Stanford researchers break million-core supercomputing barrier

In a world where quad-core computers are currently the norm, and eight-core processors available for purchase are somewhat rare, computers with more cores might a little unimaginable. A team at Stanford has dwarfed that rare eight-core processor by using a supercomputer that employs over one million cores to solve a complex problem.

The supercomputer, IBM’s Blue Gene/Q Sequoia, is located at Lawrence Livermore National Laboratory (LLNL), and sports 1,572,864 cores, not to forget its 1.6 petabytes of memory — a far cry from the MacBook Pro’s Intel Core i7 quad-core processors. Sequoia isn’t used for multi-playing a lot of MMO alts (unfortunately), as a team at Stanford Engineering’s Center for Turbulence Research (CTR) used the supercomputer’s massive amount of processors to prove that million-core fluid dynamics simulations are possible. These kinds of simulations can help researchers discover how to build quieter aircraft engines.

The computational data gathered using the supercomputer is vital in helping researchers cut down on aircraft noise, as researchers can’t exactly stick their heads inside a jet engine to see what’s going on in there while it’s in use. Director of the Center for Turbulence Research, Parviz Moin, noted that thanks to supercomputers with an enormous amount of cores, “only recently… have engineers been able to model jet engines and the noise they produce with accuracy and speed.”

In terms of how the supercomputer performs the computations, it’s similar to any standard computer with multiple cores: Each piece of the complex math is broken down, delivered to various cores, and computed. The more cores the computer has available, the more math can be simultaneously processed. Interestingly, when a computer reaches a certain number of cores — previously set at around one million — something of a processing bottleneck would manifest. Too many cores pumping out information would create a blockage, so to speak. Led by Joseph Nichols, researcher at the CTR, a team was able to refine the processes and code — dubbed CharLES — enough to bypass the bottleneck (and whatever other problems may have arisen), and put those 1,572,864 cores to worthwhile computational use. (See: Exascale supercomputer hardware is easy – it’s the software that’s holding us back)

To put the supercomputer’s massive amount of cores into perspective, the $999 MacBook Air comes with a dual-core processor. Newegg only has five eight-core processors, six six-cores, and 38 quad-cores currently available for purchase.

Along with aircraft noise simulations, the CharLES code is being used to spruce up other areas of aircraft, such as to simulate turbulent flow over a wing, and propulsion systems used in hypersonic flight. Basically, without the supercomputer’s ability to run such staggeringly complex simulations, researchers couldn’t really examine and test certain complicated theories. One day, perhaps this research will help people living near an airport not get woken up by the constant roar of engines overnight.

Source

GE develops ultra-thin, almost-silent cooler for next-gen laptops and tablets

General Electric has unveiled what seems to be the thinnest, high-performance cooler for the next-generation of ultra-thin tablets and laptops (pictured above). While this cooler obviously allows for slimmer designs (or more space for other components), it also uses just half the power of a comparable fan, granting a significant boost to battery life. Oh, it’s almost silent, too.

The technology behind GE’s cooler is called DCJ — Dual Piezoelectric Cooling Jets. DCJ basically acts as a miniature pair of bellows: Expanding to suck in cool air, and then contracting to expel hot air. GE originally invented DCJ to help cool commercial jet engines, but two years ago it seems someone had the clever idea of miniaturizing the tech for use in computers — and so here we are.

At 2:15 in the video below you can see the DCJ in action, inside a modern ultrabook laptop. The rest of the video is a bit fluffy, featuring lots of dramatic lighting and excessive use of protective goggles.

As you can see above, GE’s cooler is roughly the size and thickness of a credit card, but the press release states that the complete cooling solution (presumably including a heatsink/pipe) is 4mm. This is apparently 50% thinner than existing fan-based solutions, and obviously rather significant as we move towards tablets and laptops that are sub-8mm.

Perhaps most importantly, though, according to GE VP Chris Giovanniello, “DCJ can be made so quiet that users won’t even know it’s running.” This is partly because the tech is fundamentally different from a fan — there’s no blade whizzing through the air at thousands of RPM, and thus no buzzing or vibrations — but it’s also because DCJ supports verylocalized cooling. Instead of a complex heat pipe and fan assembly, GE suggests that you might instead have a bunch of smaller, more efficient DCJs directly attached to components that need cooling.

Moving forward, GE has already licensed DCJ tech to Fujikura, a Japanese thermal management company. GE is also providing OEMs with DCJ kits, so that they can test the technology out in next-generation tablets and laptops. Realistically, we should see DCJ cooling solutions sometime in the next couple of years.

Customizable gamepad makes for the perfect introduction to Arduino

PC gamers love their mice and keyboards, but some titles just work better with a controller. Anyone who has ever used an emulator to play games designed for consoles can attest to that. Many different USB gamepads are available, but a new Arduino-based gamepad might just persuade you to build your own, at least if you’re electronically inclined. The Arduino Esplora is a brand new do-it-yourself gamepad, and it’s a perfect my-first-Arduino project because it doesn’t require manipulating a breadboard or soldering.

The Esplora includes an analog joystick and four face buttons for traditional input, but also sports a handful of sensors: a linear potentiometer, microphone, light sensor, temperature sensor, and accelerometer. It also features a square-wave buzzer, an RGB LED, twoTinkerKit inputs/outputs, and a TFT display connector.

The Esplora circuit board is 6.5-inches by 2.4-inches, and has four screw holes so you can build a case for your controller or even mount it to a surface like an arcade cabinet. It connects to your computer through USB 2.0, and can be completely custom-programmed by using the Arduino software. It even has libraries available to take advantage of the sensors without having to write everything from scratch.

This gamepad will eventually be available for purchase through the Arduino store, but Radioshack should have it as well. It’s not online quite yet, but the Arduino blog promises it will be out soon. The official price isn’t posted on its site, but it is available through press releases. It will apparently be available online for 41.90 euros, but the retail version will cost 44.90 euros. Currency conversion varies, but we’re looking at less than $60. Until it’s online, your best bet is to visit your local Radioshack in hopes of finding one right now. Obviously, other gamepads are available in a finished state for less money, but there’s not much entertainment or educational value in that, right?

Arduino-based kits and devices continue to impress. From aiding hotel thieves to computing in space, this tiny and inexpensive platform has made electrical tinkering exciting again. It’s not only novices that are taking advantage of it to learn the ropes, but some of the brightest electrical engineers are exploiting the capabilities of the Arduino platform to make working solutions quickly and cheaply. If you’ve been eying Arduino for a while, but you’ve been too worried about the difficulty, this is the project for you.

Read up about the platform, grab the Esplora, and start hacking away.

Source

Wii U launch: Software and battery life woes, and more hardware details emerge

This is it, ladies and gentlemen: Following this weekend’s release of Nintendo’s Wii U console, the eighth generation of video game consoles is finally here. It’s been a long time coming — seven years, to be exact; the longest generational gap ever in the history of games consoles — but hopefully the wait will be worth it. With the Wii U selling out across the US, and games retailers seeing record numbers of pre-orders, early reports are generally good, with a handful of negative caveats that mostly involve the Wii U’s tablet Gamepad and a beefy, hours-long firmware update.

CPU, GPU, and RAM

With video game consoles, their exact hardware specifications are always a bit mystical. Prior to their actual launch, console makers rarely tell us more than the absolute minimum, such as “it will be backwards compatible” or “it will be much faster than its predecessor.” As soon as the console goes on sale, though, we can finally tear the thing open and find out what’s actually under the hood — which is exactly what Anandtech did.

We already knew that Nintendo had opted for an AMD GPU/IBM CPU multi-chip model (MCM) in the Wii U, but we now have the exact die sizes: 5.2mm x 6.3mm (32.76mm²) for the CPU, and 12.2mm x 12.7mm (156.21mm²) for the GPU. There is also a third, absolutely tiny (1.79mm x 1.48mm) die on the MCM, which Anandtech speculates is off-chip memory.

Wii U logic board, showing the IBM/AMD multi-chip module, and Hynix RAM

As for the exact specifications of the CPU and GPU, it’s still a bit of a guessing game. Froma previous teardown by Nintendo, we know that the Radeon-derived GPU has some on-chip memory, no doubt increasing the die size considerably. If we start with a 104mm² Redwood core found in the Radeon HD 5670, then pile on some megs of eDRAM, we approach the 156.21mm² die size of the Wii U’s GPU. This assumes that a 40nm process is being used (and considering the relative cost/scarcity of 28nm production, it’s almost certainly 40nm). A Radeon HD 5670 with on-die memory would be a fairly beastly GPU, far outstripping the Xbox 360 and PlayStation 3 — but hey, those consoles are seven years old now.

On the other hand, the tiny, multi-core Power-based CPU isn’t going to win any performance prizes. Beyond the fact that it’s manufactured on IBM’s 45nm SOI process and has on-die eDRAM, we know very little about the CPU’s exact specifications. Due to the fact that the Wii U is backwards compatible with the Wii (i.e. they have the same instruction set architecture (ISA)), the persistent rumor is that the Wii U simply takes the Wii’s Broadway core, and multiplies it by three. Throw in a few tweaks and a few process node improvements (from 90nm down to 45), and the Wii U CPU should be fairly competent. Ultimately, though, as we all know, the CPU clock speed and core count are ultimately just footnotes: For the Wii U’s new and shiny 1080p HD output, that beefy GPU is far more important.

Rounding out the hardware spec, Wii U teardowns have revealed that there’s a Samsung NAND flash chip for user storage (32GB or 8GB), a Toshiba NAND chip for non-user storage, and 2GB of Hynix DDR3-1600 RAM on a 16-bit interface surrounding the MCM, providing peak memory bandwidth of 12.8GB/s. This is twice as fast as the Wii, but for 2012 it’s pretty slow — though, with the on-die eDRAM, the DDR3 bus shouldn’t see too much usage.

Gamepad

By all accounts, the Wii U Gamepad is the centerpiece of Nintendo’s new console — and yet, curiously, every tech publisher (including ET) has failed to report on a massively vital caveat: battery life. It turns out that the Wii U Gamepad, with its large, 6.2-inch resistive touchscreen display, only has a 5.6 watt-hour battery (1500mAh @ 3.7V). Early reports suggest that the Gamepad only lasts 3-4 hours between charges — okay for casual gameplay, but it may put a serious dent in prolonged gaming sessions if you have to stop and recharge. The good news is that the battery is replaceable — and we’re sure it’s only a matter of weeks until larger, third-party batteries are made available.

As far as the Gamepad’s wireless capabilities are concerned, reviews are almost universally positive. In essence, the Gamepad is a second display for the Wii U. It can either be the sole display for the Wii U (so you can play from bed while someone else uses the TV), a second mirrored display, or a second supplementary display for gameplay gimics, menus, and so on. This wireless functionality is provided via a high-speed 802.11n WiFi connection (possibly using some variant of Miracast), which should work flawlessly up to 25 feet (7.5m) away. Some reviews are saying that it will work from even greater distances, which is neat.

Curiously, Anandtech’s teardown reveals two WiFi antennae and controllers — one for network connectivity, and one dedicated to the Gamepad, presumably to ensure that the Gamepad always works flawlessly. Early reports suggest that Nintendo’s implementation is spot on, with no sign of lag or any other weird artifacts.

Software, power consumption, firmware updates, and other errata

Rounding out the Wii U launch, we now have data surrounding the console’s software, power consumption, and a few other tidbits. Anandtech connected a power meter and found that the Wii U is rather efficient, drawing just 31.2 watts at the main menu, and 33 watts while playing Super Mario U. This suggests that IBM and AMD have performed some serious hocus-pocus to pare down their processor power envelopes.

In a move that will upset Norwegian apologists and beardies everywhere, it seems like Nintendo and Opera have finally gone their separate ways. Like almost every smartphone and tablet on the market, the Wii U features a WebKit-based browser — in this case, made by NetFront, the same company that provides a browser for the PS Vita, PS3, and Nintendo 3DS. As you’d expect from WebKit, this means that the Wii U’s browser has surprisingly good HTML5 support and JavaScript performance — though early reports do suggest that, for anything other than very simple internet usage, the browser is just too basic.

Finally, despite selling out in the US, not many Americans will have actually played on their Wii U due to a monstrous (5GB!) firmware update, which reportedly takes hours to install— and, if you lose power during that time, which happened to at least one user, you’ll have a bricked console on your hands. The firmware update adds online functionality (why isn’t this included by default?), support for Netflix and Hulu, and other important features.

Overall, the Wii U seems to be off to a good start — but it has a major advantage in that it’s the only eighth-generation console on the market. It remains to be seen if the Wii U can stand against the Xbox 720 and PS4, both of which will have significantly more power under the hood. As with the Wii sensor bar, the Wii U is relying on the dual-display Gamepad gimmick to sell consoles, which is enough for now — but just as with Kinect and Move, Microsoft and Sony will probably have their own, multiple-display setups. The Wii U will undoubtedly do well in the short term, if only off the back of first-party titles such as Mario and Zelda, but can Nintendo drum up enough support from consumers and third-party developers to ensure another seven years of success? We’ll see.

Now read: Can Sony and Nintendo’s next-gen consoles escape Japan’s stifling cultural baggage?

[Image credit: Anandtech, Louis Lamarche]

Source

Samsung’s Ativ 500T: Not the Windows 8 tablet you’ve been waiting for

Samsung’s Ativ Smart PC 500T is a $749 tablet meant to bridge the gap between the notebook industry and the Apple-dominated tablet space. The manufacturer describes the 500T as “a fully functioning PC. The Samsung Ativ Smart PC 500T runs the programs you need on a Windows 8 operating system in a sleek, lightweight form.”

Samsung has built an enormous smartphone business around delivering high-quality products at attractive price points; the Galaxy series is the beautiful result of close collaboration between engineers, designers, and the Holy Beancounters. The Smart PC 500T, in contrast, feels more like the product of a three-way turf war — and the accountants won.

The 500T’s base specs look great. It’s built on Intel’s Clover Trail platform and features a pair of 1.8GHz Atom cores with Hyper-Threading enabled. The tablet includes 2GB of RAM, lists 64GB of storage, and an 11.6-inch display at 1366×768. There’s no 3G or LTE support, but WiFi and BlueTooth 4.0 are both included.

External hardware

Note: The Samsung Ativ 500T we were sent for review lacked the keyboard that normally ships with the device. When we compare against Surface, we’re omitting comparisons to the Touch Cover. The Ativ 500T’s weak points are unrelated to its dock.

The plastic shell Samsung uses for the 500T feels sturdy, with just the right touch of flexibility. This may or may not be an accurate impression; grip the 500T too hard at the back, and the front LCD will distort in the same places. The Samsung logo at front and back is glued on and there’s a pair of thankfully unobtrusive stickers on the back. The system feels a little too cheap. The stickers aren’t visible in the stock image below, but you can see the included stylus in its slot. This is a welcome addition for navigating in Desktop mode, and Samsung gets kudos for including it.

The bottom of the tablet has the attach points for the dock as well as a magnetic connector reminiscent of those Microsoft uses for the Surface.

Port access and button placement is very hit-and-miss. Here’s the left-hand side of the 500T:

The left-hand side of the device is clean; the volume controls and mini-HDMI output are easily accessible. Mini-HDMI is esoteric enough that we wish the company had included a full-size HDMI to mini-HDMI adapter, but that’s a fairly minor point.

The top of the device is very busy. From the left, there’s the SD card slots, the USB 2 port, the physical auto-rotate on/off button, and the power switch. We’re thrilled to see a physical option for auto-rotate, but the device’s button placement makes it impossible to hold it in portrait mode from the left-hand side. The power button is easy to hit by mistake, a fact that’s particularly annoying given the network problems we experienced. The fold-away tabs make the device look cheap, especially if you’ve got multiple peripherals hooked up at the same time. There’s no way to detach them without breaking them off.

Weight and size

The 11.6-inch screen is clear and bright, but 11.6 inches is, in my opinion, the tipping point for 16:9 tablets. Even before I had the opportunity to test Microsoft’s Surface, I thought the Samsung 500T was too large and clumsy for a handheld device. Windows 8′s split keyboard helps, but doesn’t completely solve the problem.

I compared notes with Sebastian, who has an 11.6-inch Series 7 (Core i5) tablet from Samsung and substantially larger hands than me. He agreed; 11.6-inch tablets aren’t that great for on-screen keyboard work. At 1.65lbs (0.75kg) sans dock, the 500T isn’t heavy, but the 16:9 form factor make it difficult to hold in just one hand in either portrait or landscape mode. Holding the 500T up at an angle while reading/watching content in bed also aggravated the tendinitis in my right arm. That’t not Samsung’s fault, but if you have carpal tunnel, tendinitis, or a similar problem, you may want to look elsewhere.

Finally, there’s Surface. On a purely aesthetic level, Surface makes Samsung’s 500T look like a cheap knockoff from mainland China. In this case, looks are deceiving — the 500T is faster than its rival — but you’d never think so to look at them. The small difference in screen size (10.6 inches instead of 11.6) and weight (1.5 pounds vs. 1.65) matters far more than you’d think; Surface feels solid and comfortable when held in a single hand. The 500T never does.

Display

The Ativ’s display is one of the device’s strong points. 11.6 inches and 1366×768 is a good fit for basic desktop work as well as browsing in Metro applications. It’s bright and clear, with good color saturation and uniformity.

Each bar should be a individual, easily distinguished color. Good TN panels will only blur a bit at the 30-32 marker. Surface is a bit better than the 500T, but the Samsung panel is quite good. Both tablets performed well in single-color viewing angle tests as well.

We compared it against the Surface’s 10.6-inch screen for general video playback and used Lagom’s LCD tests for specific data points. Surface’s display has a slightly higher PPI (147 vs. 135), and slightly more accurate color reproduction. Both displays were adjusted to the third “Brightness” notch. That’s one of the only changes you can make to either tablet; the driver interfaces are bare-bones by even Intel’s standards. Nvidia’s comprehensive control panel, with its 2D and video playback controls, is completely DOA.

White and black saturation are slightly better on Surface; the 500T’s display is marginally easier to use outdoors. Surface is a bit slower in lag tests; viewing angles for both are excellent. Both panels are glossy. Boo.

Video playback: Pick your problems

File compatibility and format support are a big potential reason to go with an x86 tablet instead of something Windows RT-based. Microsoft’s decision not to support MKV playback natively is still a problem; there’s currently no sure-fire solution available in the Windows Store. The 500T, in contrast, has access to tools like VLC and MPC. Samsung’s decision to include PowerDVD doesmake some sense here, even if we prefer open-source solutions.

Unfortunately, other problems keep the Ativ from snagging what ought to be an easy win.

We measured CPU utilization on Surface and the Samsung 500T with multiple test files. These included:

An Xvid encode of a Walking Dead episode (343MB, 640×360, 1.1Mbps)
An Xvid encode of The Incredibles (2.05GB, 720×304, 2.5Mbps)
An Xvid encode of Shutter Island (1.7GB, 1280×544, 1.73Mbps)

We also tested one movie (Star Trek) at multiple quality settings to measure what the two platforms could and couldn’t handle. We encoded Star Trek at three different quality levels and profiles using Handbrake. These were:

1x H.264 (High@L3.0) (811MB, 720×304, 761 Kbps)
1x H.264 (High@L4.0) (6.77GB, 1920×800, 7.6Mbps)
1x H.264 (High@L4.1) (8.75GB, 1920×800, 9.8MBps)

Here’s how performance broke down in our first three videos:

Overall CPU usage is lower for Tegra, which isn’t too surprising — Hyper-Threaded cores are treated like full cores when Windows tracks CPU usage, but they don’t have a full core’s worth of resources. The Incredibles and The Walking Dead episodes played beautifully across both tablets. Shutter Island, on Surface, was something more akin to Stutter Island, for no reason we could see. Its encoding parameters and bit stream size are well within the range of the other two files, but playback was a mess.

The x86 Smart PC had no such problems. Point to Samsung. Watch what happens, though, when we fire up Star Trek.

Atom can play all three versions of our Star Trek encode, but the last two, only just barely. We had to shut down the Task Manager (keeping it open consumes 4-6% of the CPU’s processing power) and close every other open program. Unfortunately, none of our go-to applications fully support Clover Trail’s SoC; we couldn’t get a reliable read on whether or not the video stream was being properly offloaded to the GPU. Based on these figures, we don’t think it is.

GPU offload problems would also explain the lag in Desktop mode. A number of other reviewers have commented that while Metro is smooth and capable on the Ativ, attempting virtually anything in Desktop turns the system into a lagfest. We ran into the same problem; as soon as the tablet tries to juggle even basic tasks, it starts to skip. We suspect the problem here has more to do with video drivers than Atom’s CPU performance. What we’re seeing in Windows 8 is similar to early nettop problems in Windows 7. When Nvidia launched Ion, one of its major talking points was the way Windows 7′s UI didn’t stutter.

Windows 8 relies on GPU acceleration much more than Windows 7 did; it’s entirely possible that the lag and high CPU utilization during video playback are a result of an unoptimized driver stack.

So does Surface win this round? Not hardly. No MKV support meant I had to re-encode the movie into a format the Windows RT tablet could play. The 32GB tablet’s minuscule available storage meant deleting everything else just to watch a single high-quality film. We suspect the Ativ’s playback capabilities could be substantially improved with better video drivers, but as things stand, you can choose between a tablet with excellent video offload and minimal storage, and a lot of re-encoded data files — or a tablet with slightly more storage, full application compatibility, and problematic high-detail playback.

Playback power consumption

Playback power consumption favors Surface, since the two devices have 30Wh batteries.

Smidgen of storage, wonky WiFi

The Ativ Smart PC 500T has a hat trick of problems that effectively nuke its strong points. We’ve discussed the storage problem at some length, but the bottom line is this: Samsung is advertising this as a 64GB tablet and shipping it with roughly 32GB of usable storage.

That’s not just an advertising problem. The entire benefit of buying an Intel tablet is supposed to be x86 compatibility, but a lot of desktop software isn’t exactly sized to fit on a 32GB device. Office 2013 is 2GB. Start adding browsers, Photoshop, photos and video, and some work applications and poof — you’re out of room.

Cloud services like SkyDrive can be helpful here, but they only work as “extra” storage if you tell them not to synchronize with local folders. Otherwise they keep local mirrors on each device and could actually exacerbate the problem.

Still, cloud storage could help solve this problem, if the WiFi didn’t have problems of its own. The 500T’s signal strength and performance once a connection has been established is excellent. We were able to download files via wireless nearly as quickly as via wired connections, and had no problem using the tablet while several rooms (and an entire floor) away from the router.

But actually connecting to the router? That’s a major problem.

At boot or upon resume, the 500T regularly takes between 90 and 120 seconds to connect. It regularly claims to be connected to the internet, even as web pages refuse to load, as shown below.

This is infuriatingly common

Clicking on “Fix Connection Problems” almost never fixes the problem. Usually, it just makes the router vanish for an indefinite period of time. Rebooting the tablet will restore the router as a detected device, but won’t solve the connection issue. Sometimes Metro apps like News will load but Desktop Internet Explorer won’t. Sometimes it’s the other way around.

Intel confirmed to us that this is an issue, but told us the solution is stuck in Samsung’s QA process, with no targeted release date.

This bug single-handedly kills the idea that a tablet is a computing device you can pick up and use the same way you’d use your smartphone. Simply leaving the tablet on and waiting a few minutes doesn’t work; the user has to actively attempt to use the internet, have it fail, than muddy through enabling and disabling the connection.

Samsung’s update application, meanwhile, is prone to errors. The screenshot below was taken after checking both the Windows 8 Store and Windows Update for available Microsoft patches.

It made me download the update twice, for no apparent reason. I’m still not sure it actually changed anything.

Dodgy Desktop

Finally, there’s the state of Desktop performance. In Metro, the 500T shines. Application load times and overall performance are measurably faster than Surface. This is particularly true in twitch games, like Jetpack Joyride. Surface has a noticeable stutter; the Samsung keeps things smooth.

In Desktop mode, responsiveness and performance are great until you actually try to do something. Even small tasks, like simultaneously playing a video file while moving the mouse, lag noticeably. We couldn’t, for example, keep the Task Manager open while playing either of our high end Star Trek encodes. Attempting to manage both windows at once, even with one of them minimized, was too much for the 500T.

Desktop gaming is also out. While the 500T may be theoretically compatible with x86 games, the SGX545 can’t handle anything recent. Even Torchlight, in 640×480 netbook mode, with all details at their lowest values, was barely able to manage a mid-teens frame rate.

The nearly constant lag turns x86 software compatibility into more of a bullet point technicality than it really ought to be. It’s the sort of situation that might be fixable via driver updates, or might be a symptom of an underlying hardware bottleneck.

Who’s supposed to want this?

All of the companies that dominate the modern tablet industry have content stockpiles and product ecosystems. Companies like Samsung, that lack such options, have little choice but to double down on hardware capability. Compare the specs on the 500Tagainst Samsung’s own Series 3 NP350V5C, both at $749, and try not to wince. The laptop is 5.5lbs. The 500T is 3.28lbs with its dock attached. If you need something lighter, there are over a dozen Core i3/i5 SKUs in the 2lb – 3.9lb segment, including four that skate under the 3lb mark.

That’s a major problem. As much as Samsung might like to position the system as a tablet with an optional dock, it’s ridiculously underpowered compared to the laptops in its price bracket. The instant-on premise is demolished by network problems, the prospect of doing Real Work is nuked by desktop lag, performance stuttering, and a ludicrously small amount of storage.

If all these issues were resolved, this tablet might be worth $549. That’s a $120 premium over the current crop of high-end 10.1-inch (1024×600) Atom netbooks, which typically use the N2600 (1.6GHz, dual-core) and an SGX545 GPU. Then again, those systems still ship with 2-3 USB ports and 320-500GB of storage.

What would I do? Wait. At the very least, wait and see if Samsung, Intel, and Microsoft can resolve the desktop performance lag and the wireless issues. Wait and see if Samsung does anything about available storage, or even acknowledges the problem. The Ativ gets some things right, but far too much of what’s billed as basic functionality doesn’t work or comes with caveats. Some of these issues very much extend to Surface, which is why I’m not waving it around as the alternative must-have. Some of them, like the network problems, don’t.

If the Ativ is representative of the hardware other PC OEMs are shipping, it’s no wonder thatuptake is markedly worse than Windows 7. It’s not that the Ativ 500T is a bad tablet, but it’s eclipsed on every side by better devices at lower price points.

Source

IBM 3390: The world’s largest and most expensive hard drive teardown

If you lived through the ’70s or ’80s, you probably remember a time when technology was big. TVs, with their cathode ray tubes, were monstrous. Amplifiers and hi-fis were heavy, bulky, expensive things. And computers… computers were reallylarge. Case in point: The IBM hard drive pictured above, which weighs in at 38.5kg (85lbs) and stores a grand total of between 1 and 2 gigabytes.

This drive, which originates from around 1989, would’ve been teamed up with a number of other drives and slotted into a IBM 3390 Direct Access Storage Device (DASD) — a floor-to-ceiling server rack. One IBM 3390 model was capable of storing up to six drives, for a total capacity of 22.7GB. A complete IBM 3390 system had a data transfer rate of 4.2MB/sec, with an average seek time of 12.5 milliseconds. The platters probably span at around 2,500-3,000 RPM.

While it’s hard to put an exact price on a single drive, it would’ve cost somewhere in the region of $50,000 to $100,000 in 1989 — or about twice that, in today’s money. That’s around $50,000 per gigabyte — or one million times more expensive than today’s hard disk drives, which are currently priced at around five cents per gig.

With the background out of the way, I give you EEVblog’s teardown of the the hard drive. The video is rather long, so you might want to around the 8-minute mark for the case opening; the 15-minute mark for spindle motor; 20 minutes for the hard drive heads; and 33 minutes to see the head actuators in motion.

Back in the ’80s and ’90s, the IBM 3390 DASD was the pinnacle of reliable, online (as opposed to offline tapes) storage. EEVblog’s hard drive was sourced from a bank, where it would’ve stored recent transactions until they were backed up to tapes. In the ’80s and ’90s it would’ve been quite common to find systems like these in the vaults of banks, and large corporations and institutions.

The back of the IBM 3390 hard drive, showing a power and data connector

It’s worth noting that this drive isn’t big just because it’s old. In 1989, 2.5-, 3.5- and 5.25-inch hard drives already existed — but their capacities were measured in megabytes. The IBM 3390 had a large capacity, but it was primarily designed for reliability — and in the ’80s that meant going large (the platters are 11 inches across in this case). If you watch through the entire teardown, you will see that the hard drive’s only silicon chip is a tiny chip that acts as a signal amplifier. If you look closely, you’ll see a nozzle on the drive’s exterior that was used to pump the enclosure full of pressurized halon, increasing reliability. EEVblog says that these IBM drives were so reliable (and so expensive) that many institutions carried on using them well into the ’90s.

At this point, you should check out our history of computer storage. It’s amazing how hard drives were one of the most recent advances in non-volatile storage, and yet the fundamental technology has remained almost unchanged for 40 or 50 years. If you took the IBM 3390 drive and simply scaled it down a few hundred times, it would look virtually identical to a modern hard drive.

Source

Valve confirms Steam Box coming in 2013, says it will compete with next-gen consoles

Hot on the heels of the surprisingly-well-received Big Picture mode, Valve has confirmed that it will release a “Steam Box” video game console sometime in 2013.

Speaking to Kotaku at the 2012 Video Game Awards this weekend, Gabe Newell confirmed that Valve is working on a living room-friendly PC that runs Steam. Newell says that reception for Steam’s new, TV-oriented Big Picture mode has been “stronger than expected,” and the company will now work to get Steam for Linux out of beta.

Beyond that, Gaben only gave us one other tidbit about the Steam Box, stating that “our hardware will be a very controlled environment.” In other words, the Steam Box will be a console, rather than some kind of PC that you can upgrade. Curiously, Newell also said that he expects other companies to sell living room PCs that will compete with next-gen consoles from Microsoft, Sony, and Nintendo. We’re not entirely sure why 2013 is the year that living room PCs will suddenly compete with game consoles, but do let us know in the comments if you have a theory.

Valve in general, and Gaben in particular, seem to be on some kind of heroic mission to bring Steam to the mass market. Newell was one of the first critics of Windows 8, calling it “a catastrophe for everyone in the PC space.” It’s impossible to say for certain, but it certainly seems that Steam for Linux’s graduation from weekend project to full-blown endeavor neatly matches up with Gabe’s distaste for Windows 8. As far as we can tell, though, this hatred doesn’t stem from the Metro interface; rather, Valve sees the Windows 8 Store as a direct threat to Steam. In much the same way that Microsoft squashed other web browsers with its monopolistic backside, Valve is worried that the Windows Store will snuff out third-party digital distribution platforms.

What isn’t clear, though, is how the Steam Box will actually combat the Windows Store. With Valve’s focus on Steam for Linux, the current assumption is that the Steam Box will run Linux — but then the console won’t be able to play more than a few dozen games. Valve is also working on a Linux port of its Source game engine, which will certainly help matters, but that isn’t going to suddenly make Skyrim or Mass Effect compatible with Linux.

The other option, of course, is that the Steam Box will run Windows 7. This would drive the cost of the console up, but it would mean that you could immediately play Assassin’s Creed III on your TV, with your Valve gamepad.

Ultimately, though, the elephant in the room is the hardware itself. Newell very clearly tells Kotaku that the Steam Box will be a PC, rather than a custom piece of console hardware (such as the Xbox 360 or PS3). In all likelihood, this means the Steam Box will probably just be a small form factor (SFF) x86 PC — and as we know, high-powered SFFs are not cheap ($600+). By pitching it as a living room PC with more capabilities than a conventional video game console, Valve has some leeway on the pricing — but not that much.

Though, having said that, if it looks like a Companion Cube, I’d pay almost anything for a Steam Box.

Update: The wording in the original Kotaku story is rather vague. It seems to indicate that the Steam Box is coming in 2013, but Newell might instead be referring to other living room PCs. Either way, with next-gen consoles due at the end of 2013, it would make sense if the Steam Box has a similar launch window.

Source

Apple acquires TI chip engineers, moves deeper into the world of SoCs

Texas Instruments announced earlier this year that it is getting out of the SoC market, and that sadly means a little under two thousand jobs are being cut. In response, Apple has snatched up dozens of engineers from TI’s Israel-based operation. Not only were these engineers reportedly working on TI’s Open Multimedia Applications Platform (OMAP), but they were also working on chips including WiFi and Bluetooth radios. As Apple continues to move into custom chip designs, this kind of expertise becomes absolutely essential to its core business.

Apple isn’t the only one moving in on TI’s SoC corpse, though. A few months ago, rumors swirled that Amazon was interested in acquiring OMAP technology. As tablets start to become a bigger business for Amazon, it makes perfect sense to move in exactly the same direction as the Cupertino behemoth. Apple went from using stock CPUs and SoCs from other companies to making custom CPUs and whole SoCs itself in just a few generations,thanks to the purchase of PA Semi back in 2008. With part of TI’s resources, Apple will undoubtedly expand its capabilities in the next generations of SoCs. These engineers could possibly even be tasked with making the radios small enough to be crammed onto a single low-power chip in the relatively near future. That would certainly help battery life substantially, and it would give Apple quite an advantage.

Apple’s legal war with Samsung has strained relations between the two companies. Unfortunately for Apple, Samsung fabs the A5 and A6 SoCs, and there aren’t many alternatives because of the large scale needed. Rumors have been floating around that perhaps the Taiwan Semiconductor Manufacturing Company (TSMC) would become Apple’s go-to for SoC production. Now there is even talk about Intel fabricating ARM chips for Apple with the catch attached that Apple would switch the iPad to x86 chips. As it stands, these have just been rumors bandied about. There hasn’t been any hard evidence pointing towards Apple’s future in the SoC space.

Since Apple has a huge war chest and is continuing to acquire expertise, it stands to reason that Apple could even open its own SoC fabs and operate them free of outside dependencies. While it would put Apple in control of its own destiny, and it does have other upsides, it is far from an ideal situation. Apple likes to stay, wisely, in the high-profit world of selling devices. Getting down to the level of component manufacturing isn’t really Apple’s wheelhouse, and has a fair amount of risk associated with it because they don’t have experience in the field. That said, if it can’t work out a satisfactory deal with another company in the coming months and years as contracts end, and new chip designs are ready for production, it might be forced into becoming its own fab whether or not it’s something it wants to do.

The future of iOS device internals is being decided now in meetings, emails, and calls. We’ll soon see if the status quo of highly customized chip designs made by Apple and fabricated by Samsung remains, or if something more interesting is afoot. In any case, it’s just good to see that TI’s talented Israeli engineers won’t be going to waste. Apple continuing to push the envelope with SoCs makes the whole industry better no matter who does the fabrication.

[Image credit: AutomaticDefence]

Source

Is Intel about to become Apple’s ARM SoC foundry?

If an industry analyst is to be believed, Apple may soon move the production of its ARM processors from Samsung to Intel. This dovetails neatly with the news earlier in the year that Intel was moving into the foundry business — but it isn’t quite that simple: The same analyst says that, as part of the deal, Apple will have to transition its iPads away from ARM to an Intel x86 SoC.

This information comes from Doug Freedman, an analyst at RBC Capital. The exact detailsare as follows: Intel would take on the production of the ARM SoCs found in the iPhone (and presumably the iPod); in exchange, Apple would switch to an x86 SoC in the iPad.

According to Freedman, Cupertino is already in talks with Intel, but we should stress that nothing is confirmed. At first blush, you might think this development is completely insane — and to be honest, it really is. There could be a grain of truth, though. We know that there have been growing tensions between Apple and Samsung (which currently manufactures every iPhone/iPad SoC), due to patent warfare and increasing competition in the consumer electronics space. Over the last few months there have been numerous rumors that Apple would move its A-series SoCs from Samsung to another foundry.

The question has always been, though, who would take on the fabrication of Apple’s SoCs? With a volume of around 100 million chips per year, there are only a few possible choices, once you remove Samsung from the equation: TSMC, GlobalFoundries, and Intel. Most analysts and pundits originally assumed that Apple would move to TSMC, but there are lingering questions about whether TSMC has the spare capacity to continue producing chips for Qualcomm, Nvidia, and AMD, while taking on another 100+ million from Apple. GlobalFoundries is a possibility, especially if Apple stumps up some capital investment for additional capacity at an advanced process node.

And then there’s Intel, with its fancy, industry-leading, 3D FinFET 22nm process. Earlier in the year Intel announced that it had begun producing chips for Achronix and Tabula, two small-time FPGA makers — the first time that Intel has acted as a foundry (manufacturing chips for third parties). Apple is a whole different ball game, though. With Achronix or Tabula we are talking about tens of thousands of chips — with Apple, it would be hundreds of millions. Would Intel really take on Apple’s SoC business? Intel probably has the capacity, and through the FPGAs it has shown that it has the technologies and processes to act as a foundry.

The main thing to remember, though, is that switching from Samsung to Intel (or TSMC or GloFo) would be a massive move for Apple. Changing from one process to another (say, from 45nm to 32nm) can take months to move from design, to tape-out, to actual production — but moving from one foundry to another is a gargantuan task. Apple’s chip designers have intimate knowledge of Samsung’s foundry work flow: The limitations of the HKMG process, the caveats of the software, and other gotchas that can only be uncovered through experience. Depending on Intel’s foundry work flow, such a switch could take a very long time to execute.

If Apple does move, though, there’s a lot of money at stake. Freedman estimates that Apple’s chip business to be worth $2 billion per year (415,000 wafers at $5k each) — a sizable sum, though still only a fraction of Intel’s $55 billion annual revenue.

For me, though, the biggest giveaway that Freedman’s sources are incorrect is that tidbit about Intel forcing Apple to move its iPad to x86. It certainly makes sense from Intel’s perspective, which is desperately trying to break into the mobile market — now with Clover Trail, which is fairly exciting, and next year with Bay Trail, which will probably blow the doors off everything else on the market. I’m not sure if I can picture Apple splitting its iOSplatform into two different architectures, though, just to get into bed with Intel. Maybe that’s the only option, though, if Apple really wants to drop Samsung and TSMC doesn’t have the capacity.

Source

Self-healing, self-heating flash memory survives more than 100 million cycles

Macronix, one of the world’s largest producers of flash memory, has produced a new kind of flash memory that can survive more than 100 million program/erase (PE) cycles — most likely long enough to persist until the end of human civilization. By comparison, the NAND cells found in conventional flash memory — as in commercial SSDs — generally have a lifespan of just a few thousand PE cycles.

For such a huge advance you would expect an equally vast technological leap — but in this instance, that’s certainly not the case. Macronix just adds a bit of heat — literally, each of Macronix’s new memory cells contains a heating element that can deliver a jolt of 800C (1472F) heat to the cell, healing it and preventing wear-out. Furthermore, 100 million PE cycles is a low-ball estimate: In reality, Macronix’s new flash might survive billions of cycles — but it would take so long to test that the company doesn’t yet know.

Why does heat fix a flash memory cell? It’s all down to the physical structure. NAND flash is constructed from floating-gate transistors, which are exactly what they sound like. Basically, the control gate (which controls the flow of electricity across the transistor) floats above an additional oxide layer. In effect, the bit value of the cell is stored in this floating gate. To trigger the gate — to change the bit value — a certain threshold of current is required to jump through the oxide layer. Over time, this oxide layer degrades, eventually causing the cell to fail.

By applying heat, this oxide layer can be annealed, returning it to its base state. Macronix has known about this annealing effect for years — but historically its testing involved putting a bunch of memory chips in an oven and baking at 250C (482F) for a few hours. Obviously this wasn’t the best solution for consumer electronics, and so a new method had to be devised. Ta’da: Macronix’s NAND memory cell with built-in heat plates (pictured right).

In these new heat-assisted cells, the heating plates flash on for a few milliseconds and then turn off. This process does consume extra power, but the heating only has to be done infrequently — so it could be done while plugged into the wall, over night. As a curious side effect, Macronix also reports that hot memory allows for faster erasing — though the company doesn’t seem to know why this is. “Further down, this may evolve into a ‘thermally assisted’ mode of operation that gives both better performance — such as the faster erasing — and better endurance flash memory,” says Hang-Ting Lue of Macronix to IEEE Spectrum.

As you can imagine, this heat-assisted memory cell’s structure and footprint is quite different from conventional NAND cells. Macronix doesn’t give us a timeline for commercial availability, which is a hint that current NAND flash processes probably aren’t capable of producing the new memory. Given the world’s love affair with mobile devices and high-speed flash storage, though, it’s almost certainly a matter of when Macronix’s heat-assisted memory will come to market, not if.

Now read: Cache of the titans: SSD storage accelerators from Intel and Corsair face off

[Image credit: IEEE]

Source