ARM Cortex-A15 explained: Intel’s Atom is down, but not out

Share This Article

2
inShare

For the past two years, smartphone advances and the tablet market have been driven by ARM’s Cortex-A9. The diminutive mobile processor has proven to be extremely capable, but it’s also getting a bit long in the tooth. Custom chips from Qualcomm and Apple have taken the wind out of A9-based hardware in recent months, which makes the Cortex-A15′s debut downright timely.

The best way to understand the features and capabilities of the Cortex-A15 is to keep the chip’s heritage firmly in mind. Each iteration of the Cortex-A family has increased CPU execution efficiency and performance, but always with the goal of remaining in a low-power envelope. The Cortex-A15 is meant to continue this tradition, but it’s also the first chip ARM has ever explicitly aimed at the low-end server market.

The logical evolution of Cortex-A9

The A15 is exactly what you’d expect from a company that wants to target very different usage scenarios with the same processor. It’s a 32-bit chip that supports 40-bit physical addressing, multiple power domains, hardware-level virtualization, and several new instructions to the ARMv7 ISA. The L1 cache is twice as wide (128-bit, up from 64). The Cortex-A15 can decode three instructions per clock cycle (the A9 could only decode two), and it can issue eight micro-ops per cycle as compared to the A9′s four.

The A15′s branch predictor is more advanced than the A9′s, it can execute a greater range of instructions out-of-order, and it can execute 128-bit NEON/SIMD instructions in a single cycle. It can execute a pair of Load/Store commands simultaneously and supports multiple clock domains.

These enhancements make the Cortex-A15 significantly more powerful than the Cortex-A9; Anandtech’s recent benchmarks of a new Chromebook from Samsung show it outpacing Intel’s dual-core/quad-threaded Atom as well. This is unsurprising. Virtually all of theimprovements Intel has made to Atom since it launched the core in 2008 have been on the power efficiency side of the equation; the chip’s performance has scarcely budged.

One of the Cortex-A15′s most important tools isn’t actually part of the CPU. Server chips require sophisticated linkages and cross-communication capabilities, but it didn’t make sense to build those components into the base CPU design when the average smartphone/tablet would never need them. ARM’s CoreLink CCI-400 (Cache Coherent Interconnect) is a separate block of silicon that connects the CPUs, MMUs, graphics, Ethernet, and memory controller.

The CCI-400 is integral to ARM’s plans to push into servers. Each component attaches to the CCI using a 128-bit data path and the unit is designed to run at 50% the clock speed of the Cortex-A15s. That works out to a clock speed of between 500MHz and 1.25GHz. ARM is talking up the CCI-400 as a key component of big.LITTLE, its pairing scheme that combines Cortex-A15 and Cortex-A7 CPUs on the same SoC to maximize execution efficiency, but we expect the chip will primarily be used to connect dense server SoCs.

Putting the A15 in context

Based on the figures we’ve seen to date, the Cortex-A15 is one of the fastest — possibly thefastest — mobile architecture currently on the market. Much depends on implementation; smartphone/tablet performance can vary considerably depending on cache size, RAM speed, the number of memory channels, and of course the power envelope.

The Cortex-A15 will show up in mobile phones, but this slide, from ARM’s own presentations, puts a major emphasis on tablets, notebooks, NAS’s, routers, and servers. For now, the A15 will likely be confined to high-end phones. Even there, battery life will take a hit, if ARM’s guidance on frequency is anything to go by.

For now, the best way to think about the Cortex-A15 is as the architecture that puts the greatest emphasis on performance, with a subsequent hit to power consumption. It may not be as flexible as the Cortex-A9, Krait, or even Atom, but it could rack up huge wins in tablets and low-end Windows RT notebooks, where batteries are larger.

Does the Cortex-A15 threaten Chipzilla’s mobile plans?

There are people who will look at the performance gap between Atom and the Cortex-A15 and conclude that ARM has won the day, game, set, and match. This is inaccurate. The current 32nm Atom parts (Medfield and Clover Trail) draw significantly less power than the old 45nnm hardware. More importantly, there’s the fact that no one will be squeezing a 1.7GHz Cortex-A15 into a phone any time soon.

Intel’s first efforts in phones and tablets have been aimed at carving out a competitive space for itself in midrange markets. The Cortex-A15, in contrast, is a high-end part. It won’t face serious competition from either x86 manufacturer until Intel launches its 22nm Atom refresh and AMD’s 28nm Kabini debuts. Both of these events are scheduled for mid-2013.

We expect ARM’s various partners to make more noise around the Cortex-A15 in servers, cloud devices, and tablets/notebooks than smartphones. It’s not a threat to Intel’s enterprise interests at the moment, but it could present Santa Clara with a problem long term if ultra-dense servers, such as those produced by AMD/SeaMicro, start gnawing into the traditional x86 market.

Chinese chipmaking upstarts are gearing up to challenge Qualcomm and Intel

Share This Article

0
inShare

If you follow developments in the US smartphone and tablet market, you are at least passingly familiar with the big names in the industry. Apple, Qualcomm, Samsung, and Nvidia are the major ARM developers, while Intel’s recent Medfield, Clover Trail, and just-unveiled Silvermont products offer an increasingly significant challenge to the ARM tetrarchy. AMD has its own plans as well, though these are largely confined to the W8 tablet space.

Samsung, Qualcomm, and Nvidia account for the lion’s share of the high-end mobile market, but they’re not the only players. Two other fabless design manufacturers, AllWinner and MediaTek, are increasingly hungry for a piece of the lucrative mobile pie.

MediaTek moves to challenge

MediaTek was founded in 1997 as a spinoff from Taiwan’s United Microelectronics Corporation, or UMC. It started off as a controller chip manufacturer for CD and DVD drives but has pushed into mobile devices and wireless products. Total sales for 2012 in US dollars were $3.3 billion — and while that’s not in Samsung or Qualcomm territory, it’s nothing to sneeze at. The company has recently begun shipping several higher-profile products, including its MT6572 SoC. The less-than-sexy name cloaks a dual-core Cortex-A7 SoC at 1.2GHz with an integrated WiFi radio, Bluetooth, GPS, and FM radio tuner all on board.

It’s a bit exuberant, but here’s the feature set.

This is MediaTek’s second Cortex-A7 product — it launched a quad-core Cortex-A7 SoC late in 2012. One reason the chip stands out is that MediaTek has eschewed ARM’s big.LITTLE strategy to explicitly focus on the little end of the equation. ARM has mostly positioned the Cortex-A7 as a low-power companion for the Cortex-A15 and a way to improve total deviceefficiency by using a svelte low-power core as much as possible. MediaTek believes the Cortex-A7 has enough horsepower to drive a lower-end smartphone on its own, and has designed chips on 28nm technology to help make that happen.

MediaTek’s idea for these chips is to create a market around lower-end smartphones whose users care about battery life but still want an acceptable level of performance. The Cortex-A7, meanwhile, should be up to the challenge. Benchmarks of the quad-core version point to an acceptably snappy chip that trades some synthetic performance for less power consumption — exactly what MediaTek is targeting. If the dual-core variant does so as well, it could be strong competition for lower-end Snapdragons and Intel’s Atom.

The high-end quad-core variant, the MT6589T, recently broke cover and is proving to be a potent competitor for slightly older smartphones. Recently published benchmarks show the chip edging out the HTC One X (the Nvidia Tegra 3 version) and surpassing the Google Nexus 10 and Nexus 7 in the AnTuTu benchmark. These are only two tests, to be sure, but the Taiwanese vendor is clearly capable of fielding hardware that can compete at more than the bottom end of the market.

Allwinner amps up

Allwinner is a much younger company. Founded in 2007, it employs some 500 people, almost all of which are engineers. Unlike MediaTek, it’s a mainland China company. It’s been making waves for several years, starting with its A1X processor series. Newegg has a hefty list of Allwinner-equipped products, and while plenty of them are substandard schlock in the $80-$100 range, there’s an increasing number of devices pushing into $150-$200.

Allwinner’s latest A31 processors are still based on 40nm technology, but the quad-core Cortex-A7 designs should help limit total power consumption. In late March, the company announced its A31s SoCs, which are designed for so-called “phablets” — phones that range from five to seven inches. Equally as interesting to consumers and investors alike is that the company reportedly sells its SoCs for as little as $8-$10 each. That’s half of what Nvidia has reportedly charged for Tegra 3, and substantially less than Intel.

Devices like the upcoming Onda tablet might look like iPads, but they aren’t. Anyone arguing that these products are going to make a near-term chomp into Samsung, Qualcomm, Nvidia, or Intel is indulging in flights of fancy. Less fanciful is the fact that the ultra-cheap Android tablet market is growing up fast. When I tested the Walgreen’s Maylong tablet in December 2010, I was anything but impressed. The system used a resistive touchscreen, a 533MHz ARM9 core, 256MB of RAM, and 2GB of NAND flash. Even then, “unimpressive” was an understatement.

The Onda, at $145, is a quad-core Cortex-A7 design with 16GB of NAND, 1GB of RAM, a 1200×800 display (up from 840×480), an 8-inch screen, and a 3G modem. Video is via the SGX544MP2, and while that’s not a cutting-edge solution these days, it’s a much stronger product than what we’d have expected in these devices before.

Are we going to see these products launching stateside any time soon? For now, the answer is “no.” They’re not ready. Reading reviews of some recent low-end devices, a number of readers complain of a variety of problems — one person’s USB ports don’t work, another can’t get WiFi functional, a third has no trouble with WiFi or USB, but sound is heavily distorted. These are quality control problems that dog low-end manufacturing and they have to be fixed before these devices are ready for Western markets.

What the growth of these companies shows, however, is that there are other ARM licenseesthat could emerge as major players in their own right. This is particularly true in India and China, where much of the long-term smartphone growth is expected to take place.

Windows 8.1 will be a free upgrade...

Windows 8.1 will be a free upgrade, will begin public beta testing on June 26

• By Sebastian Anthony on May 14, 2013 at 1:27 pm
• 33 Comments

Share This Article

2
inShare

Microsoft has announced that Windows 8.1, aka Blue, will be a free upgrade for all Windows 8 users. Microsoft has also confirmed that it intends to release a public preview of Windows 8.1 on June 26, the first day of the Build developer conference — and if everything goes to plan, you should be able to get your hands on Windows 8.1 sometime in the fall (it should release to manufacturing in August). As expected, there’s also a Windows 8.1 update for Windows RT.

For users who don’t already have Windows 8, you’ll have be able to buy a standalone version of Windows 8.1 — which will be priced the same as Windows 8 ($200 for the full product, or $120 for the upgrade). Microsoft has previously confirmed that Windows 8.1 will be available as an easy upgrade from the Windows Store, much in the same way that Apple distributes OS X upgrades via the Mac App Store. It isn’t clear if there will also be boxed versions of Windows 8.1, or if 8.1 will be a purely digital product.

Windows 8.1 (Microsoft, like Picasso, has now officially ended its “Blue” period) is a Windows 8 feature and service pack. Like a service pack, Windows 8.1 will roll up all of the Windows 8 security updates and fixes into a single installer. In this regard, Windows 8.1 is conventional. What is unconventional, however, is the release of a service pack that also significantly alters the interface, bundled apps, and overall user experience — i.e. a feature pack. Normally we would have to wait three years for Windows 9 for major changes, but Windows 8.1 represents the beginning of Microsoft’s shift to a faster, annual release cycle. This change, which affects almost every Microsoft software product, is due to the mobile ecosystem — and thus Microsoft’s biggest opposition — moving a lot faster than the desktop PC. Put simply, Windows 8 can’t complete in the tablet space if Microsoft only puts out a new release every three years, while Apple and Google release new OS versions every year.

As for what features Windows 8.1 will actually include, there’s still very little to report. There are strong leaks that indicate that the Start menu and button will return, and a leaked early build of Windows 8.1 shows some changes/updates to the Metro interface. There have also been repeated hints that Windows 8.1 will bring 7- and 8-inch tablets into the fold (at the moment, Windows 8 essentially requires a 10-inch display). Beyond that, Microsoft has merely said that Windows 8.1 will “respond to consumer feedback.”

In other news, Microsoft says it now has 70,000 Metro apps in the Windows Store — and, speaking at the JP Morgan Technology, Media and Telecom Conference, Windows CFO Tami Reller says that Microsoft won’t drop Windows RT/ARM support, despite Intel’s incoming Bay Trail chips.

If you don’t want to wait for Windows 8.1 to address your feedback, be sure to check out our collection of Windows 8 tips and tricks.

MIT constructs synthetic analog computers inside living cells

Share This Article

1
inShare

Early computers could perform fairly complex calculations just by arranging a few analog circuits. By choosing appropriate resistor and capacitor values, arithmetic summers, integrators, and differentiators can be easily constructed using a single amplifier, so long as the answer is less than the supply voltage. Indeed this is, at least in part, how early computers were able to track and deploy countermeasures to incoming missiles. It is even possible to take a couple of transistors and turn them into a precision multiplier. Inspired by these simple circuits, researchers at MIT have created synthetic analog computers that run genetic machinery — in other words, living cell calculators. In addition to arithmetic, these computers are also able to also perform more complex functions like taking logarithms, square roots, and even do power law scaling (evaluate functions of x raised to a certain power). While these machines are not as convenient as any inexpensive calculator, they can process numbers up to four digits, and are a heck of a lot smaller.

Up until the mid-seventies, the old slipstick, or slide-rule, was the dominate calculating tool. In its most basic form, it uses two logarithmic scales to allow rapid multiplication and division. By exploiting feedback in gene circuits, the researchers could implement logarithmic-linear (think of old-fashioned log-linear graphing paper) computations over four orders of magnitude. To do this, a continuous range of input values are represented by things like the amount of certain kinds of sugars, or other molecules present in the cell.

For addition or multiplication, each of these inputs turns on a gene that manufactures an optically-detectable molecule called green fluorescent protein (GFP). For subtraction or division, one of the inputs instead suppresses the expression of GFP. Depending on how each input is rigged to behave in the genetic circuits, the different functions can be accurately encoded. The answer is read out by measuring the total fluorescence of the cell.

We recently reported on the creation of biological transistors inside cells at Stanford. These “transcriptors” were then used to build up various kinds of digital logic gates. Combining these kinds of digital circuits with the analog circuits described here, powerful sensing and control platforms can be created on the very small scales. Digital may reign supreme inside the neat little microcosm of a computer, but to interface to the real analog world, conversion elements for both directions will become essential.

One thing these machines could do would be to add a whole new level of precision to the regulation of genes. Biological clocks, as naturally implemented both at the cell and organism level, normally do things like control circadian rhythms, or cue the developmental programs that build the creature. These oscillators work fairly well, but lack the accuracy of the crystal timepieces used in computer chips. For the moment the researchers are hosting their biological computers in bacterial cells. They would like to develop analogous circuits that operate in mammalian cells, where these functions might be brought into better use.

If at some point we are to construct the “molecular ticker tapes” imagined by the creators of the new BRAIN Initiative to map neural activity, building these kinds of circuits into the readout devices would provide powerful methods of regulation and control. Having a complete brain activity map written into pieces of DNA inside each neuron is a nice idea, but to read that information out, and do something with it, we will need get the neurons to speak our language.

Google’s new underwater Street View...

The technology behind Google’s new underwater Street View

• By David Cardinal on May 17, 2013 at 9:10 am
• 0 Comments

Share This Article

1
inShare

Street View is great for visualizing destinations before you arrive, but increasingly it is also being used to take people places they may never be able to go in person. Richard Vevers and Underwater Earth have partnered with Google to begin creating the full 360-degree views of coral reefs needed to make them part of the Street View universe. Vevers explained to us during a session at Google I/O that, “people don’t value what they can’t experience. Coral reefs are disappearing and need help. By sharing them through Street View we hope to generate interest and awareness, encouraging their preservation.” In addition, the large volume of data being collected by the project will also serve as a scientific baseline for future studies of how reefs and other sensitive ocean bottom areas are faring.

As you might expect, underwater Street View requires some fancy engineering. Because of the way lenses behave underwater, the team needed to design a new camera rig. By using three very-wide-angle-equipped cameras, the current Seaview cameras — dubbed SVII — can capture left and right images that are merged into a 360-degree panorama, as well as a direct downward image. The downward image is used for compatibility with existingscientific databases, which are primarily composed of images taken by shooting straight down. You can see one underwater Street View below — for the full collection, hit upGoogle’s Ocean Street View website.

Not your ordinary underwater camera housing

The individual cameras on an SVII are sealed in a sphere, which is placed on the front of a torpedo-shaped hull that provides propulsion and battery power. The rig is steered by a trained diver using its compass. It also includes a tablet for controlling the cameras — so they can remain in their watertight housing as much as possible — as well as GPS equipment and other sensors. The mapping process requires capturing an image about every three seconds — spaced about two meters apart — along the bottom. After each dive, the 3,000 or so “fisheye” images are assembled into 1,000 rectilinear panoramas.

Currently the team only has four cameras — costing about $50,000 each — which has limited them to exploring only part of the Great Barrier Reef. They aim to expand the project until they have covered all the world’s coral reefs, but that will require more money and more technology. In particular, Vevers is hoping to be able to make citizen scientists out of the 10 million recreational divers by allowing them to capture images with their own underwater cameras — perhaps even smartphones in underwater housings. In the meantime, it also thinks volunteers could be helpful in cataloging the sea creatures found in the images. Over time, their goal is to use the human-powered results as data for a machine learning system that can automate the process.

Other areas of interest for the Underwater Earth team, in addition to expanding its database of 150,000 images, are creating 3D models of the reefs, and partnering with autonomous submersible creators to deploy a robotic version of its rig that can stay down longer and doesn’t require a trained diver.