Another article has been posted about how Qualcomm’s Snapdragon platform is ideal for low cost, long battery life, small form factor notebook PCs. We heard about this a few weeks ago but Brooke Crothers of CNet visited Qualcomm to hear more detail about what’s going on. The article gives me a chance to dive into the details and give some thought about what’s happening here in both technology and market terms. Is it significant or not?

First of all a little background about Qualcomm’s Snapdragon. Its a small-form-factor, mobile computing platform (think of it as a ready-to-use computer on a tiny motherboard a bit like the image you see below-right.) that includes a CPU core which is based on a licensed ARMv7 architecture. ARMV7 is the architecture used in the ARM A8 Cortex CPU design that you can now find in the new Archos devices, the Open Pandora and BeagleBoard (image below-right) projects. Snapdragon has been a four year, $350 million project. It’s not clear how many snapdragon versions there are but the one that CNet are talking about is the new dual-core QSD8672 capable of clock speeds up to 1.5Ghz. The platform also includes the following features: (Details from Qualcomm.)

• WWAN, Wi-Fi and Bluetooth connectivity
• Seventh-generation gpsOne® engine for Standalone-GPS and Assisted-GPS modes, as well as gpsOneXTRA^TM Assistance 
• High definition video decode (720P)
• 3D graphics with up to 22M triangles/sec and 133M 3D pixels/sec
• High resolution XGA display support
• 12-megapixel camera
• Support for multiple video codecs
• Audio codecs: (AAC+, eAAC+, AMR, FR, EFR, HR, WB-AMR, G.729a, G.711 , AAC stereo encode)
• Support for Broadcast TV (MediaFLO^TM , DVB-H and ISDB-T)
• Fully tested, highly-integrated solution including baseband, software, RF, PMIC, Bluetooth, Broadcast & Wi-Fi

Looking at the article from CNet, these tech specs might be referring to a previous Snapdragon platform because the CNet article mentions 1080p video and an ATI graphics core.

As JKKMobile pointed out , one of the important things to note about this platform is its inability to run Windows desktop software meaning that it’s restricted to mainly Linux-based distributions or WindowsCE-based operating systems but lets take a look at a few other aspects of the platform first.

Performance

According to some references on the Internet that I’ve been able to cross-check, raw single-core CPU performance of the platform, an important factor in rendering browser-based pages and applications, is in the order of 2000 Dhrystone MIPS at 1Ghz. Obviously a dual-core version, clocked at 1.5Ghz means we’re in the region of 3000-5000 MIPS which means, if you take a look around for Dhrystone tests on the Atom N270 used in netbooks, its right in the same ballpark. The N270 appears to return about 4000 Dhrystone MIPS. These figures don’t indicate anything about the real-world performance but they do tell us that, given good memory and storage performance the results are good enough to run a basic desktop OS.

Power Efficiency

This is something I think we should be careful to remain realistic about. While ARM cores are extremely efficient and idle at extremely low levels, we’re at the point where CPU power usage on Atom and ARM-architecture CPUs aren’t a world apart from each other. When taken as a ratio of total system power drain, including screen, radios, DC-DC components and storage, the CPU is well under 50% of the equation. In 10" screen devices, it’s even less. 20% maybe. The major power advantage here is gained through on-board integration. Packing processing cores, communications silicon and memory tightly together, unifying the power-saving methodology across the board and writing efficient firmware is they key. Qualcomm have a lot of experience in that and I expect to see well-designed Snapdragon-based netbooks in 2009 to be running in the order of 5W  about half the power that an N270-based notebook will use. Due to the high level of integration, the boards will be smaller too. The solution gives designers the option to make the devices smaller or run with longer battery power. With more space for batteries, you could even see devices where thin Li-Poly take up most of the underside of the netbook and offer capacities that will allow you to leave your netbook on all day. An always-on netbook is a seriously interesting prospect!

Cost

It’s difficult to do a complete estimate on this but experience tells me that in quantities, the highly integrated Snapdragon platforms would be significantly cheaper than a full Atom-based motherboard build-out.

Markets

This could go two ways. Initially I would expect to see manufacturers use the new Snapdragon platform to make cheaper Linux-based netbooks that will go into some of the emerging markets in India and China. These markets are less sensitive to desktop OS types due to the shorter history of XP and Vista. We will see some low-cost and long-battery life versions coming to ‘our’ markets (I speak to 80% of readers when I say that; Americas, Europe, Japan) but without XP, the devices won’t get the traction and channels that the Intel-based devices will. There’s one thing that could change this though and I know for sure that people are working on this. High-quality consumer focused, easy-to-use, reliable, branded, thin operating systems. Operating systems that completely hide their roots and offer a slick UI, rich applications suite, content and branding. So far we haven’t seen anything but there are a few projects and companies out there that are working on this. The Open Handset Alliance work has potential. Moblin too. They will need a good company to ‘finish’ the product though and the branding will need to come from a big name in order to give it traction. It will need that all-important app-store too. The one OS that could really change the game though is WM7. I don’t really expect this to happen but if it is an OS that offers a large-screen experience and the opportunity to work with native document formats and in both a ‘business’ and ‘consumer’ mode, there’s no reason why it can’t be used as a netbook OS.

Back to that article now. (Sorry, I got carried away. I actually wanted to write a quick re-blog article!) CNet mentions a few companies working on solutions:

Companies including Acer, Asus, and Toshiba are planning devices based on Snapdragon, according to Qualcomm. Acer, for example, will initially introduce Qualcomm’s Gobi 3G modem into its devices, then gravitate to products based on Snapdragon.

Of course we need to take that with a pinch of salt considering the fact that there are already supposed to be 15 ultra mobile devices out there running on Snapdragon.

Gizmodo have just completed a browsing speed and accuracy test with three high-end windows mobile devices using Pocket IE and Opera 9.5. The results should hardly be a surprise. There isn’t a single reasonable result among them with page load times well over a minute in many cases and very few of the devices rendering the pages well.

In the test, Gizmodo used the Sony Xperia, HTC Fuze, Samsung Omnia and Samsung Epix. Some of the newest WM-based phones you can buy.

Opera 9.5 appears to have turned in a better level of quality and speed than Pocket IE but there’s still a bunch of ‘fails’ in there which would turn off anyone thinking of relying on the given combo.

We’ve done similar tests here in the past which have proven that, on average, with some of the best ARM-based devices you can find and under good conditions, average page load times are twice as long when compared to on low-end ultra mobile PCs. We’ve even done some extensive Opera Mobile 9.5 testing and can confirm that while it does render well, it needs a lot more horsepower underneath it than the average smartphone can provide. Nothing in the smartphone world, including the iPhone, comes close to the speed and accuracy of even the lowest-level ultra mobile PC or Intel-based MID so once again I hear myself saying; If you or your business relies on fast, accurate access to Web-based resources through a browser, don’t risk problems or waste time by using a sub-standard solution. Don’t try and push everything onto one device. Buy a dedicated device. If not for the speed and quality, do it to preserve battery life for your important voice calls!

Take a read of the article and the HUGE bashing that WM gets from author, Matt Buchanan. Its a fun read!

Source: Gizmodo Via Friendfeed

Qualcomm demonstrated a Snapdragon based netbook in London last week and gave us a few hints as to how netbook-style devices based on it might turn out. Silicon.com, who published images and information from the event, asks if it’s the next-gen netbook.

‘Alternative-gen’ is more like it in my opinion.

We probably won’t see many of these in the western world as we like our processing power (way) too much but I do expect these to be a success in 2009 and 2010 in other markets. Markets that haven’t really had the chance to use laptops and who’s main form of communication is a GSM phone could benefit from this.

We could see one or two ultra-light, ultra-thin devices here but for them to be a success, they are going to have to offer levels of portability, user interfaces and battery life that only a few companies are in the position to be able to achieve.

As for battery life, in normal Internet-connected scenarios, there won’t be much advantage as screens and radios take the lions-share of the battery. There’s a lot of talk about multi-day battery life but the people that talk about this fail to mention that it’s standby life. Once you start switching those transistors on and off to decode today’s intensive web pages, power drain rises into the same ballpark as ultra low power X86-based devices. Even Qualcomm themselves are being honest about this.

Snapdragon will enable mini laptop style devices – with screens of between nine and 12 inches – to run for between four to six hours without needing to be charged, according to a spokesman.

The last time I spoke to Texas Instruments, they were talking about similar figures too. The 3-4W total power drain range is about the limit of technology for 8-12″ screens in the short term.

Personally, I think the more interesting market, is the mobile Internet device market where 5″ screens enable the best balance between size, portability and quality for video, navigation and our old friend, the world wide web. It’s here, where screen power drain is much smaller, that the new ARM-based platforms could make a bigger impact. Those ARM-based devices were supposed to be in the market already but appear to have been delayed. According to word-of-mouth reports that i’m hearing though, devices are on their way. Indeed, I wouldn’t be surprised to see Apple putting out a larger, more powerful iPod web and video-focused device on these architectures in 2009. ComputerWorld thinks so too. Unless Apple have something up their sleeve with PA-Semi

Word has been spreading around lately that the second-gen iPod Touch [Portal page] has a faster CPU than any of the previous Touch devices (including the iPhones). This was a bit strange when I first read it because I own an iPhone 3G, and it was easy to see that in comparison, a friends second-gen iPod Touch was visibly faster when there were on-screen GUI actions. I thought that because it was so clear that it was just a known fact that the second-gen iPod Touch was faster. And yet here I am writing this post to let people know that in fact, the second-gen iPod Touch has a faster clock speed than the original iPod Touch, as well as the iPhone EDGE, and iPhone 3G.

From what I’m looking at, it appears as though the ARM CPU in the second-gen iPod Touch is different than those in the other Touch family devices (at first I thought it was simply a change in clock speed). Second-gen iPod Touch: 523MHz, up from 412MHz in the rest of the family of devices. Regardless of the slightly updated CPU, I think it is possible for Apple to turn up the clocks on the other devices to reach the same speed, however there are other important factors to consider before doing that, the biggest of which is battery life. It seems as though the second-gen iPod Touch’s hardware is quite a bit more efficient than the original iPod Touch. Even with the more powerful CPU, the second-gen iPod Touch has better battery life than the original. Here is a list from greatest speed to least for the Touch family:

1. second-gen iPod Touch
2. iPhone 3G
3. iPhone EDGE
4. first-gen iPod Touch

In my personal experience with all of these devices I would say that the speed differences between the iPhone 3G, iPhone EDGE, and first-gen iPod Touch are entirely unnoticeable. Only when you compare these devices to the second-gen iPod Touch can you see a difference.

The real question is why didn’t Apple mention this increased CPU speed (they didn’t mention it IIRC) especially if they managed faster speed with improved battery life? Maybe they meant to limit the CPU to the same clock as the rest of the devices but it slipped by them? Hopefully we won’t see games that cater to the second-gen iPod Touch’s faster CPU. At this point, every piece of software is interoperable across the entire Touch family, it would be a shame to see software that ‘works’ on the other three devices, but is really meant to be run on the second-gen iPod Touch.

Following through behind the ARM / Ubuntu announcement last week is another shot in the direction of Intel’s MIDs. Adobe’s Flash 10, probably the most commonly used multimedia plug-in in the Web world and a significant part of the full web experience, is coming to the ARM architecture. Not only will you be seeing full Flash 10 support on the next generation of evolved smartphones and mobile Internet devices based on the ARM Cortex core which are starting to trickle in now via specialist devices such as the Archos 5, Open Pandora and the next generation Nokia Internet Tablet but existing ARM11 architectures will also see a port. In addition to the Flash 10 announcement, Adobe have announced that AIR will be available so applications like twhirl and the ebay app could end up in your hand within the next year.

The press release talks about being able to ‘remove the barriers to publish content and applications seamlessly across screens.’ and points towards the website – Openscreenproject – that was announced earlier this year.

“Adobe Flash is the leading video format on the Web today, and this collaboration with ARM is another important step towards bringing the complete Web experience to mobile devices worldwide, inch said Gary Kovacs, general manager and vice president, Mobile and Devices at Adobe.

Read: TGDaily

Read: ARM Press

The crossover is finally happening. The smartphone CPU designer is moving into the PC market while the PC CPU designer is moving into the smartphone market. 

ARM are announcing a partnership with one of the best-known names in desktop Linux and are making a very significant move into the world of low-cost, low-power, connected computing with Canonical who will port the full desktop version of their Ubuntu OS to the ARMv7 architecture ready for release in April 2009. 

Press-release and more, below.

Read the full story

Its nice to see that we might not have to wait too long for a better browser as part of Windows Mobile but is this going to be another case of ‘nice software, shame about the hardware?’ I say that because it was the conclusion I came to with Opera 9.5 a while back. The accuracy, UI and features were vastly improved but the bottle-kneck is the processor. There’s really no way around it. To convert todays very complex html, script and media that arrives through your Internet connection into a useable screen of information takes a lot of transistor switching. ARM9 and ARM11-based devices just don’t have that power. ARM Cortex and Atom do.

Show me a ‘Pocket PC’ a-la HTC Universal (left) with a 4.5″ 800×480 screen, 3G and Wifi, WM6, 5-hour online battery life, running on a a Cortex A8 core and we can talk. The Archos 5 does a fairly speedy job. The Open Pandora device too so we know there’s no hardware limitations. It’s just a case of putting the right software with the right hardware!

Source: Download Squad

No prizes for guessing what company said that! ZDnet highlights it with "Intel slams ‘slow’ iPhone ARM CPU."  Gizmodo follows up. Engadget too (along with a lot of comments) and now it’s reached the front page of Techmeme. Oops!

I’m all behind Intel for their work in squashing the X86 architecture down to smartphone-sized levels with Moorestown and there’s an element of truth in the fact the the iPhone is underpowered for Internet apps but the way I see it is that it’s not really about CPUs anymore and as such, it comes across rather uncool to focus on it, especially when your partners are behind schedule on getting mobile Internet devices out of the door and you still have work to do to reach smartphone levels of power efficiency.

Both ARM and Intel have reached similar (consumer acceptable) territory in terms of watt/performance [*1] with their respective core architectures and yes, Intel’s solutions are probably more powerful and will definitely be attractive to the power-user but that’s a tiny part of the equation that goes together to make a thrilling consumer device. Intel’s main task now is about the integration of the CPU, GPU, controllers and radios into the smallest space possible with the highest platform efficiency. Both ARM and Intel’s ecosystem is highly capable of achieving that but there’s even more to consider. Industrial design,marketing and most importantly, software.

Intel are betting on one of the most fragmented software environments out there – Linux. They want to create a new, mobile-focused stack with it and surround it with quality ISVs. ARM’s partners want to use Linux too but they already have well-supported stacks with the same ISVs and big dev communities around them to. To drive a new Linux stack you need control, lead (in-house, paid, full-time) developers, Linux distribution partners and, if you want to take advantage of the existing application base, the skills of the people that wrote them. That means you need to be Linux-geek-cool and you need to show the dev community that you are a caring, sharing type. You also need to have an easy channel for them. An app, store. It’s critical now, not only for the developers, but to enable an important revenue stream in e-commerce for Intel and the partners. Moblin doesn’t have either of those two elements. Highlighting your advantages is one thing but making statements that attack the other side (where some of your community sits) won’t win you any hearts in the open-source world.

To be fair, I wasn’t there and haven’t been able to hear the comments in context and having interviewed Pankaj Kedia a few times, I know that he knows what he’s talking about. Add Steve Jobs’ recent comments into the mix and the stones that came from the ARM camp a few months ago and you can understand why these comments happen. Intel’s ultra mobile products are good and getting better and there may even be an Apple product in the works that gives Intel this confidence but when I hear comments like this, it just sounds cheap.

*1 The latest ARM-based devices have almost closed the 9-second penalty I demonstrated a year ago. See this article about the Archos 5.

Update. Intel has corrected its comments in a statemnet here.