hand at a time when every ARM partner is going multi-core. 2011 will be the year of the multi-core smartphone, at a time when Intel can't even field a single single-core CPU into a single smartphone. Even if their next generation atom is multicore and manufactured on expensive 22nm silicon at the end of 2011, Intel will have to lose a significant amount of money that they won't be making if they were fabbing SSDs instead, by selling this atom at a knock down price to win customers.
Software may appear to be quite rosy on the surface of the intel architecture. They have Windows, Meego, a port of Android and various other flavours of Linux. Compare this to ARM, who have every mobile OS - some of which are private to particular companies, and will never be ported to Intel - along with Palm WebOS inside HP, the latest leading edge of Android, and Windows Phone 7. The one thing ARM lacks is "big" Windows. Surely a company of Microsoft's stature can port an OS to the ARM architecture? There is absolutely nothing technical stopping them from doing so. Windows is not like an open source OS, dogged by fragmentation. The source trees - for MS's sake - should be well managed, and there be enough expertise within the company to allow for a port to not only take place, but also for a fresh start to be made on a new platform if it so chooses, dumping all legacy code along the way that makes Intel's compatibility with the past not an advantage, but an albatross around their necks. It gives MS the chance to clean up a mature OS and make a clean break on a new architecture. If MS make this port happen - and there's every reason to believe they should be, then I'd be worried if I were Intel. The Nufront 2GHz ARM Cortex-A9 based laptop being shown in that link will be quite favourable up against a common everyday laptop.
Watching the wagons circling around Intel's server business must be equally worrying. Although it may appear that history is not on ARM's side, this may be a misconception and the opposite may be true. It is certain that intel may have a performance edge, but at what price?
Their competition in the server space with AMD is an old show that's been travelling through town for years, and people have seen it so many times that they are getting bored of it. The real disruptive innovation in the server room will be when they can take out the cooling equipment, and get just the right amount of performance to get the job done. Can intel and AMD duck low enough - quick enough - to avoid the punch coming at them from below?
So many companies are now entering the server market with experience of low power design that the inevitable step of evolution will be that the disruption will occur. But it won't just occur in power - it will also occur on price. Companies that rarely charge over $50 for a chip will now be glad to be selling chips into a new market at 4 times their traditional going rate, but still undercutting Intel by as much as 5 times. Imagine the impact on the bottom with every server sale lost to one of these new competitors? How long will it take before the investor community starts worrying about the effect this degradation in price will have on their investments?
People often say that the one thing that matters little in the equation of running a server farm is the cost of the CPU, in comparison to the cost of the electricity. That's all well and good from the server makers perspective, but that argument doesn't help Intel, who wants to maintain their high value margins. If someone can sell a product that Intel can't match on power that removes the largest expense in the data center, then where else can that leave Intel? The erosive effect on Intel of a lost server sale is much worse than the accretive effect in the balance sheets of their competitors.
Their purchase of Mcafee looks more and more like Intel is aligning itself with a new world where they need to diversify out of the microprocessor business. There doesn't seem to be much of a future in software based security for them either, though, if you look at a technology that is hiding secretly inside every single one of those high end smart phones - and the introduction of another technology demo by non other than the CEO of Google.
TrustZone is a security feature built into the hardware of every ARM CPU in every smartphone. It creates an unbreakable layer at the system level in the silicon where there are distinct secure and non-secure zones for peripherals, memory, even the architectural state inside the CPU. There is secure software that runs only in secure state, and manages accesses to the secure world. This is a little known technology outside of semiconductor circles, but one that ARM will surely be looking to tap as more payments are made through our mobile devices.
The volume of payments to be made through mobile devices are likely to skyrocket, thanks in no small part to Google's Android 2.3/3.0/gingerbread. In an interview, Google CEO Eric Schmidt announced that Gingerbread would have Near Field Communication (NFC) capabilities, allowing it to read and write information back and forth between terminals and other devices through the air.
The encryption and NFC blocks in the silicon that manage these transactions - and the keys that are used for encryption as well - will all be secured under ARM's TrustZone system security. Even the screen will be secure and the part of the OS that listens to the screen taps, e.g. turning them into numbers for a PIN - cannot be spied by malicious software, because only trusted software is inserted and can operate in the secure zone from the point of manufacture onwards.
Intel's Mcafee brand is a hackneyed dinosaur of yesteryear, where such refined security technology never existed, necessitating a brute force fix in software. When security is built into the silicon, this becomes unnecessary and it is up to the silicon provider to secure the system, not the OS.
What would happen, though, if Apple chose to switch to Qualcomm, the undisputed heavyweight of baseband? The answer is quite clear - it would be another blow to Intel's stillborn mobile strategy. Sure, they can sell baseband now alongside the Atom processor, but if their atom mobile strategy falls short of expectations, then alongside the Mcafee purchase, it makes Intel look more like IBM - punching as hard as it can at multiple targets, trying to leave a bruise.
And of their mobile hardware and software? Atom may be going through a lengthy refurbishment inside their offices - see this LinkedIn profile - but will it be good enough? Intel are trying hard to make a dent with Meego but even that won't be a safe haven from ARM since Linaro appeared on the scene.
The first problem Intel may have is they may be trying to aim too low with their next atom, while ARM is very aggressively and deliberately moving up into Intel's coveted server space. ARM already have products that satisfy the niche the current atom is trying to enter better than atom ever can. Can Intel really tune their performance/power to the levels that ARM can, talking as they do to so many partners who use their products every day?
The real clincher is that Intel don't have the upper hand at a time when every ARM partner is going multi-core. 2011 will be the year of the multi-core smartphone, at a time when Intel can't even field a single single-core CPU into a single smartphone. Even if their next generation atom is multicore and manufactured on expensive 22nm silicon at the end of 2011, Intel will have to lose a significant amount of money that they won't be making if they were fabbing SSDs instead, by selling this atom at a knock down price to win customers.
Software may appear to be quite rosy on the surface of the intel architecture. They have Windows, Meego, a port of Android and various other flavours of Linux. Compare this to ARM, who have every mobile OS - some of which are private to particular companies, and will never be ported to Intel - along with Palm WebOS inside HP, the latest leading edge of Android, and Windows Phone 7. The one thing ARM lacks is "big" Windows. Surely a company of Microsoft's stature can port an OS to the ARM architecture? There is absolutely nothing technical stopping them from doing so. Windows is not like an open source OS, dogged by fragmentation. The source trees - for MS's sake - should be well managed, and there be enough expertise within the company to allow for a port to not only take place, but also for a fresh start to be made on a new platform if it so chooses, dumping all legacy code along the way that makes Intel's compatibility with the past not an advantage, but an albatross around their necks. It gives MS the chance to clean up a mature OS and make a clean break on a new architecture. If MS make this port happen - and there's every reason to believe they should be, then I'd be worried if I were Intel. The Nufront 2GHz ARM Cortex-A9 based laptop being shown in that link will be quite favourable up against a common everyday laptop.
Watching the wagons circling around Intel's server business must be equally worrying. Although it may appear that history is not on ARM's side, this may be a misconception and the opposite may be true. It is certain that intel may have a performance edge, but at what price?
Their competition in the server space with AMD is an old show that's been travelling through town for years, and people have seen it so many times that they are getting bored of it. The real disruptive innovation in the server room will be when they can take out the cooling equipment, and get just the right amount of performance to get the job done. Can intel and AMD duck low enough - quick enough - to avoid the punch coming at them from below?
So many companies are now entering the server market with experience of low power design that the inevitable step of evolution will be that the disruption will occur. But it won't just occur in power - it will also occur on price. Companies that rarely charge over $50 for a chip will now be glad to be selling chips into a new market at 4 times their traditional going rate, but still undercutting Intel by as much as 5 times. Imagine the impact on the bottom with every server sale lost to one of these new competitors? How long will it take before the investor community starts worrying about the effect this degradation in price will have on their investments?
People often say that the one thing that matters little in the equation of running a server farm is the cost of the CPU, in comparison to the cost of the electricity. That's all well and good from the server makers perspective, but that argument doesn't help Intel, who wants to maintain their high value margins. If someone can sell a product that Intel can't match on power that removes the largest expense in the data center, then where else can that leave Intel? The erosive effect on Intel of a lost server sale is much worse than the accretive effect in the balance sheets of their competitors.
Their purchase of Mcafee looks more and more like Intel is aligning itself with a new world where they need to diversify out of the microprocessor business. There doesn't seem to be much of a future in software based security for them either, though, if you look at a technology that is hiding secretly inside every single one of those high end smart phones - and the introduction of another technology demo by non other than the CEO of Google.
TrustZone is a security feature built into the hardware of every ARM CPU in every smartphone. It creates an unbreakable layer at the system level in the silicon where there are distinct secure and non-secure zones for peripherals, memory, even the architectural state inside the CPU. There is secure software that runs only in secure state, and manages accesses to the secure world. This is a little known technology outside of semiconductor circles, but one that ARM will surely be looking to tap as more payments are made through our mobile devices.
The volume of payments to be made through mobile devices are likely to skyrocket, thanks in no small part to Google's Android 2.3/3.0/gingerbread. In an interview, Google CEO Eric Schmidt announced that Gingerbread would have Near Field Communication (NFC) capabilities, allowing it to read and write information back and forth between terminals and other devices through the air.
The encryption and NFC blocks in the silicon that manage these transactions - and the keys that are used for encryption as well - will all be secured under ARM's TrustZone system security. Even the screen will be secure and the part of the OS that listens to the screen taps, e.g. turning them into numbers for a PIN - cannot be spied by malicious software, because only trusted software is inserted and can operate in the secure zone from the point of manufacture onwards.
Intel's Mcafee brand is a hackneyed dinosaur of yesteryear, where such refined security technology never existed, necessitating a brute force fix in software. When security is built into the silicon, this becomes unnecessary and it is up to the silicon provider to secure the system, not the OS.
Thursday, 8 July 2010
Nokia/Intel flip-flop
In June of last year I wrote a piece about the deal between Intel and Nokia. At the time, I wrote that the announcement was probably more about Netbooks than smartphones, and that the real winner might have been Intel because they could obtain 3G chipset IP on license from Nokia.
Times have moved on, and there was a Nokia netbook announced, and there was no smartphone.
The main thing to come out of the deal was the merging of two failed linux distributions: Nokia's Maemo and Intel's Moblin have merged to make Meego. I don't know what the name means, but presumably the inclusion of the word "go" as part of the name is some way to suggest it as a portable OS.
However, I then noticed the other day that Nokia sold it's entire 3G chipset business to Renesas in a shock move whose implications are still not clear. In the pleasantries of the press release, Nokia and Renesas will be collaborating on future HSPA+/LTE chipsets (a continuation of a past collaboration), but all of the employees in that division will transfer entirely under Renesas.
So this is a firesale of an asset that Nokia on the one hand can do without (they can source 3G chipsets competitively on the open market), but on the other hand they can make obvious symbiotic use of in their handset business.
Is Nokia's market share dipping below the point where it makes sense to have 3G in-house? Is it not possible any more to amortize the cost of that R&D across other parts of the business and still make a profit?
Back onto the original subject of Nokia and Intel, what does this mean for Intel's sourcing of 3G components now? The original deal was to include Intel putting a respirator around the nose and lips of Nokia's 3G business - has this fallen through?
If Intel's traction in smartphones keeps looking as limp as it currently does, there's little demand for this notional part of the deal, and Intel may be forced into sourcing parts from the open market, instead of having a two way love fest with the Finnish giant.
Times have moved on, and there was a Nokia netbook announced, and there was no smartphone.
The main thing to come out of the deal was the merging of two failed linux distributions: Nokia's Maemo and Intel's Moblin have merged to make Meego. I don't know what the name means, but presumably the inclusion of the word "go" as part of the name is some way to suggest it as a portable OS.
However, I then noticed the other day that Nokia sold it's entire 3G chipset business to Renesas in a shock move whose implications are still not clear. In the pleasantries of the press release, Nokia and Renesas will be collaborating on future HSPA+/LTE chipsets (a continuation of a past collaboration), but all of the employees in that division will transfer entirely under Renesas.
So this is a firesale of an asset that Nokia on the one hand can do without (they can source 3G chipsets competitively on the open market), but on the other hand they can make obvious symbiotic use of in their handset business.
Is Nokia's market share dipping below the point where it makes sense to have 3G in-house? Is it not possible any more to amortize the cost of that R&D across other parts of the business and still make a profit?
Back onto the original subject of Nokia and Intel, what does this mean for Intel's sourcing of 3G components now? The original deal was to include Intel putting a respirator around the nose and lips of Nokia's 3G business - has this fallen through?
If Intel's traction in smartphones keeps looking as limp as it currently does, there's little demand for this notional part of the deal, and Intel may be forced into sourcing parts from the open market, instead of having a two way love fest with the Finnish giant.
Wednesday, 5 May 2010
Intel Atom announcement - knee Jerk reaction?
Is Intel striving for relevance in the mobile computing age?
I don't know the answer to this question, but a number of things have happened in the last week that indicate a possible motive for Intel to re-iterate information that was already out in the public domain.
First of all, HP have announced their intention to buy Palm. This is at a time when HP were showing their "Slate" device, which was based around Atom and Windows 7. Now that HP have made a strategic investment in Palm and brought a fully formed mobile platform into their organisation, it was only going to follow that they cancelled their Slate initiative - although this is just rumour at the moment, but probably makes sense.
Secondly, LG have announced that they have no intention of releasing the GW990 Moorestown based smartphone that was announced at CES.
Is Intel's recent re-iteration of the existence of Moorestown a ploy to counter the above news, and refocus attention back onto their intentions towards the ultra-portability market?
I don't know the answer to this question, but a number of things have happened in the last week that indicate a possible motive for Intel to re-iterate information that was already out in the public domain.
First of all, HP have announced their intention to buy Palm. This is at a time when HP were showing their "Slate" device, which was based around Atom and Windows 7. Now that HP have made a strategic investment in Palm and brought a fully formed mobile platform into their organisation, it was only going to follow that they cancelled their Slate initiative - although this is just rumour at the moment, but probably makes sense.
Secondly, LG have announced that they have no intention of releasing the GW990 Moorestown based smartphone that was announced at CES.
Is Intel's recent re-iteration of the existence of Moorestown a ploy to counter the above news, and refocus attention back onto their intentions towards the ultra-portability market?
Intel Moorestown - the pied piper of Hamlet? Doubtful...
Intel's claim is that they will have X86 developers following them in droves to provide their software on mobile devices. I don't buy Intel's arguement about developers following them onto smartphones purely because of experience on the desktop. The arguement is several-fold.
Firstly, Intel don't talk about Windows for the smartphone. Yet the entire back catalogue of software they'd like to associate with their mobile platform almost universally comes from the Windows stable. The one thing that Intel can do that ARM partners cannot do is test the viability of Windows 7 on a smartphone form factor, since Moorestown is an X86 platform so in theory should be able to boot this OS. My opinion is that having done this, they have found it to be a poor user experience, because they have gone down the route of developing their own Linux distribution to fit into the space that Windows is not being trumpeted to fill. Have you ever heard Intel saying their smartphones will run Windows? No, and my opinion is that you won't hear that any time soon.
Secondly, Meego is a merge of Nokia's Maemo, and Intel's Moblin. I have a Maemo device, and there are not that many applications available for it. Nor are there many for Moblin. Meego has only just been released, and it will suffer for a long time without applications - as its parent OSes have done for a long time (especially maemo) - until devices exist in the market place to encourage developers to take part. The beauty of the iPad is the legacy of iPhone applications, and the fact that Apple was partnering with new developers on the larger iPad form factor before its release to take advantage of it in new ways - leveraging a portable legacy unrivalled in the industry. By contrast, Intel's OS offering in this space is starting from a poor base and trying to compete with the iPhone OS and Android, both of which have traction and mindshare in the marketplace. [And why does Android have a growing, and decent, applications base? Because there are devices out in the market that are now receiving critical acclaim...]
So, Intel are effectively having to start from scratch on the software front in the mobile space, and in my opinion they are trying to soften the blow of having to go down this route by referring to their X86 desktop legacy as a potential solution to a software gap that is as yet unplugged, even if that legacy doesn't exist.
The third leg on this stool is the issue of platform ownership. At present, a mobile phone manufacturer will decide what access developers have to their mobile platforms. The two most popular platforms with application marketplaces/stores are the iPhone and Android, with WebOS deserving a mention because of its recent acquisition by HP through its parent company Palm.
If the platform is an openly adopted one like Android for example, there is still an element of control that a manufacturer has to add or subtract the Android marketplace from the phone. If the platform is a closed one like the iPhone OS, the applications store is closed and scrutinised by the company, in order to control the user experience (although they are an exception here).
When you come to Linux, the story gets quite murky. There is no Linux applications store, but if there were it would have to support multiple platforms and distributions. The problem has to be narrowed down to just Meego from Intel's perspective, where the intention is for it to span ARM and Intel platforms in order to build - at present - Nokia-only devices. So far, I know of nobody who has taken Nokia's old open source Maemo OS and used it to make a device of their own except for Nokia, so questions abound regarding the scope of Intel's Meego OS effort in the and its ability to penetrate the marketplace.
The question on this third leg really is whether or not Intel will enforce the inclusion of their applications store into any device shipping with Meego? And also, will anyone really adopt Meego, given the histories of Maemo and Moblin?
The fact that Intel cannot draw upon any of their legacy desktop software for the mobile space - because Mobile is not based around windows - is being hidden behind a smokescreen of talk about the legacy of development that Intel has in the desktop world with Windows. If you waft the smoke away, I don't think that Intel has a strong story here.
Firstly, Intel don't talk about Windows for the smartphone. Yet the entire back catalogue of software they'd like to associate with their mobile platform almost universally comes from the Windows stable. The one thing that Intel can do that ARM partners cannot do is test the viability of Windows 7 on a smartphone form factor, since Moorestown is an X86 platform so in theory should be able to boot this OS. My opinion is that having done this, they have found it to be a poor user experience, because they have gone down the route of developing their own Linux distribution to fit into the space that Windows is not being trumpeted to fill. Have you ever heard Intel saying their smartphones will run Windows? No, and my opinion is that you won't hear that any time soon.
Secondly, Meego is a merge of Nokia's Maemo, and Intel's Moblin. I have a Maemo device, and there are not that many applications available for it. Nor are there many for Moblin. Meego has only just been released, and it will suffer for a long time without applications - as its parent OSes have done for a long time (especially maemo) - until devices exist in the market place to encourage developers to take part. The beauty of the iPad is the legacy of iPhone applications, and the fact that Apple was partnering with new developers on the larger iPad form factor before its release to take advantage of it in new ways - leveraging a portable legacy unrivalled in the industry. By contrast, Intel's OS offering in this space is starting from a poor base and trying to compete with the iPhone OS and Android, both of which have traction and mindshare in the marketplace. [And why does Android have a growing, and decent, applications base? Because there are devices out in the market that are now receiving critical acclaim...]
So, Intel are effectively having to start from scratch on the software front in the mobile space, and in my opinion they are trying to soften the blow of having to go down this route by referring to their X86 desktop legacy as a potential solution to a software gap that is as yet unplugged, even if that legacy doesn't exist.
The third leg on this stool is the issue of platform ownership. At present, a mobile phone manufacturer will decide what access developers have to their mobile platforms. The two most popular platforms with application marketplaces/stores are the iPhone and Android, with WebOS deserving a mention because of its recent acquisition by HP through its parent company Palm.
If the platform is an openly adopted one like Android for example, there is still an element of control that a manufacturer has to add or subtract the Android marketplace from the phone. If the platform is a closed one like the iPhone OS, the applications store is closed and scrutinised by the company, in order to control the user experience (although they are an exception here).
When you come to Linux, the story gets quite murky. There is no Linux applications store, but if there were it would have to support multiple platforms and distributions. The problem has to be narrowed down to just Meego from Intel's perspective, where the intention is for it to span ARM and Intel platforms in order to build - at present - Nokia-only devices. So far, I know of nobody who has taken Nokia's old open source Maemo OS and used it to make a device of their own except for Nokia, so questions abound regarding the scope of Intel's Meego OS effort in the and its ability to penetrate the marketplace.
The question on this third leg really is whether or not Intel will enforce the inclusion of their applications store into any device shipping with Meego? And also, will anyone really adopt Meego, given the histories of Maemo and Moblin?
The fact that Intel cannot draw upon any of their legacy desktop software for the mobile space - because Mobile is not based around windows - is being hidden behind a smokescreen of talk about the legacy of development that Intel has in the desktop world with Windows. If you waft the smoke away, I don't think that Intel has a strong story here.
Thursday, 29 April 2010
Apple change of strategy?
With all of the recent chatter regarding the disappearance of Intrinsity from the technology community, a question opens up about the strategy of the company that has in all likelihood bought them: Apple.
It all started with the announcement of the Hummingbird processor by Samsung - see my January 27th post about the ipad for a link in the comments to the press release surrounding the Samsung/Intrinsity collaboration to create a 1GHz implementation of the ARM Cortex-A8 processor.
When the iPad was announced, and the clock speed was said to be 1GHz, and it runs the iPhone OS, and the timescale was about right for Hummingbird to be in production, it was clear that the great performance of this hardware is in part down to the excellent implementation skills possessed by Intrinsity, since this chip is in all likelihood an exact match for a Hummingbird enabled System on Chip (SoC).
So why would Apple buy Intrinsity? I think the answer lies somewhere between their need for full custom silicon across the board (including at the processor RTL micro-architectural level), and the need to proceed quickly to market with the fastest possible implementation of off-the-shelf components, built in a semi-custom manner using Intrinsity's techniques.
Intrinsity possess a mix of technology and knowhow that allows them to do a really good job of making designs operate faster. They create a custom version of a given off-the-shelf processor design that addresses the slowest paths through the chip, and then adds special fast logic into those paths to speed them up. This makes the logical path between the clocked elements of the processor shorter, allowing the clock to run faster.
Because a strategy that focusses around Intrinsity relies on feeding them with other people's silicon designs (they do not actually design anything - they customise the generic design descriptions for fast operation in a given semiconductor companies process node), this suggests to me that Apple might be continuing with their use of ARM's licensable processor IP for the foreseeable future.
If Apple sticks with optimising off-the-shelf components in ways that other people cannot - now that they own Intrinsity - what does this say about their suspected ambitions towards designing their own in-house ARM processor?
I think the answer is two-fold.
Firstly, Apple are getting good performance out of ARM's "soft" cores by being clever in how they are implemented. This is quite low-hanging fruit, since the whole design and verification task has been done by ARM in creating the CPU in the first place. The "only" thing Intrinsity would have to do is identify the areas of the design that were holding back the top-end frequency, and make optimisations to those for speed.
Secondly, it isn't easy designing a processor from scratch. I believe that Apple retain a large percentage of the CPU design talent that they obtained from PA Semi - it would only have been the top-brass involved in business development that would have left to form Agnilux (now bought by Google). However, PA Semi's last ARM implementation would have been the experience gained from working on StrongARM at DEC and the ARM architecture has certainly moved on to include many new features. For an architect to get up to speed on that, and then work out what the CPU will look like is non-trivial, and bringing a team up to scratch on how to build that - and verify it - is not a two year task. It can take up to 4 years to turn around a CPU from scratch, especially if you've not worked with the architecture either for a long time, or on it's latest incarnation.
It also opens the question of what Apple thinks they can achieve over and above what ARM can achieve in CPU design? ARM are experts at designing their own CPUs, and have a spectrum of nifty high-performance implementations available in the applications CPU space. Give or take wanting a CPU with better performance or lower power, I can only think that Apple have decided to take their time on designing their own processor, and will stick for the foreseeable couple of years using optimised ARM macrocells.
If/once Apple do successfully complete the creation of their own CPU, the Intrinsity guys would then be deployed to analyse it, and make its implementation go faster. However, if you have that many in-house implementation engineers working alongside your designers, it would be possible to feed information to the designers about where the worst paths were in the design, and get them shortened before moving from the design description into the synthesis and layout stages of the project. This to some extent negates the value of having Intrinsity on board, because their skills lie in taking a pre-verified box off-the-shelf that you cannot change in any way, and making logically equivalent cycle accurate implementations of it that are faster and possibly lower power than if they were just synthesised automatically into standard bulk-CMOS. As such, it is a little bit questionable how much of Intrinsity's unique abilities can co-exist with the presence of an in-house design team, and what percentage benefit can be gained by using Domino logic on a design that is as fast as it can be in its RTL implementation.
So I've said it before in a previous post and it is worth repeating here - PA Semi designed CPUs in Apple devices will take many years to appear. Now I can add that they will only be marginally faster than what opther companies are capable of now that Intrinsity co-exist alongside a design team working towards a common cause. Just where I think Apple's own processor designs might wind up appearing in the product line is a matter for another post. But one thing is sure - Apple are intent on crafting their own silicon, and the game is changing still.
It all started with the announcement of the Hummingbird processor by Samsung - see my January 27th post about the ipad for a link in the comments to the press release surrounding the Samsung/Intrinsity collaboration to create a 1GHz implementation of the ARM Cortex-A8 processor.
When the iPad was announced, and the clock speed was said to be 1GHz, and it runs the iPhone OS, and the timescale was about right for Hummingbird to be in production, it was clear that the great performance of this hardware is in part down to the excellent implementation skills possessed by Intrinsity, since this chip is in all likelihood an exact match for a Hummingbird enabled System on Chip (SoC).
So why would Apple buy Intrinsity? I think the answer lies somewhere between their need for full custom silicon across the board (including at the processor RTL micro-architectural level), and the need to proceed quickly to market with the fastest possible implementation of off-the-shelf components, built in a semi-custom manner using Intrinsity's techniques.
Intrinsity possess a mix of technology and knowhow that allows them to do a really good job of making designs operate faster. They create a custom version of a given off-the-shelf processor design that addresses the slowest paths through the chip, and then adds special fast logic into those paths to speed them up. This makes the logical path between the clocked elements of the processor shorter, allowing the clock to run faster.
Because a strategy that focusses around Intrinsity relies on feeding them with other people's silicon designs (they do not actually design anything - they customise the generic design descriptions for fast operation in a given semiconductor companies process node), this suggests to me that Apple might be continuing with their use of ARM's licensable processor IP for the foreseeable future.
If Apple sticks with optimising off-the-shelf components in ways that other people cannot - now that they own Intrinsity - what does this say about their suspected ambitions towards designing their own in-house ARM processor?
I think the answer is two-fold.
Firstly, Apple are getting good performance out of ARM's "soft" cores by being clever in how they are implemented. This is quite low-hanging fruit, since the whole design and verification task has been done by ARM in creating the CPU in the first place. The "only" thing Intrinsity would have to do is identify the areas of the design that were holding back the top-end frequency, and make optimisations to those for speed.
Secondly, it isn't easy designing a processor from scratch. I believe that Apple retain a large percentage of the CPU design talent that they obtained from PA Semi - it would only have been the top-brass involved in business development that would have left to form Agnilux (now bought by Google). However, PA Semi's last ARM implementation would have been the experience gained from working on StrongARM at DEC and the ARM architecture has certainly moved on to include many new features. For an architect to get up to speed on that, and then work out what the CPU will look like is non-trivial, and bringing a team up to scratch on how to build that - and verify it - is not a two year task. It can take up to 4 years to turn around a CPU from scratch, especially if you've not worked with the architecture either for a long time, or on it's latest incarnation.
It also opens the question of what Apple thinks they can achieve over and above what ARM can achieve in CPU design? ARM are experts at designing their own CPUs, and have a spectrum of nifty high-performance implementations available in the applications CPU space. Give or take wanting a CPU with better performance or lower power, I can only think that Apple have decided to take their time on designing their own processor, and will stick for the foreseeable couple of years using optimised ARM macrocells.
If/once Apple do successfully complete the creation of their own CPU, the Intrinsity guys would then be deployed to analyse it, and make its implementation go faster. However, if you have that many in-house implementation engineers working alongside your designers, it would be possible to feed information to the designers about where the worst paths were in the design, and get them shortened before moving from the design description into the synthesis and layout stages of the project. This to some extent negates the value of having Intrinsity on board, because their skills lie in taking a pre-verified box off-the-shelf that you cannot change in any way, and making logically equivalent cycle accurate implementations of it that are faster and possibly lower power than if they were just synthesised automatically into standard bulk-CMOS. As such, it is a little bit questionable how much of Intrinsity's unique abilities can co-exist with the presence of an in-house design team, and what percentage benefit can be gained by using Domino logic on a design that is as fast as it can be in its RTL implementation.
So I've said it before in a previous post and it is worth repeating here - PA Semi designed CPUs in Apple devices will take many years to appear. Now I can add that they will only be marginally faster than what opther companies are capable of now that Intrinsity co-exist alongside a design team working towards a common cause. Just where I think Apple's own processor designs might wind up appearing in the product line is a matter for another post. But one thing is sure - Apple are intent on crafting their own silicon, and the game is changing still.
Wednesday, 21 April 2010
ARM takeover by Apple? Rubbish
The speculation about ARM being taken over by Apple is unfounded. ARM's business model runs contrary to everything Apple stands for. Apple are a closed, secretive, paranoid business intent on guarding every aspect of every product. ARM has an open, partner driven business of supplying intellectual property to the whole of the electronics industry. If Apple bought ARM and kept them entirely to themselves, they would be stripping the entire electronics industry of the only standardised supplier of IP, leaving the army of ARM partners and customers completely in the lurch, without any further prospect of plugging their roadmaps in future. This would have the consequence of leaving Intel as the only architecture left inside the marketplace. Other niche suppliers such as MIPS would then be the only companies left in the silicon IP business, each of whom do not have anywhere near the level of ecosystem that ARM has to fight off the insurgence of Intel. This takeover would devalue the price paid for ARM to such an extent that the acquisition alone would be suicidal for both Apple and ARM.
Pure rubbish, espoused by people in the city who don't have a clue about ARM, or what a poor fit it would be for Apple.
Pure rubbish, espoused by people in the city who don't have a clue about ARM, or what a poor fit it would be for Apple.
Subscribe to:
Posts (Atom)
Posts
