Why Flash failed on Android, and what it means for Adobe

Something happened yesterday that seemed unthinkable a year ago — Adobe gave up on Flash Player for mobile devices. As you were chugging through another Wednesday, Adobe quietly removed the Flash plug-in from the Google Play Store, effectively admitting defeat. The sad thing is, it should have happened a long time ago — Flash on Android was doomed from the start.

Promises, promises

I distinctly remember being incredibly excited when Adobe announced the impending release of Flash for Android. It was 2010 and the phone to have was the Nexus One. The early video demos looked fairly good, with video playing smoothly on the Nexus, along with promises that the battery impact wasn’t too bad. Unfortunately, this was all smoke and mirrors.

The initial release of Flash was only good for the novelty effect. “Look, video playing in a web page!” we exclaimed. The video playback was choppy much of the time, and battery life was severely impacted. Excuses were made; this was the first release, and the Nexus One, while at the top of the hardware heap, was still only running a 1GHz chip.

Hotfixes continued to roll out, but the bugs were not squashed. By the time Android 3.0 Honeycomb was announced, Adobe was sounding like a broken record. We were told that the advancements in Flash 10.2 combined with Honeycomb would be the video experience we always wanted. Finally Flash would take advantage of multi-core ARM chips and GPU acceleration.

You can probably guess what happened next. Flash was still a painful experience on Android. Sure, it technically worked a bit better. You could watch video and interact with Flash objects on web pages. However, you had to deal with the low framerates and general scrolling lag when Flash was enabled. By the time Flash 11 came out in late 2011, no one cared anymore.

Despite a small contingent of Flash defenders, the larger Android community has moved beyond Flash. The writing was on the wall, and a few months ago, Adobe announced that it was ceasing development on Flash for Android. In some ways, that felt like the end, but it wasn’t until the app was erased from Google Play that we could actually say Flash is dead on mobile devices.

The truth about mobile

Adobe was right to want a part of the burgeoning mobile device ecosystem. It correctly ascertained that video viewing on smartphones was going to be a big deal, but the same factors that made Android phones popular made Flash very difficult to implement.

The main method of interacting with a smartphone is the touchscreen, and Flash was a fundamentally poor experience on a touchscreen. Flash content, especially video, assumes you have a mouse cursor that can hover, click, and drag. There is no equivalent to this on an Android phone. If you try to drag a video progress bar, for example, you will just scroll the screen. The buttons in Flash content were also far too small for use on an inexact capacitive touchscreen.

Adobe tried to get around this by allowing for full screen video with Flash, but I found that this rarely worked much better. There was also a failed push to get more touch-friendly video player interfaces used in Flash content. There just wasn’t enough interest in mobile Flash to make this happen.

The performance and battery life issues were probably the most vexing ones Adobe had to contend with. Flash was designed to run on desktop x86 systems. The lower-power ARM chips that made smartphones great also hindered Flash’s performance. Getting its plug-in to work on ARM devices was like hammering a square peg into a round hole. Adobe did it, but it was a mess.

You’ve probably watched plenty of videos on your mobile device, and they worked fine. That’s because ARM devices have hardware video decoding for H.264 video. Virtually all the HTML5 video you’ve ever come across is just an embedded H.264 video in an MP4 wrapper. Phones are very good at decoding this kind of file without draining the battery. Flash could never compete with this kind of built-in performance.

Adobe is not a small company, but it was unable to tweak and optimize the bits enough to make video play acceptably. Flash was never going to work well on Android, no matter how hard Adobe tried.

Adobe’s future

Each time I’ve tested Flash on Android, it has been harder to find good test pages. In the two years since Flash appeared, there has been a seismic shift toward HTML5 video. Sites like the New York Times and Time.com used to serve up Flash content to Android devices, but eventually moved to HTML5.

If Adobe could have made Flash work, it would have been a nice alternative. You can put any kind of video behind Flash and it will work just fine for the user wanting to watch a quick video. HTML5 with H.264 works very well, but it’s narrow. The H.264 codec is patent encumbered, which has led some to push for Ogg or VP8 to be the new standard. But again, we have devices with H.264hardware decoding.

Flash is dead on mobile, but it still has a place in the desktop world. HTML5 is a very cool technology, but it exposes the video files to the user. For copyrighted content with DRM, Flash will still be a desirable option. Flash games also occupy a large, but shrinking segment of online entertainment. It’s going to take time for all this to shift to HTML5, and Adobe hopes its “Flash Next” project in 2013 can hold onto these two footholds a little longer.

Adobe still has a presence on Android, as well. Photoshop is a juggernaut of a brand, and Photoshop Touch is a great app. I expect to see Adobe pushing this product harder on mobile platforms; maybe even getting a stripped down Photoshop Touch on phones in addition to tablets. The company also has the AIR platform to fall back on. AIR is a framework you can get in the Play Store (as well as on desktops) that supports Flash-like apps. There aren’t a ton of these on Android, but again, Adobe is going to promote it.

Adobe probably could have spent twice the man hours on the development of Flash for Android, and it still wouldn’t have been good enough. Even when Flash came in handy, it was never a good experience. I was never enthused to come crashing into a Flash web page, and most other users weren’t either. Flash didn’t work on Android, but at least Adobe tried to get out in front of a trend. You have to at least give them credit for that.

Source

Why Nvidia will rush to Tegra 4 as fast as it can

When Nvidia finally announced the Tegra 3 platform a few months ago, it was undeniably the best ARM-based system-on-a-chip (SoC) to be had. However, as other manufacturers begin to detail next generation ARM chips, Tegra 3 is already starting to lose a little of its luster. Last year’s Tegra 2 SoC had an advantage for a number of reasons, but the Tegra 3 will be on a level playing field, and the competition will be tough.

Early reports from MWC indicate that Nvidia stamped the first samples of Tegra 4 (codenamed Wayne) in December, and have sent them to OEM partners to work with over the coming months. Wayne will run multiple Cortex-A15 cores with a smaller 28nm manufacturing process, whereas Tegra 3 still uses Cortex-A9 at 40nm. Improved Tegra 3 parts with higher clock speeds are also on the way.

Prior to 2011, building a product with a Tegra chip was essentially the kiss of death. The early Tegra parts powered uninspiring devices like the Microsoft Kin, Zune HD, and Samsung YP-M1. None of them did particularly well, and many prototype smartbooks running Tegra were cast aside as well. All that changed with Tegra 2.

Nvidia’s Tegra 2 SoC was running two ARM Cortex-A9 cores and had a solid GPU behind it. Nvidia was able to get Google to officially bless its platform by making it the reference design for Android 3.0 Honeycomb. Because of the closed nature of Honeycomb, OEMs were forced to use Tegra 2 until Android 3.2 expanded support. By that time, Tegra was firmly entrenched and powering not just tablets, but phones like the Motorola Atrix 4G, Droid X2, and LG Optimus 2X.

For its encore, Nvidia has opted to up the ante on cores with the Tegra 3 rather than move to a new architecture. Tegra 3 still uses Cortex-A9, but runs four of them on a single die. As manufacturers like Qualcomm begin showing off upcoming chips, Nvidia might have cause to worry. Snapdragon S4 is benchmarking very well with its two Krait cores, and Qualcomm has plans for even more powerful versions of the chip in the coming months.

Krait is a processing core that is unique to Qualcomm, and is designed to compete with Cortex-A15, not the older A9 used in Tegra 3. If this is indicative of what manufacturers can do with A15, OMAP5 and the revamped Exynos could also prove problematic for Nvidia’s Tegra 3.

Another concern for Nvidia’s current Tegra is that it has traditionally not played nicely with LTE modems. In early 2011, that was not terribly important as LTE was just beginning to roll out in the US market. In 2012, consumers are starting to expect LTE in devices, and Tegra still lacks an integrated LTE radio.

It is telling that HTC’s US variant of the One X will have a Snapdragon S4 standing in for the Tegra 3 precisely because of the LTE support. Asus is also using Snapdragon S4 in the 3G/4G version of the new Transformer Infinity 700 tablet. This puts Tegra 3 at a disadvantage when compared to Snapdragon S4, and possibly future SoCs.

Nvidia has recently put out its first SoC with an integrated radio chip in the ZTE Mimosa, but this is just a Tegra 2. A version of the Tegra 4 codenamed “Grey” is expected in early 2013 that will have a built-in 4G LTE Icera modem along with Cortex-A15 cores. Can Nvidia really wait a year to fully answer the SoCs launching this Spring? Nvidia is standing behind Tegra to the end, so the company is sure to give it a good shot.

Read more at VR-Zone

Source

Why Qualcomm’s Snapdragon S4 has the competition on the defensive

After months of teasing and hinting, Qualcomm has finally taken the wraps off of its new Snapdragon S4 mobile system-on-a-chip (SoC). The MSM8960, featuring the Krait core, is the first ARM chip in the company’s stable of next generation mobile parts, and it’s got some official benchmark numbers that are sure to impress. Perhaps most importantly, the new Snapdragon is pulling off these striking feats of computational might with just two cores as other manufacturers are working toward four.

Qualcomm, unlike most ARM chip makers, does not license the architecture for its application processors directly from ARM Holdings. Instead, it designs its own cores in-house and simply licenses the ARM instruction set. At the most basic level, Qualcomm’s Krait core is more advanced than Cortex-A9, the current standard from ARM, and it may even best the next-generation Cortex-A15 core.

Snapdragon S4 is going to be significant the next time you’re shopping for a device for several reasons, not least among them the raw speed. With two 1.5GHz Krait cores, Qualcomm’s new part was able to thoroughly thrash dual-core ARM chips in most tests, and even beat Nvidia’s Tegra 3 quad-core SoC in benchmarks that don’t rely heavily on multi-core optimization. A device running on this new breed of Snapdragon will have much more power to play with and lower power consumption thanks to a 28nm manufacturing process. Other ARM SoCs are still at 40nm.

Snapdragon S4 is also going to support a wide array of hardware on the chip itself. It has always been Qualcomm’s policy to integrate the cellular modem with the SoC, whereas other manufacturers rely on external components for that. The S4 will support on-die LTE, HSPA+, and CDMA radios — making it the first ever SoC to integrate LTE radio, and also the first 28nm LTE part. That meanslonger lasting, thinner, lighter devices.

Video and gaming performance might be the one stumbling block for the first run of S4 chips. Qualcomm opted not to wait for the Adreno 3xx GPUs to be ready for use, so the MSM8960 uses the Adreno 225, which is close in performance to other GPUs currently on the market. The 225 does add dual-channel video memory and a unified shader architecture, but the real magic will come when future S4 chips ship with Adreno 305.

Real world battery life claims will have to wait to be settled as the development platform Qualcomm is handing out is not tuned for battery performance. Each manufacturer thinks it’s on the right track, though. Snapdragons are capable of dynamically altering the clock speed of each core individually to respond to system demands. Qualcomm says this leads to better battery life in most situations. Nvidia says its system of switching cores completely off when not needed, and running a low-power companion core is better.

If anything is made clear by these first Snapdragon S4 tests, it’s that you should not only be concerned with the number of cores. That makes little more sense than the gigahertz race years ago. A chip’s features and architecture can make the user experience better regardless of the number of cores. Nvidia might have thought getting to quad-core first assured it victory, but Qualcomm’s Krait core is more similar to the next generation Cortex-A15 than it is to Tegra’s A9. TI and Samsung are expected to launch dual-core Cortex-A15 SoCs this year, but Qualcomm is also planning to ramp up to a quad-core Krait this year.

The first dual-core Snapdragon S4 devices will be debuted at Mobile World Congress in just a few days, among them the rumored Asus PadFone. You will be able to get phones and tablets running the S4 in your hot little hands in the first half of 2012.

Read more at PCMag

Source

How Intel’s Medfield will dismantle ARM

As I watched Intel CEO Paul Otellini at CES 2012 announce that both Motorola and Lenovo would be shipping x86 Android smartphones, just one question echoed around my skull: So what? Reading throughthe details of Medfield, the SoC that that provides the foundation for Intel’s smartphone push, the same question began to reverberate.

You see, Medfield isn’t head and shoulders above its Cortex-A9 and -A15 competitors. The 1.6GHz Saltwell x86 core is better in some areas — mostly-single-threaded tasks like JavaScript execution and website rendering are stand-out — but in other areas like audio and video playback, where ARM chips have dedicated hardware codecs, the difference is negligible. Intel was also careful to not make any direct comparisons between Medfield’s power consumption and the latest chips from Qualcomm, TI, Nvidia, Samsung, and Apple. If the power consumption figures were better than ARM, you can be sure that Otellini would’ve mentioned it. Still, the figures Intel gave for 3G talk time, multimedia playback, and standby were in line with devices like the Galaxy S II and iPhone 4S.

So far, so good — but to actually answer the question of why Medfield matters we have to zoom out from the technical details and look at the bigger picture. The biggest announcement yesterday wasn’t Medfield — it was that Motorola has signed an extensive “multi-year, multi-device strategic partnership” with Intel to produce Android-powered smartphones and tablets. We might not know the exact details of how Intel’s x86 smartphone platform compares with ARM, but Motorola does. For months, Motorola will have been privy to the exact specs of Medfield, including the all-important power consumption figures. For months, Motorola will have been studying the architecture of Medfield, analyzing its pros and cons, building prototypes, and seeing how it compares to the ARM SoCs it currently uses. In short, if Motorola didn’t think that Medfield was competitive, it wouldn’t have agreed to make a bunch of Medfield-powered phones and tablets.

In six months, then, the smartphone industry will go from being completely dominated by a single architecture to a competitive, two-horse race.

To understand how significant this is, you need to understand how constrained smartphone and tablet makers currently are. Except for Qualcomm, which makes its own ARM-compatible cores, every other ARM chip on the market uses reference CPU cores (Cortex-A9, -A15, -A7) from ARM Holdings, a fabless UK company that licenses out chip designs. TI, Samsung, and Nvidia simply take these cores and dress them up differently to produce OMAP, Exynos, and Tegra. Actual performance and power consumption varies very little between chips, with the only real difference being additional features like built-in LTE radio, or a faster GPU — or in the case of Apple’s A4/5/6 SoCs, the removal of a few unnecessary features to reduce its power footprint.

With a bona fide x86 smartphone SoC on the market, everything changes. Competition will drive prices down. If you thought smartphones were developing quickly, be prepared for even more new features from both ARM and Intel as they strive to differentiate. Consumers, for the first time ever, will actually have a choice — just like AMD and Intel on the desktop, you will be able to pick a smartphone or tablet with a CPU that best suits your needs.

Most importantly, though, ARM must be painfully aware that Intel has matched the latest and greatest Cortex-A9 with its first ever attempt at a real SoC, and at 32nm no less. While competition from Samsung and Qualcomm are pushing out with limited supplies of 32/28nm chips, Intel has already announced that it will be shrinking its ultra-low-power Atom chips to 22nm in 2013 and 14nm in 2014. Say what you like about ARM chips being inherently more efficient, but the shift to 22nm and 14nm will mean that Intel’s ULV chips simply cannot be beaten. ARM can innovate until its blue in the face, but at the end of the day it will be impossible to compete with Intel’s far superior manufacturing process — and heck, on top of all that, it’s not like Intel doesn’t know how to make a good CPU.

In short, ARM & Co. are in a very unenviable position. Intel now has a beachhead and its terrifying, industry-clobbering, 60%-gross-margin wehrmacht will surely follow. You might only see a few Medfield-powered scouts in 2012, but by the time the 22nm Silvermont tanks roll around in 2013 and x86 is better than ARM across the board, expect to see dozens of Santa Clara design wins. Unless TSMC, IBM, GloFo, or Samsung uncover some kind of ancient scroll that details the magic of 14nm, come 2014 and the Airmont core, I wouldn’t be surprised if Intel completely dominates the Android smartphone market. Following the Motorola partnership, I would also be very surprised if Microsoft and Apple aren’t working on x86 versions of their mobile operating systems, too.

Source

Intel Medfield 32nm Atom SoC power consumption, specs, and benchmarks leak

Specifications and benchmarks of Intel’s 32nm Medfield platform — Chipzilla’s latest iteration of Atom and first real smartphone- and tablet-oriented SoC — have leaked, and if I held stock in an ARM-based company like Qualcomm, Nvidia, or Samsung, I’d be a little jittery right now.

According to the leak, the Medfield Tablet Platform (i.e. the base specs of the SoC) consists of a 1.6GHz CPU, 1GB of DDR2 RAM, WiFi, Bluetooth, and FM radios, and some kind of GPU. The smartphone variant will probably be clocked slower — and also there’s no mention of whether a GSM/LTE radio will be baked into the chip.

Benchmark-wise, according to VR-Zone, this 1.6GHz Medfield chip scores 10,500 in CaffeineMark 3, a Java-based cross-platform benchmark. Nvidia’s Tegra 2, by comparison, scores just 7,500; Qualcomm’s Snapdragon MSM8260 scores 8,000; and Samsung’s Exynos (which one?) scores 8,500. These scores might sound impressive, but it’s important to note that Nvidia, Qualcomm, and Samsung’s SoCs have all been out for a year — and the chip that’s most likely to contend with Medfield in the short term, Tegra 3, doesn’t have any CaffeineMark data yet. Benchmark data is useless in the absence of real-world, hands-on testing, too; will Intel’s x86 version of Android be as optimized as the ARM version? We don’t know anything about the GPU, either.

The leaked specs continue with some very tenacious power consumption figures: The Medfield Tablet Platform currently has a 2.6-watt TDP when idling, which peaks up to 3.6W when playing 720p video. The final chips, which ship early next year, aim to cut this down to 2W and 2.6W respectively. This is in-line with the latest ARM chips, though again, we’ll need to get our hands on some production silicon to see how Medfield really performs.

Suffice it to say, if the x86 Medfield really can outperform its strict, utilitarian ARM comrades, and if Intel has polished x86 Ice Cream Sandwich until it shines, 2012 might finally be the year that the ladies and gentlemen of Santa Clara, California win some of that incredibly juicy mobile computing market share.

Source

HDD Pricewatch: Manufacturers slash warranties, 1TB drive deal

The flood waters are receding and manufacturers are bringing factories back online, but the major HDD distributors haven’t issued much new guidance as far as when they expect to be fully operational again. The big news this past week came from Intel, which lowered its guidance on Q4 results by a billion dollars as a result of the HDD shortage.

AMD, however, claims to see no such problem. Speaking to MarketWatch, AMD CEO Rory Read stated that HDD supplies were “going pretty well” and that “in 1Q and 2Q, maybe you see some manifestations [of shortages,] but I wouldn’t bet against the supply chain. They’re very resilient.” Investment group Nomura also claims that Intel’s problems are bigger than the hard drive shortage, writing “we think weak sell-through is also contributing to the $1 billion shortfall. We see softness in China, continued demand for ARM-based more power-efficient devices, and low volumes for ultrabooks.”

AMD’s lack of trouble (assuming Read was being honest) may have more to do with the preparedness of the company’s primary vendors than any obfuscation on Intel’s part. As for Nomura’s claims, Intel’s ultrabook push is still in its infancy, the company has set an aggressive target for next year’s sales, but has made no claims about 2011. ARM chips aren’t replacing Intel chips in notebooks, and tablets aren’t eating into notebook sales.

So what is going on in the HDD market? Not much, though there’s a major short-term caveat to that statement. Here’s what prices look like.

The Samsung Spinpoint F3 (in red above) is currently $149 at Newegg, but the site is offering $40 offwith the use of promotion code EMCJHJE29. That brings the 1TB down to $109, and while that’s still nearly twice the price such drives were going for this past fall, it’s much cheaper than any other option we’ve seen in months. This deal expires on December 21, so if you plan to use it you’d best do so quickly.

Aside from the Spinpoint deal, prices have generally held steady. The 2TB Caviar Black is back down to $249 after rising as high as $279 just before Thanksgiving. Seagate’s 1TB Barracuda has fallen to $129, while the other drives haven’t budged. The Hitachi Touro Desk Pro saw a modest price increase up to $129, but the external USB 3.0 2TB drive remains a much better deal than any of its internal counterparts.

We’ve supplemented our previous chart with a graph to show how current prices compare with historical levels. The VelociRaptor 600GB and Hitachi Touro Desk Pro are the only two drives whose prices aren’t running at least 50% higher than back in September. At present, analysts continue to predict that drive shipments will impact prices well into 2012 with levels only slowly returning to pre-flood points.

Seagate, Western Digital slash drive warranties

Both Seagate and WD win this year’s “Suspicious Timing” award for drastically slashing HDD warranties in the wake of the Thailand floods. In WD’s case, there’s at least an obvious reason: The company has stated it expects to take at least a $275 million hit against its Q4 results as a result of the flooding. Seagate, however, is the manufacturer least affected by the floods and the one expected to see the greatest revenue gain as a result of the shortages.

There’s also a difference in scope. Western Digital cut warranties on the lower end of its consumer products — Caviar Blue, Green, and Scorpio Blue drives will carry two-year warranties as of January 2, down from three years previously. Warranty lengths on Caviar/Scorpio Black drives and VelociRaptor drives remains at five years.

Seagate, on the other hand, is cutting warranties on its Constellation and Momentus XT families (down to three years, from five). The real slash is aimed at the Barracuda family, where the warranty period will now be just one year (down from five) as of December 31.

Both manufacturers have cut warranties before and hid behind vague explanations when they did it; Seagate currently claims that this move is being made “to be more consistent with those commonly applied throughout the consumer electronics and technology industries.”

The reality is simpler: This is a decision to protect profit margins by cutting down on RMA costs. In Western Digital’s case, it’s at least somewhat understandable. Seagate is further maximizing its already excellent market position by cutting product costs. Both companies deny that the HDD floods have anything to do with this decision.

There’s an alternative, somewhat darker explanation for this behavior. It’s possible that Seagate and WD have slashed warranties because they’re worried about the reliability of the drive components they’re currently using. HDD component manufacturers were also swamped by the flooding; the drive motor company Nidec was particularly affected.

These circumstances created an enormous gray market for HDDs and their constituent parts. It’s absolutely possible that the major drive companies have cut warranties because they’re not as confident in the reliability of their hardware as they would be under normal circumstances.

Source