Zdnet

Why I dumped my Droid

十月 13, 2010

Why I dumped my Droid

I wanted to love the Droid X. I really did. The specs are awesome, and it got rave reviews from friends and colleagues, who overwhelmingly recommended it as the best of the Android phones. Even Paul Thurrott enthused about his Droid X, calling it “hands-down superior to anything designed in Cupertino.” So I picked one up from Verizon, secure in the knowledge that I had 30 days to decide whether to keep it or send it back.

Initially, I found a lot to like about this device. But two full weeks of carrying it around the Western U.S. on an assortment of business and pleasure trips was enough for me to decide, with no hesitation, that for me, Droid doesn’t cut it.

What went wrong?

Initially, at least, I was seduced by the device—or, more accurately, by Verizon’s network, which actually works in my office. That’s a pleasant and striking contrast to AT&T, whose signal doesn’t penetrate to my home office at all and only reaches about half a bar on the other side of the house. My wife, who loves her iPhone 3GS, has learned that the best way to make it through an entire conversation on the iPhone is to sit in one spot on the couch and try not to move around too much. If I could use it in my office, the iPhone might be an option, but I can’t, so it isn’t.

Update: A commenter asks why I don’t just get an AT&T MicroCell, which connects to the local WiFi network to send and receive phone calls. I tried it earlier this year, but unfortunately, the layout of our home doesn’t allow the signal to reach both the office and the living areas at the same time. In addition, it adds $20 a month to the already high phone bill. So it’s a nonstarter.

Eventually, though, the novelty of being able to use a mobile phone toactually make phone calls wore off, and I had a chance to evaluate this device and the rest of the Android platform on its merits. And that’s where things began to fall apart.

For starters, I realized the device was just too big for my taste. That big, bright, high-resolution screen is part of its appeal, but it also makes it awkward to stash in a side pocket and a little too big to fit comfortably in a shirt pocket.

And although the Droid X is impressively fast at most tasks, it pays a familiar price for that performance, generating a level of heat that I found literally uncomfortable. It wasn’t hot enough to fry an egg, but on more than one occasion I flinched when I touched the back of the device.

Whatever it was doing with the CPU was having an impact on battery life as well, because I had a hard time getting through a working day (much less from sunup to sundown on a long midsummer day) before the battery died. Generally, I found that it would quit before the nine-hour mark, which simply isn’t acceptable for business travel. Using the turn-by-turn GPS seemed to be a particularly big power gobbler. This was particularly inconvenient on one extended driving trip, where the Droid battery conked out about a half-hour before we reached our destination, and we had to use a backup phone to call for directions.

I’m sure I could have figured out how to squeeze more battery life out of the Droid by tweaking its settings, but just finding those settings was what finally sent me over the edge.

The Android OS, at least as customized by Verizon on this device, had dozens of icons spread out over seven screens. Some were part of the OS, some were third-party apps, and the whole thing looked like it had been designed by a committee of mad scientists.

The first app I downloaded was a Task Killer, which I used to shut down all of the running apps (more than 20) that were starting by default on this phone, including Skype, which I don’t use. But within a minute or two, all of those apps restarted themselves. Gee, thanks, Verizon.

Ultimately, what unsold me on the Droid X in particular and on the Android platform in general was its complexity. I understand why geeks love this OS. Like Linux, it’s a tweaker’s playground. I’m sure I could have solved all my problems by rooting the device, downloading a new kernel, and starting from scratch.

But the last thing I want is a device that requires constant babysitting.

So, after spending a few hours trying to figure out how to make everything make sense, I surrendered, packed the whole thing up, and shipped it back.

And the search continues.

—————————————————————————————————–

ARM’s Race: Why your next laptop might not have Intel inside

十月 12, 2010

By dennistai

ARM’s Race:

Why your next laptop might not have Intel inside

By John Morris | September 29, 2010, 12:05pm PDT

It’s no secret that the performance margin between smartphones and netbooks is shrinking. But soon it could disappear altogether. Over the last few weeks, ARM Holdings and its customers have announced new products that promise to not only power a wave of “super-phones” coming to market in 2011, but also push the architecture into new markets such as tablets, netbooks and laptops and even servers.

Unlike Intel, ARM doesn’t manufacture or sell chips. Instead it licenses its recipes to chip designers who customize them and add other ingredients to develop more complex SOCs (System-On-a-Chip). These customers can either manufacture the SOCs themselves or hire a foundry to fabricate the chips. Nearly all cell phones use ARM-based processors developed in this manner.

Most current smartphones use processors based on ARM’s Cortex-A8. This platform was initially designed for chips manufactured on a 65nm low-power process with a single CPU core running at 500MHz to 600MHz. Texas Instruments’ OMAP 3430, the processor in the originalMotorola Droid and Palm Pre, is an example. Newer implementations from companies such as TI’s OMAP 36xx and Samsung’s S5PC110-some of which are manufactured on the more advanced 45nm node-can reach speeds up to 1GHz. These processors power the most popular smartphones such as the Apple iPhone 4 (and iPad), Samsung Galaxy line, and Motorola Droid X and Droid 2.

ARM’s follow-on, the Cortex-A9, is only just beginning to enter production. The key difference is that the A9 supports symmetric multi-processing in chips with up to four cores (ARM refers to the multi-core versions as MPCores). TI’s OMAP 4430, which has two 1.0GHz Cortex-A9 cores and Imagination Technologies’ PowerVR SGX540, is schedule to be in production this year. Earlier this month, Samsung announced its Orion processor, which will also have two 1.0GHz A9 cores, but the company claims it will have 5x the graphics performance of its current S5PC100, which uses the PowerVR SGX540. Orion will be in production in the first half of 2011. ST-Ericsson, a joint-venture of STMicroelectronics and Ericsson, has two SOCs using Cortex-A9 cores, both of which integrate ARM’s Mali-400 graphics and a 3G baseband. The U8500 for high-end smartphones, which has dual 1.2GHz A9 cores, will be in production later this year. The U5500 for mid-range smartphones, which will have slower frequency A9 cores and graphics, should be in production in mid-2011. Nvidia’s Tegra 250, which two 1.0GHz A9 cores combined with Nvidia’s graphics technology, should show up in smartphones from Motorola, LG Electronics and HTC in early 2011.

The BlackBerry Playbook, which RIM announced this week, will use a 1GHz dual-core A9, though for now the company isn’t saying which one. Most BlackBerries use Marvell processors. Both Marvell and Qualcomm are among a handful of companies that have ARM architectural licenses, meaning they can design chips that are compatible with the ARM v7-A instruction set, but are customized from the ground up rather than starting with ARM Cortex cores. This allows these companies to design processors that operate at faster speeds and/or more efficiently, though the design process takes longer and is more costly.

Last week, Marvell announced the latest addition to its Armada line. The Armada 628 has three general processing cores-two running at 1.5GHz for high-performance and one operating at 624MHz for low-power-plus six additional accelerators for 3D graphics, video encoding and decoding, audio, encryption and image processing. It is currently sampling, but Marvell hasn’t announced when it will be in production. Qualcomm has had a lot of success with its Snapdragon processor in products such as HTC’s Evo 4G, HD2 and Droid Incredible, Google’s Nexus One and Sony-Ericsson’s Xperia X10, and it will soon release more powerful versions based on 45nm process technology, The first ones, which will be in production this year, will have a single Scorpion core running at 1.0GHz for smartphones and up to 1.3GHz for tablets and netbooks. Next year, Qualcomm will start production of dual-core Snapdragons running at 1.2GHz for high-end smartphones and 1.5GHz for tablets and netbooks. Qualcomm’s advantage is that it combines the processor with a 2G/3G cellular baseband on a single chip.

Exactly how much faster these smartphones and other mobile devices will be depends on a lot of factors including the process technology, frequency, graphics and other processing engines for tasks such as video. The current A8 is capable of around 1,200 Dhrystone MIPS (million instructions per second), according to ARM’s specifications, and I’d estimate that a dual-core A9 is capable of perhaps 4,500 Dhrystone MIPS, or nearly a 4x improvement. (Dhrystone is a standard test that measures integer performance.) Another advantage to these dual-core ARM processors is that they will support full 1080p video and most will have HDMI-out to drive an external HDTV. ARM says the dual-core Cortex-A9 is competitive with the single-core 1.6GHz Intel Atom processor in most netbooks.

That brings me to another interesting Cortex-A9 announcement. Earlier this month, ARM announced that Nufront, a technology company based in China, had developed an SOC with a 2.0GHz dual-core A9 designed for laptops and netbooks. Manufactured using a 40nm process, the NuSmart 2816 chip also includes ARM’s Mali-400 graphics, a memory controller, a 1080p video co-processor and the required I/O components to build a laptop. Nufront said it will demonstrate laptops using the NuSmart 2816 at the Consumer Electronics Show in January. The announcement illustrates not only how ARM’s performance is reaching up into the PC world, but also how even a small start-up with little processor experience can use one of ARM’s “hard macros” and a contract manufacturer to bring a chip to market quickly and at a relatively low cost. I expect to see more companies in Asia pursue this strategy, leading to lower-cost tablets and laptops.

After the A9, ARM plans to make a big leap in performance and features with its Cortex-A15, previously known by the code-name Eagle. Designed for 32nm and beyond, the A15 will support up to four cores in a single cluster running at speeds up to 2.5GHz, though smartphones and other mobile devices will use dual-core versions running at up 1.5GHz. ARM says the A15 will deliver about 5x the performance of a current smartphone with a single Cortex-A8 core while using about the same amount of power. The A15 will also include several new features-system coherency for clusters, hardware support for OS virtualization without modifications to the kernel, larger memory addressability, and memory error detection and correction-that make it clear that ARM is after not only high-end handsets, but also tablets, laptops, servers and communications infrastructure products such as wireless basestations. Samsung, ST-Ericsson and Texas Instruments are the “lead licensees,” but don’t expect to see products based on the Cortex-A15 until late 2012 at the earliest.

Of course, neither Intel nor AMD is sitting still. Intel has finally released a dual-core version of Atom for netbooks, the 1.50GHz N550, and Acer, Asus, Fujitsu, Lenovo, LG, Samsung, MSI and Toshiba have all started shipping products using it. At the Intel Developer Forum earlier this month, the company was also showing some tablets based on its Moorestown Atom Z6xx SOC, which Intel has said is comparable in terms of performance to an ARM Cortex-A9 MPCore dual-core processor. Theperformance looks promising, but it still draws too much power for handsets, and I don’t expect to see many smartphone designs until Intel releases the successor, the Medfield platform, sometime in 2011. Early next year Intel also plans to release the Oak Trail platform for tablets. Around the same time, AMD will launch its Ontario dual-core processor, manufactured by TSMC on a 40nm process, for netbooks and ultra-thin laptops.

It’s too early to tell whether ARM will succeed in “moving up the stack” and pushing its technology into PCs and servers–just as it is too early to tell whether Intel will be able to successfully push x86 down into smartphones and tablets. The companies have very different businesses, and the head-to-head clash between the two can be a bit overblown. In areas where x86 operating environments and applications are entrenched (and growing), such as the bulk of the server marker, Intel and AMD are likely to maintain much of the business. In other categories such as tablets, netbooks and laptops where users are increasingly comfortable with other operating systems and applications-Apple’s iOS, Google’s Android and Chrome OS, BlackBerry-ARM could make real inroads. One thing’s for sure: the choices for mobile computing are about to become much more complicated.

http://www.zdnet.com/blog/computers/arms-race-why-your-next-laptop-might-not-have-intel-inside/3994?tag=nl.e539

————————————————————————————————————————