Since battery technology isn’t really enabling us to pack more than a few watt-hours into our portable devices, companies like Intel, Nvidia, and Texas Instruments are working hard at making their chips and processors more efficient. Apple is acknowledged to be the leader here — their vertically-integrated device creation process (and the mysterious A5) gives them the control they need to maximize battery and get that critical 10-hour claim. TI is hoping that an upcoming family of its OMAP chips will take some of the pressure off manufacturers and enable current batteries to last from dawn till dusk.

Speaking with Fudzilla, TI’s Brian Carlson mentioned that the 2013 OMAPs will use a 20nm process and should enable “true all day computing.” What does that mean, exactly, when the most popular device is sporting a 10-hour battery? That’s all day in most people’s books, even with plenty of video usage. Carlson didn’t put a number on it, but it’s clear that TI wants to focus on efficiency and not power. Nvidia, on the other hand, is in an arms race with Intel (an ARM race if you will) to push the power of its mobile processors as high as possible.

The question is: what will the tablet and mobile world look like in 2013? Probably not radically different, but an all-day battery might not be much of a draw by then. At any rate the improvement of memory and cache handling and other performance tweaks is always welcome.

As Tom’s Hardware points out, there were indications of TI’s intent to sell off the OMAP brand after the release of OMAP 5 in 2012. A nice bidding war between Nvidia and Intel would bring in a hell of a payday. Discussing plans for a 2013 release, however, seems to moot speculation on that point; maybe TI felt they were better off playing the game than cashing out.

Now, it’s important to differentiate this 3D method from others under investigation, like IBM’s. This isn’t a multi-layer solution, merely a more complicated shape for the single layer of transistors we know and love. I say merely, but of course sculpting things at a near-atomic level is no joke. So what exactly is the advance here, and what will it enable?

If you’d rather not read, check out the video above. It does a good job of explaining, though there’s a bit of preamble you can skip.

Essentially, what’s happened is this: Intel has used nano-scale architecture to move the conductive electron layer actually inside the metal component of the gate, in which the insulating layer is embedded.This illustration from their announcement deck shows it well, and the following picture is the progression of the fabrication process:

Surrounding the electron flow on three sides, as opposed to just one, allows for improvements in just about every aspect of the transistor. Intel claims a 50% power use reduction when under load, and a 37% performance increase at low voltage, presumably from the increased rate at which these gates can open or close.

This is quite a serious leap for Intel, and immediately puts AMD further into peril, since low-power combos like Fusion have been its biggest advantage of late. It still has the lead in graphics (Intel isn’t even competitive there), but this major boost in efficiency by Intel could put them in danger on their home turf, things like all-in-ones and ultraportable laptops.

And what about ARM? This advance probably won’t hurt them. A more efficient x86 architecture is still an x86 architecture, and right now ARM is so prevalent in the portable computing market (mobile, tablet, embedded) that even a large bump to Intel’s prospects like this would take years to propagate, during which time competitors like Nvidia (Intel co-nemesis) will continue to throw in with the non-x86 option. That said, Android is perhaps the most likely to break from the pack, but that depends entirely on manufacturers, not Google, ARM, or Intel.

Intel’s roadmap for the 22nm process first mentioned Ivy Bridge, the shrink-down of current Sandy Bridge Core processors. And then there’s the low-power segment, with “Future” entrants into the various portable segments of the market. The recently-released Oak Trail generation of Atom processors will be succeeded by Clover Trail before the 22nm shift, but it’s possible they could still make it happen in mid-2012. And Intel is so marginalized in smartphone processors that I wouldn’t even hazard a guess as to their strategy there. Actually, I’d say they’ll push the mid-size segment with their low-power stuff, and then try to shrink it — the opposite of Google’s approach with Android.

The processors are due to be manufactured in the second half of 2011, though there’s no information on how or when they’ll be marketed, made available, or integrated with existing product lines.

Whatever the case, this new tech is exciting and fascinating, and I look forward to the new devices and components it enables. Nano-scale research is worth reading about no matter who’s doing it, or to what end.

Cloverview is going to be the name of Intel’s next-generation Atom processors for tablets according to a hint dropped during a speech today at the Intel Developer Forum.

The chip represents Intel’s direction of lowering energy consumption, especially for mobile processors, by using a 32-nanometer manufacturing process. Cloverview will join two other new 32-nanometer Atom chips still in the works; Cedar Trail for netbooks, Medfield for low-end smartphones and tablets.

Intel has to get serious about mobile processors; ARM processors are considered more efficient, which is why they are in more tablets. As far as we can tell they won’t be going away either.

Researchers in Europe were able to create such a processor by using 4,000 organic transistors. The processor sits on top of flexible plastic foil measuring about 2 x 2 cm. As of now, its performance is a bit weak; it can only run one simple, 16 instruction program which has to be hardcoded in.

It’s pretty basic, but if this proof of concept can follow Moore’s law like its silicon sister, then in a few years we could be wearing electronic clothes.

This week at the International Solid-State Circuits Conference, European researchers showed off the world’s first flexible microprocessor made with organic semiconductors. Right now, processing power is similar to what was found in the 1970′s, but the advantage is that the processor is flexible. What’s so good about the research is that it can lead to fully bendable displays and sensors; something we could begin to see in clothing, gadgets and biomedical applications.

The bendable conducting plastic contains only 4000 transistors (todays, have hundreds of millions). You can imagine what this could bring.

[via slashgear]

I’ve just read Anandtech’s thorough analysis of the chips, and I think he’s right in recommending the i5-2500K. You’re really getting just a huge amount of performance for a bargain price ($216 as they quote it, but it will vary), and even some of the lower-end chips will do heavy HTPC duty now.

Hot Hardware and HardOCP both have good roundups of current reviews (as well as reviews of their own) so give ‘em a read if you’re thinking of upgrading.

Actually, Intel’s CTO (you have to be kind of smart to have that job) said at the time that we’d hit the wall sometime in the next ten (now eight) years. Certainly that hasn’t happened yet, as it seems that Intel already has 15nm chips on their roadmap. 15nm is a ridiculously small amount of space; if you were 15nm tall, you’d constantly be getting knocked over by atoms.

In order to maintain Moore’s Law, at some point we’ll have to change things drastically, like stacking processor layers and cooling them with nanocanals. I know, sounds like science fiction, and it is — today. But 15nm-based processors were science fiction a couple years ago, and here we are.

[via CNET]

[via]

Interestingly, they chose a rather old GPU and a comparatively new CPU to compare: an Nvidia GTX280 and a Core i7 960. Maybe they chose on price parity? Whatever the case, they found that while indeed the GPU advantage was not as great as suggested, it was significant: 2.5 times faster on average and up to 14 times faster in certain situations.

Nvidia took the opening for all it was worth:

It’s a rare day in the world of technology when a company you compete with stands up at an important conference and declares that your technology is only up to 14 times faster than theirs.

To be fair: we can’t really expect unbiased judgment from either side, and the tests in the paper are too complex to be boiled down to a “oh it’s this much faster” talking point. I just think the drama is hilarious.

[via TG Daily]

That said, however, there’s a lot more to a device than making the right hardware decisions. The Samsung Wave, it has been revealed, uses a very similar Samsung-made (obviously) 1GHz ARM A8 processor, putting it as near Apple’s A4 as can be. But do you think you’re going to have a similar user experience? No, silly! So much depends upon the UI designers and coders that the processor is really only a secondary consideration.

In other words, the processor (magical A4 or not) doesn’t make anything work on its own — it just provides a sandbox for the coders and designers to work in. That’s very important, but it’s very far from the whole story. This is why there are puzzling differences between phones with similar specs. Could a memory leak in a program not get fixed in Android 1.6 on a G1, where perhaps an outdated RAM caching technique is still in use? Sure, this kind of thing happens all the time. Apple is unique in that they control almost the entire process of creating a device from start to finish. You see something similar in the Zune HD, which I loved, but it also has an advanced ARM processor with a special name, and again, it’s perhaps the least visibly important part of the device.

Ironically, when the iPad and iPhone 4 come out, a large amount of attention is paid to the processor and display, things which Apple almost certainly had very little to do with! Sure, their curation of the devices’ hardware is impeccable, but the “magic” doesn’t come from the A4, any more than the snappiness of Google something comes from the brand of memory they use in their servers. On that note, it is only fair to add, however, that Google’s unorthodox server technique does deserve some credit, but their database and crawling algorithms are the real stars.

If you want to give credit where credit is due, thank ARM and Samsung for making excellent hardware, and thank Apple for taking advantage of it correctly. I have my issues with Apple, but one thing they’ve always managed to do is get the absolute most out of a given hardware setup — RIM, HTC, Sony, Nokia, all these guys would take the same bits and get an inferior product. It really is just a fact at this point.

The A4 is a nice piece of hardware, don’t get me wrong — but there’s more to a device than its processor, and it seems like a lot of the other stuff gets lost in the background when the spotlight shines on a single aspect. When it comes to the A4, it seems like the spotlight is completely arbitrary, since so much of the A4 is shared by other devices on the market right now. A little perspective shows where the credit truly lies in a great device.

Remember back in the day when you could go into a store and compare two clock speeds and come out with the right machine for you? Everything else was immaterial. Hard drives could be upgraded, memory could be added, but clock speed was the number you lived or died by. 1.8GHz was better than 1.5GHz every day of the week, right?

Those days are over. Moore’s Law, the idea that “the number of transistors that can be placed inexpensively on an integrated circuit doubles every two years” is nearly over, which is why multiple cores are now appearing in consumer PCs. When you have multiple cores, clock speed doesn’t make any sense. A 2.5GHz quad-core machine, to the average Joe or Jane, seems to mean that the PC has four cores running at 2.5GHz each. That means it’s 10GHz, right? Right?

Wrong, but that doesn’t matter anymore. AMD – and, to some extent, Intel – has been moving steadily away from clock speeds for years and this is the year they leave them behind entirely. Take a look at AMD’s Vision page. This new naming convention offers choices based on use case. For example, Vision Premium offers the ability to:

Run several applications at once
Play mainstream games
Convert CDs to MP3s
Perform basic photo editing
Watch Blu-Ray/HD movies
Use a webcam

Then, when you click on “Show me Vision Premium Laptops,” you are led to a list of PCs and laptops that are quite disparate in their power and specs. The assumption is that if a laptop falls under a Vision tranche, you’re good. The only number that now matters is price. If you pay more, you get a bit more. And if you want to dig into the specs, you can, but the only things that are going to change the price are the storage and memory components. The processor and, presumably, the graphics solution, will stay the same.

I, for one, welcome our Vision overlords. The question “What laptop should I buy?” can be answered with a wave of ones hand: go over there, pick a pretty one. That’s essentially what we’re dealing with these days, anyway, the commoditization of computing hardware. If the netbook revolution has taught us anything it’s that consumers don’t want power, they want the package. Let the geeks cling to their graphics cards and overclocking while the rest of the world wanders into Best Buy, drops a credit card, and wanders out happy.

I know I sound a bit facetious, but I’m not. This is 2010. Computing, in a way, is flat. When the iPad and its ilk are on the same level playing field as an Alienware monster laptop, you really can’t look at speeds and feeds anymore. The question now is “Does this do video?” or “Does this play games?” and you’d say, “Well, it has a 2GHz processor and discrete graphics and 4GB of RAM, so yes.” Now you say “Does it run Windows 7? Then it can probably run a few good games.” There are thousands of games, after all, and most of them run fine with a bit of coaxing. Do you want the ultimate in performance? Build it yourself or go for Vision Black. The choice is yours and it’s gotten much easier.

My hope is that Intel begins adopting this nomenclature for their processors as well. By segmenting the product line into a few understandable blocks, you reduce complexity and, in the end, assist the consumer more than anyone will ever know.

So what’s going on here? Is Apple jumping ship, or just teasing in order to get a rise out of Intel and NVIDIA? Maybe a little of both.

If you’ve followed computer hardware over the last 10 years, you probably remember a period of AMD ascendancy five or six years ago, just before Intel’s Core series came out. It was also a period of great strife in the video card wars, with ATI flagging and NVIDIA thriving. Apple switched horses to X86 and Intel at a good time, and although the vast majority of the credit goes to Intel for creating an excellent product, the alliance with Apple surely helped popularize both platforms. AMD swallowed ATI some time after and has been experiencing mixed success against its rivals.

Apple’s decision at the time was prescient; they liked what they saw on Intel’s roadmap, not what was available at that moment. It’s just possible that AMD is pitching Apple in a similar way, although sneak peeks of their 2010 technologies aren’t really that impressive. Their major innovations are being pushed back to 2011, it seems, and they’re relying on graphics integration and “total package” strategies to sell their way through 2010. But for a good friend like Apple, maybe they have something special. It’s not out of the question. Bobcat-based iPad? I’m thinking no, but it’s fun to play pretend.

Furthermore, while Intel clearly has the lead in performance, the latest update by Apple suggests that raw power really isn’t the focus for its notebook line. After all, what use is a high-end processor if you don’t have graphics and RAM to match? Instead, they’ve opted for a full-system solution, something they laid the groundwork for in Snow Leopard. OpenCL and Grand Central Dispatch are in prime position to be deployed as serious OS tools, but we don’t see that happening with the latest MBPs. Could they be a smokescreen? To be honest, I doubt it: that’s far too conspiracy-theory to really put any thought into, but there is a grain of truth to it. If Apple doesn’t have at least a little secret plan, why aren’t they leveraging the OS tech they’ve worked so hard to create?

Some are floating the idea of actually splitting the lineup: having AMD on the low end notebooks to save cost, and putting Intel in the high end, to capitalize on performance claims and big budgets. That’s a good strategy… for someone like Dell or HP. They’re all about choice, personalization, and budget. Like Lancelot (Percival?), Apple’s strength is in its purity, and although that’s been under attack recently with the fragmentation of the iPhone platform, I feel sure that Apple wants to keep its lineup as limited as possible. Claims of merit ring hollow when you extol both sides at once. Apple wants its judgment to be categorical: we use these processors, this screen, this material, and here is why.

What’s left? Well, I think everyone’s first impression was right: Apple is flirting with AMD to make Intel and NVIDIA jealous. And when it meets with them next time, Apple will cite all the sweet nothings and sexy promises AMD made. After all, AMD does have a compelling platform and it suits Apple’s apparent new strategy quite well; even if AMD knows Apple is using them as a tool to inflame someone across the room, it’s a good chance to get a word in.

And why, exactly, have I written 700 words about mere rumored meetings between a supplier and a buyer? Well, I think we all know that Apple is the Brangelina of the tech world (the Edward? I don’t know), but more importantly Apple is such a huge mover in the notebook sector that intrigues like this really do matter. It’s not just a question of gigahertz. This is the kind of industry gossip that can alter (however invisibly to the end user) the landscape of personal computing hardware.

Gigabyte released a couple of BIOS updates recently, and Tech Connect spotted the news: The X6 1035T will have 6 cores, and be running at 2.6 GHz, and the X6 1055T is going to run at 2.8GHz. There was some news about the new Phenom II X6 as well; the 1075T will be clocked at 3.0 GHz, which is about 333 MHz slower then Intel’s new i7 980x. Bummer for AMD, but we’ll see who the real winner is when the pricing comes out; Intel’s new chip sells for $999 right now.

The issue I see with it is this, though: the water is drawn to the pattern on the silicon, right? More so than to itself. So once it reaches the silicon, what will pull it away? It seems like water would simply coat the silicon in a single-molecule layer, and then the rest of the water would roll downhill as normal. But hey, who the hell am I?

Well. The latest advancement is that IBM is thinking of replacing the conductive copper channels in today’s chips with light. But it’s more than a simple fiber-optic setup or something along those lines; the idea is that a single photon would set off an electron cascade, meaning that a significant charge could be effected on the far side of a gap, for the energy cost of sending out a single photon. The usual “this would enable computers a gazillion times faster” talk follows.

I see one major problem with this. We have systems like this in our bodies; IBM’s design is almost biomimetic. Our brains in particular can take a single “photon” (an action potential) and have it effect a huge change on a target neuron — if that neuron is all charged up and ready to receive. But a few action potentials too many and the charged molecules that allow for a rapid multiplication of electrical power will be exhausted, and it takes some time to recharge. You can actually see this for yourself, literally: stare at a light for a second and the little phosphenes that appear before your eyes are a result of certain cells in your eyes being unable to “recharge” fast enough to propagate the correct signal.

There would be less risk of that if your eyes were plugged into the wall, of course. But on a scale small enough, there may be issues of heat or charge fatigue that would get in the way as in the brain. Anyhow, there’s not much we can say at this point, since the research is at the “CG explanation on YouTube” stage, meaning we won’t hear from them for another couple years.

[via PC World]

I don’t pretend to know what the process was for creating the A4, or Tegra 2, or any other ARM-licensed system on a chip. But a billion dollars is a hell of a lot of money to put into R&D for a single device. Apple would have to sell four or five million devices just to break even. I don’t know, maybe they really did spend this kind of money. The estimate just seems a bit high to me, and it comes off as sort of arbitrary in the NYT article. It seems that the bulk of the design work must have been done by ARM — which was why Apple licensed from them instead of leaning on their own PA Semi IP. I’d take this particular story with a grain of salt.

[via Apple Insider]

Beyond this (ARM licensing the CPU and GPU to Apple), it’s mostly speculation. But there’s room for processor shrinks and even overclocking; the A9 has been clocked to 1GHz by everyone because it stays cool, but if Apple (or anyone) were to apply some better cooling, they could take it all the way up to 1.3GHz.

The chip itself, which contains other components not yet revealed, is supposed by many to be manufactured by Samsung, like the one in the 3GS, which incidentally contains the A9′s predecessor, the A8, clocked to 600MHz.

[via Apple Insider]

The processors range in speed from 1-2GHz and all feature an 800MHz frontside bus.

According to VIA’s press release:

“Based on the 64-bit superscalar ‘Isaiah’ architecture, VIA Nano 3000 Series processors deliver the most compelling thin and light notebook computing experience with their rich HD entertainment capabilities, including support for flawless playback of high bit-rate 1080p HD video, as well as low power consumption resulting in longer battery life.”

The new processors are currently available to system builders and are expected to show up in retail markets in the first quarter of next year, presumably to take on Intel’s upcoming line of PineTrail processors.

Here’s the full press release:

VIA Introduces New VIA Nano 3000 Series Processors

VIA’s fastest and most power efficient processors yet deliver richest mobile and all-in-one desktop computing experience

Taipei, Taiwan, 3 November 2009 - VIA Technologies, Inc, a leading innovator of power efficient x86 processor platforms,　today introduced its new VIA Nano 3000 Series processors, bringing enhanced digital media performance and lower power consumption to Windows 7 thin and light notebook and all-in-one desktop PC markets.

Based on the 64-bit superscalar ‘Isaiah’ architecture, VIA Nano 3000 Series processors deliver the most compelling thin and light notebook computing experience with their rich HD entertainment capabilities, including support for flawless playback of high bit-rate 1080p HD video, as well as low power consumption resulting in longer battery life.

With a host of advanced features including 64-bit support, advanced CPU virtualization technology, SSE4 for enhanced multimedia processing, and the industry-leading encryption and security capabilities integrated in the VIA PadLock™ Security Engine, VIA Nano 3000 Series processors also provide a secure, high-performance solution for emerging cloud-based computing environments.

“With the VIA Nano 3000 Series, we are launching our fastest and most power-efficient processors yet,” commented Richard Brown, VP International Marketing, VIA Technologies, Inc. “Coupled with our market-leading digital media chipsets, they enable the richest experience across a broad range of mobile and all-in-one system designs.”

VIA Nano 3000 Series
VIA Nano 3000 Series processors are built on the successful 64-bit, superscalar architecture that powers the VIA Nano 1000 Series and 2000 Series processors, which have been adopted by leading OEMs worldwide for a growing number of market-leading mini-note, small form factor desktop, and energy-efficient server designs.

Available at speeds from 1.0GHz to 2.0GHz, VIA Nano 3000 Series processors deliver up to 20% higher performance using up to 20% less power than current VIA Nano processors and boast a number of new features including support for the SSE4 multimedia instruction set and VIA VT virtualization technology.

Fully compatible with all Microsoft operating systems, including the new Windows 7, as well as all popular Linux distributions, the VIA Nano 3000 Series processors use the NanoBGA2 package, making them pin-to-pin compatible with VIA Nano 1000 Series, VIA Nano 2000 Series, VIA C7, VIA C7-M and VIA Eden processors for easy upgrades of existing designs.

VIA Nano 3000 Series Availability
VIA Nano 3000 Series processor samples are currently available for OEMs and motherboard vendors, and will enter mass production in Q1 2010.

You can see me in some of the shots. I’m famous! The music is a little over-the-top, but you know what else is over the top? Pouring liquid helium on a piece of silicon to let it process twice as fast as it’s supposed to.

Last week AMD invited CrunchGear down to Austin to check out an overclocking event they were holding, at which many, many liters of liquid nitrogen and the much colder liquid helium would be consumed by thirsty processors. They asked us, however, not to video the entire event, since they’d have their official video coming out shortly and some of the technology being used was still in development. No problem, we said, we’ll just dip our pulled-pork tacos in the spare liquid nitrogen.

A lot of stuff actually went down there. I’ve always wanted to get up close and personal with some of the weird cooling solutions these overclockers use, as not only do they just look cool, but they result in performance numbers hugely ahead of anything possible with traditional methods. And it’s not like you can just buy these setups at the store; one of the overclockers in attendance, name of K|ngp|n, is actually an accomplished metalworker and manufactures the pots he uses himself. Pots being the containers which hold the coolant and pass the temperature difference on to the processor, you know. The thickness, material, and surface design of the pot is important to where and how fast it accepts or gives off heat.

I know you didn’t click through looking for a lecture on physics, but this stuff is crazy enough to warrant your attention. It seems that processors for the last few generations have had a sort of allergy to extremely low temperatures. At very low temperatures like those created by liquid nitrogen (LN2), silver and gold do just fine as conductors, and in fact their conductance is actually increased a little around that temperature (a little over -200°C, or -323°F) if I remember right. But beyond that temperature, their conductance drops like a rock.

Copper, however, is an excellent superconductor at even lower temperatures. In fact, liquid helium (LHe4) brings it just about to its maximum conductance temperature: around -250°C. That’s not too far from absolute zero; LHe4 boils at 4 kelvins, making it one of the coldest substances we have access to. And that’s what we were pouring all over the Phenom II. It turns out that the Phenom IIs thrive like no other under these insane conditions, and although it doesn’t make overclocking a cakewalk, it does make some things possible which weren’t before.

So enough with the science, what was going on there? Well, basically the overclockers set up their rigs, cooled them as much as possible, and hit the benchmarks until either the world record or the motherboard gave out. They were aiming for 7GHz and 50,000 in 3DMark05, which are ridiculous goals but totally doable when you’ve got 600 liters of LH4 at your back.

Some interesting things I noted: the overclockers insulated their motherboard with kneadable eraser, which is a great insulator and doesn’t stick to components. LHe4 is used up at a rate of almost a liter a minute, so K|ngp|n would cool his processor down as far as he could get it with LN2 before dropping in the far more expensive LHe4.

The pots they used were super-specialized; the LHe4 one has a special design for dealing with LHe4′s tendency to evaporate instantaneously on contact with anything, and the LN2-only pot was the first see-through one in the world. It’s got two walls of thick lexan with what is essentially everclear filling up the middle. The alcohol had the lowest freezing point of any clear liquid they tested, and I found out to my great surprise later that yes, it does get extreeeeemely cold when it’s buffering a boiling pot of LN2 for a few hours.

Unfortunately the guys didn’t hit their 7GHz and 50,000 goals, but did they get damn close. K|ngp|n posted his results over the weekend: 49,089 3DMarks and 6893MHz. Here was his setup and posted results:

• ASUS CROSSHAIR III and M4A79T-deluxe motherboards
• AMD Phenom II 955 ES LHe4 cooled
• CORSAIR DOMNATOR GT and 1800CL-7s
• AMD 4870X2′s in quad X-fire stock clocks & cooling

Next time, maybe. Thanks to AMD for inviting us out. Their official video, with more details on the overclockers themselves, will be coming out shortly and we’ll link it up then for the hardware enthusiast CrunchGear readers. There’s a more extensive gallery here if you want to drool over their setups more.