Speeds and Feeds

While we're all familiar with the steady increase in the number of cores in mainstream PC and server processors, the corresponding progress in the embedded-processor market has been anything but steady.

With mainstream PC microprocessors standardizing on four-core designs such as Intel's Core i7 and leading-edge server chips ranging from 8 to 16 cores, single-core chips are no longer competitive. For embedded systems, however, one core may still be the right answer; if more are needed, the choices range up into the hundreds.

The Tilera Tile-Gx100 combines 100 independent 64-bit integer processor cores and cryptographic accelerators with memory, network, and PCI Express interfaces.

(Credit: Tilera Corporation)

The latest announcement in the many-core embedded processor market is Tilera's Tile-Gx family, which combines 16 to 100 64-bit integer processor cores with cryptographic accelerators and off-chip interfaces for memory, networking, and PCI Express. I met with Tilera before last week's announcement to discuss the technical and business issues related to the Tile-Gx.

The technical details
San Jose, Calif.-based Tilera is eager to set itself apart from the many other chip companies competing in its target markets. Unlike most embedded processors with high core counts, for example, Tilera's design allows its cores to operate truly independently, even to the extent of running different operating systems if needed. More commonly, groups of tiles will be combined to run a single task that is part of a larger workload. In this way, one chip can operate like a cluster of multiprocessor systems.

Between this distinction and the fact that cores in the Tile-Gx family are a full 64 bits wide, Tilera claims the Tile-Gx100 is the "world's first 100-core processor." I think that's just a little too broad a claim, personally, since companies such as Clearspeed and Xelerated have previously made similar claims for their chips. Even more significantly, the Tile-Gx100 doesn't exist yet. It won't be a real product until early 2011, according to Tilera's current schedule.

Tile-Gx processors aren't something most CNET readers will ever knowingly use, though these chips will likely, eventually, help carry traffic over the public Internet and through larger corporate networks. But they do provide an excellent example of the issues facing PC processor vendors as core counts continue to grow.

Consider the Tile-Gx100 block diagram shown above. It's easy to imagine that this chip can get a lot of work done. Every core can run up to three instructions per cycle at up to 1.5GHz. It has dedicated hardware accelerators for cryptography and network packet processing. The network interfaces can implement up to eight 10Gb Ethernet ports. The chip also has four DDR3 memory interfaces; to reduce DRAM accesses, every core has 320KB of local cache memory. (The total amount of cache memory in the Tile-Gx100, about 32MB, matches that of IBM's Power7 processor, which has only eight cores.)

The need for balance
It's not so easy to keep all these resources busy, however. The more complicated a chip gets, generally speaking, the more difficult it becomes to make full use of its resources. This is what we often call the balance between hardware and software.

Tilera will offer four products in the Tile-Gx family with 16, 36, 64, and 100 cores and corresponding differences in memory and networking support. This range of products helps meet the needs of different applications, but each product still needs a particular balance of application requirements for maximum efficiency.

So here lies Tilera's great challenge--finding software applications that need a large amount of CPU performance and that also:

1. Are highly parallel, so they can keep many cores busy.
2. Don't need much (if any) floating-point math, since the Tile-Gx doesn't do that.
3. Can benefit from cryptographic acceleration.
4. Consume large amounts of network bandwidth.

Tilera wants customers to think of its chips as "general-purpose" processors, but as this list shows, they're better for some purposes than for others. As PC processors reach higher core counts and integrate more functionality, they too will become more sensitive to application requirements. Integration eventually reaches a point where additional complexity adds no practical value. And the closer PC processor vendors approach that limit, the more difficult it will become to sell their latest, greatest, most complicated chips.

Network processing is the most natural fit for Tilera's capabilities, particularly high-level services like virus scanning as I discussed in September (see "Insatiable demand for mobile data challenges industry"). Internet service providers rarely provide such services for PC users, since PCs can do their own scanning--but mobile phones and other Internet appliances often can't, so these services are seeing increasing demand.

The networking market, unfortunately, is not large enough to support a company like Tilera. Although there is a lot of networking equipment sold each year, each company in the business has its own ideas about how this processing should be done. A single chip design could never capture the majority of this potential demand.

Further, the larger equipment vendors often have policies in place against relying too heavily on individual suppliers, especially small start-ups. They will commonly design different products using different chip-level technology so that the failure of a single supplier--or the purchase of a supplier by a competing equipment vendor--will have only a limited effect on their bottom line.

New business opportunities
Tilera is working to develop new markets for its current TilePro and future Tile-Gx parts. The most significant of these new markets is cloud computing, which may favor solutions like Tilera's that offer higher performance per watt.

That's the metric Tilera touts most heavily for the Tile-Gx, promising 10 times the performance per watt of Intel's Westmere-EP, a six-core 32nm processor that Intel will begin shipping in 2010, which is aimed at high-efficiency servers. (Incidentally, I commend Tilera for making this comparison; Westmere-EP is exactly what they'll be competing against. Too often, chip companies will try to make themselves look better by comparing next year's products with last year's competition.)

Although 10x is a critical multiplier in this business (see my post "The factor factor"), such an advantage doesn't necessarily guarantee success. Tilera has done everything it can to minimize the difficulties associated with software development by adopting industry-standard development tools such as GCC and Eclipse, but its Tile chips still can't run Windows and it just can't match the developer relationships that companies like Advanced Micro Devices and Intel have established.

Fortunately, Tilera is small and relatively efficient for a chip company. Last month, Tilera announced that Quanta Computer invested $10 million in the company based on Quanta's interest in cloud computing. Tilera said it has enough funding to reach cash-flow breakeven in 2011, assuming the Tile-Gx reaches market and achieves the kind of success Tilera predicts.

I remain skeptical, but hopeful. I think there's no question that in the long run, there will be plenty of demand for complex, many-core processors like Tilera's. But will Tilera still be around by that time? And in the long run, once this demand develops, larger companies such as Intel will have their own offerings.

Can Tilera carve out a market niche that it can defend against such strong competition? I just don't know, but I'm always glad to see people trying new ideas.

How would you like a single-chip microprocessor with more than four times the performance (on some applications) of Intel's best Core i7?

Then consider that up to 32 of these chips can be directly connected to form a single server, achieving four times the built-in scalability of Intel's next-generation Nehalem-EX processor.

That's IBM's widely anticipated Power7, which it described at last week's Hot Chips conference. But if you're interested, you'd better be prepared to spend a lot more than four times as much per chip. IBM isn't talking about pricing, but large Power servers can cost more than $10,000 per processor.

IBM's forthcoming Power7 server processor has eight cores, manages 32 threads, and includes 32MB of on-chip embedded DRAM cache. Power7 also has the highest levels of off-chip bandwidth ever achieved by a microprocessor.

(Credit: IBM)

What makes the Power7 so powerful? Each chip has eight cores, and each core supports four-way multithreading. There's 32MB of level-3 cache on the chip, made using embedded DRAM (eDRAM) cells. Most CPUs use SRAM for cache because it's generally easier to combine with high-performance logic, but DRAMs--with only one transistor per bit--offer compelling density advantages. IBM spent years developing a new kind of eDRAM that would work with SOI (silicon on insulator) manufacturing processes, and the Power7 is the most advanced product to use the new technology.

Interestingly, the Power7 cores run much more slowly than those in the Power6 processor, which I wrote about here in 2007 ("Live from Hot Chips 19: Session 1, IBM's Power6"). The Power6 was designed to run very fast using a long CPU pipeline in order to deliver the highest possible performance on each thread of execution.

Maybe that strategy didn't work out as well as IBM hoped, because the Power7 returns to a more traditional microarchitecture with a shorter pipeline and much lower clock rates--though IBM didn't say exactly what those rates would be.

IBM did, however, promise that the Power7 would be roughly four times as fast as the Power6, chip for chip. Since it has four times as many cores, each of the new slower-clocked cores must still deliver about as much performance as those in the previous generation.

Chip-level performance must always be matched by off-chip connections lest the incoming data or outgoing results be bottlenecked by a too-slow channel. Accordingly, the Power7 is equipped with eight I/O channels for DRAM, each of which connects to an off-chip buffering device that splits the channel into two 64-bit DRAM interfaces. All together, IBM says the Power7 has 180 GBps of DRAM interconnect that can sustain over 100 GBps of effective memory bandwidth.

There's another 50 GBps of peak I/O bandwidth and a staggering 360 GBps of peak bandwidth used to let each Power7 chip communicate with others. The DRAM connected to each chip is thus shared across larger systems.

Combining these figures, IBM says a single Power7 has 590 GBps of total off-chip bandwidth. This isn't the real number, since many of those bytes are used for error-correcting codes and other overhead, but it's still pretty impressive.

So is Power7's die size: 567 square millimeters for 1.2 billion transistors. That's nearly a square inch! IBM says that if the 32MB L3 cache had been manufactured using SRAM, the transistor count would have been 2.7 billion instead.

Still, Power7 wasn't the only high-end chip talked about at Hot Chips.

Rainbow Falls, a record for core count
Sun Microsystems was there to describe its forthcoming Rainbow Falls chip, which I assume will be marketed as the UltraSparc T3. The chip has 16 cores, each of which is reportedly able to manage 8 threads.

Sun's primary Rainbow Falls presentation focused on details of Rainbow Falls' internal and external interconnects; a second talk described the cryptographic coprocessors present in each of the chip's cores. These coprocessors--one for modular arithmetic (commonly used in public-key cryptography) and a cipher/hash unit to accelerate bulk ciphers like AES and secure hash algorithms--provide many times the performance of pure software implementations.

Fujitsu was also at Hot Chips to describe its eight-core, 2GHz Sparc64 VIIIfx processor, the latest in a long series of impressive designs from the company. Fujitsu quoted a peak performance figure of 128 GFLOPS (billions of floating-point operations per second) with a typical power consumption of just 58 watts. It did not, however, provide sustained performance or worst-case power consumption figures.

AMD, Intel vie for high-volume servers
Few of us will have direct exposure to the IBM, Sun, and Fujitsu chips. A pair of presentations from Advanced Micro Devices and Intel described products that will be much more widely available.

AMD launched its six-core Opteron processor code-named "Istanbul" earlier this year (see Brooke Crothers' coverage from June). Next year the company will begin shipping a new Opteron model currently code-named Magny-Cours (after a racetrack in France). Magny-Cours will consist of two Istanbul chips in a single package, with twice as many DRAM interfaces to support the new processor's increased performance.

AMD also teased the audience with another mention of a new processor core design that has been under development there for several years: "Bulldozer," which is now targeted at 32nm process technology. This new core will incorporate new x86 instruction-set extensions which will probably not be adopted by Intel (a strategy that reminds me of AMD's old 3DNow extensions).

But saving the best for last--best, that is, from the perspective of anticipated sales--Intel's talk on Nehalem-EX showed just how far Intel has been able to push the technology envelope for high-volume servers.

Nehalem-EX is an eight-core version of the existing quad-core Nehalem design. The new chip also has 24MB of L3 cache done in old-school SRAM. By my calculations, about 60 percent of the chip's 2.3 billion transistors are in this cache alone.

Nehalem provides four links to external DRAM buffer chips supporting two DDR3 DRAM interfaces each (much like the Power7 solution) and four QuickPath Interconnect links that provide direct "glueless" connections for up to eight-processor systems (64 cores, 128 threads). Intel is also working on an external Node Controller chip for systems with up to 2,048 Nehalem-EX processors.

The aggregate bandwidth numbers for Nehalem aren't as mind-boggling as those for Power7, but they're still far beyond anything available for PC-architecture servers today. Based on the presentation, I estimate Nehalem could boast over 85 GBps of peak memory bandwidth and 100 GBps of chip-to-chip bandwidth, some of which must be allocated to I/O.

I expect the raw number-crunching performance of the Nehalem-EX cores to be roughly on the same level as Power7's cores. The lower ratio of bandwidth to processing power for Nehalem-EX reflects a different design target, not a design shortfall--and most importantly, a much lower selling price. There will presumably be versions of Nehalem-EX priced similarly to existing Xeon MP products, which currently top out at $2,301 each in small volumes, but that's a very reasonable price to pay for the market's most advanced x86 server processor.

I just saw an interesting piece over on the EE Times website (here) written by David Carey, president of Portelligent, an analyst firm well-known for doing teardowns of popular electronic products.

The XO-1 laptop from the OLPC Foundation

(Credit: OLPC Foundation)

Here are some of the key points I learned from the article:

According to Portelligent, the LCD is pretty similar to previous transflective LCDs (that is, LCDs that can work from transmitted backlight or reflected ambient light) except for the arrangement of the color subpixels. I've seen nothing particularly remarkable about mine. It offers poor off-axis image quality, as I described in my first look (here), and doesn't even work as well as the transflective LCD on my Sony DSC-T1 digital camera, which displays color in reflective mode. The XO-1's display may be more power-efficient than previous transflective displays, but Carey doesn't report information on that point in this article.

Portelligent did report whole-system power figures, though, seeing 6W to 7W of power consumption during "full-tilt operation," whatever that means, exactly. In my own brief tests I saw power consumption in the range of 4W to 5W when the machine was doing almost nothing. Any machine with such a narrow range of power consumption isn't exhibiting good power management.

Carey's article (and a great illustration, here) goes on to provide some good information about the specific chips found in the unit they took apart. The article is also an advertisement for a full report Portelligent sells with all the details of the teardown, but I don't mind that at all in cases like this where the free article stands on its own.

Earlier this week, Apple bought a Silicon Valley chip-design firm named PA Semi for a reported price of $278 million.

Apple will get four things for sure:

• People-- PA Semi has a medium-size team that knows how to work together and produce chips.
• Processes-- PA Semi has design tools and procedures that can be used to design new chips.
• Patents-- PA Semi did some unique innovative design work for its chips, and there must be some interesting patents on this work.
• A product-- the PWRficient 1682M, which has found a few design wins in military electronics and other systems.

And perhaps a couple more:

• A PowerPC architectural license-- which would allow Apple to make PowerPC processors
• Protection against litigation-- if Apple made promises to PA Semi or its investors, PA Semi's inevitable failure as a standalone company might have resulted in legal action against Apple.

Okay, let's take these one at a time.

PA Semi's people are chip designers. Apple is not a chip company. It used to make chipsets for its PowerPC-based systems, but now that Apple is getting processors and chipsets from Intel, it has much less need for chip designers.

On top of that, PA Semi's people are designers of high-complexity, medium-power microprocessors with unusual custom I/O interfaces. These skills are even less useful to Apple, at least for its current product line.

The same reasoning applies to PA Semi's design processes. They just aren't what Apple needs right now.

Whatever patents PA Semi might have (I haven't looked into this) are probably not very useful to Apple, since it doesn't make processors and doesn't need this kind of legal leverage over its processor vendors. If the deal was primarily about patents, the purchase price would be completely unreasonable; no untested patent portfolio is worth anywhere near $278 million.

PA Semi's one product isn't selling in high volume and never will. I assume the company was working on future products but there's no reason to assume these would be any more successful.

The PowerPC license would only be valuable to Apple if it planned to make PowerPC processors. It has no particular reason to do that, and we don't even know if PA Semi's license is transferable to Apple.

Apple just doesn't need PowerPC chips. It moved the Mac line away from PowerPC. Intel is doing a reasonably good job of providing both low-power and high-performance x86 chips. PowerPC chips-- especially those like PA Semi's-- aren't a good fit in cellphones like the iPhone.

And that leaves us with litigation protection. Apple knows just what's at stake here. Apple invested in a previous PowerPC design firm, Exponential Technologies, and after Exponential went under, it slapped Apple with a lawsuit alleging that Apple deliberately sabotaged its efforts to sell chips to both Apple and Mac clone makers.

Could PA Semi possibly have a basis for a complaint against Apple? To me, this is nigh-impossible. How could Apple have made the same mistake with the Exponential situation so fresh in its institutional memory? The most generous thing I can say about this theory is that it isn't impossible.

But all of this leaves me with three choices:

• Apple is getting into the microprocessor business. That would be crazy.
• Apple made unwise promises to PA Semi. That's also crazy, but maybe slightly less crazy since Apple's done it before.
• I'm missing something. I have a pretty good idea what kinds of value are generated by a startup chip company... but it's possible.

And I'll throw in one bonus possibility raised by a friend of mine:

• Steve Jobs felt guilty that PA Semi didn't work out. No comment.

I like the Belkin company. I remember when Belkin was basically nothing but a cable company, and by that, I mean a company that makes electrical and optical cables. They made good cables, and still do.

The Belkin F6C550-AVR UPS.

(Credit: Belkin International)

But in recent years they have expanded into a wide variety of consumer products. I've always assumed Belkin's expansion was driven, at least in part, by the success of Monster Cable, which has made a lot of money selling expensive cables that (in my opinion, at least) are not always worth the price.

I imagine Belkin's engineering-oriented management deciding that selling well-engineered cables at a reasonable price would provide effective competition for Monster's marketing-focused strategy. I have no idea whether it went that way or not, but at least it's a sensible theory.

I sometimes wonder sometimes if Belkin isn't expanding too quickly. The company seems to be introducing new product lines faster than some companies introduce new products. That can lead to problems with product quality, customer support, and other critical business functions.

I've bought quite a few Belkin gizmos over the last several months, including the TuneTalk Stereo that I reviewed here back in September. (And also see this update to that review.) The TuneTalk Stereo was incorrectly marketed as compatible with the newly-released iPod classic, and although Belkin advised me that an upcoming iPod update would likely restore compatibility with the classic, my TuneTalk Stereo still isn't fully functional. Maybe the current models work fine; I don't know.

But I've continued to buy Belkin products, including several audio and video cables and a USB hub that I've had no problems with.

Unfortunately, just this week I've bought two Belkin products that didn't live up to my expectations. One was a fairly trivial little item, the F9H600-03 Surgemaster power strip. I needed a power strip so I could plug in some "wall wart" power supplies that wouldn't fit on the back of a UPS, and this product was the cheapest power strip with surge suppression available at my local Home Depot.

Since they were only about $8 each, I bought two of them; I figured I'd find a use for the other at some point. But when I plugged in the first one, it immediately made a loud snapping sound inside and released a puff of vile-smelling smoke. I was so eager to get it out of the house that I immediately wrapped it up in the Home Depot bag and took it out to the trash. I simply wasn't going to give it the chance to stink up my car in order to return it.

The other power strip worked fine, and I doubt it'll ever go bad. It's hooked up to an outlet on the year-old APC Smart-UPS that services my Power Mac G5, and three little power bricks are plugged into it in turn.

For many years, I've bought all of my UPSs from APC; I've always been happy with that brand. But when I went UPS shopping for one to put in my bedroom to protect the XO-1 from the One Laptop Per Child organization that has taken up residence on my nightstand, I bought a Belkin instead.

At least at my local Fry's on that day, the Belkin F6C550-AVR was a better deal than any of the APC UPSs in stock. The Belkin product was, by far, the cheapest model that offered an alarm disable function. APC's cheaper models don't have that feature, but if I'm going to be sleeping next to a UPS during an extended power outage, I need to be able to shut off the alarm.

Anyway, I got the Belkin model home and hooked it up to my XO-1, my alarm clock, the charger for my Palm Treo, and a 20W flourescent lamp.

After it had been plugged in for a few hours, I tested it by unplugging it from the wall. And wouldn't you know it, the flourescent lamp-- a cheap model without a starter or ballast, admittedly-- shut off for a second then turned itself back on. Subsequent testing produced the same result. That's clear evidence that the switchover time from AC power to the internal battery isn't as short as it ought to be. I tested one of my APC UPSs with the same lamp, and the lamp never flickered.

I didn't return this product either, since it meets my simple needs in this application. The XO-1 won't notice a brief power interruption since it has its own battery, but I wouldn't use the F6C550-AVR to protect a desktop computer.

Three incidents, especially when they involve unrelated problems in different products over the space of five months, don't really establish a trend. But I have to admit I'm going to be a little less likely to buy Belkin products going forward.

Except the cables. I still like Belkin cables.

I enjoy reading the personal blogs of Scott Adams (the creator of Dilbert) and John Dvorak (PC Magazine columnist and host of Cranky Geeks), but I don't expect to learn anything there. The entertainment is value enough.

Today, however, I was surprised to see these two gentlemen linking to the same story on Next Energy News covering Toshiba's announcement of a "200 kilowatt" nuclear reactor only "20 feet by 6 feet" in size. Such a reactor could be installed in a garage-sized building and shared among the houses on just one residential block, the apartments in one large building, or a single good-size corporation headquarters. With maintenance-free operation and the price of the generated energy estimated at 5 cents per kilowatt-hour, this announcement appeared to undermine the usual arguments against nuclear power.

Run through the basic numbers, as one commenter on Dvorak's blog did, and you come out with annual operating costs around $87,500 and a total cost over 40 years of about $3.5 million. Heck, never mind powering the neighborhood; I know a lot of people in Silicon Valley who'd build one into their houses.

Alas, the rest of the important numbers--the ones not covered by Next Energy News--don't work out so well for the Valley's wealthy. According to some information I found on the Encyclopedia of Earth, the reactor in question is called the Rapid-L, and the 200-kilowatt electrical output is just a small part of the reactor's thermal power production of 5 megawatts.

So even if your McMansion is filled with enough electronic gizmos to use up that 200-kilowatt power rating, there's no way it can dissipate 5 megawatts of thermal power. That's enough to heat over 200 homes during a 27° F (-3° C) cold snap. You'll just have to share.

But if you're one of the Silicon Valley multimillionaires who built mansions in Idaho because you love fly-fishing, you may be in luck; just divert part of your trout stream to provide cooling water for the reactor. You'll never need to turn off that big plasma TV again, and even the fish will be happier in the warmer water.

I'm not going to try liveblogging the keynotes today as I did yesterday. There's just never enough content in IDF keynotes to justify the effort.

The first keynote for Day 2 here at IDF was from Dadi Perlmutter, senior VP and general manager of Intel's Mobility Group. His theme was "Breaking the Barriers of Mobility."

He presented the results of a survey that showed the top needs of mobile computer users:... Read more

"Power to the people" was a popular rallying cry among anti-establishment activists in the 1960s.

"Power from the people" appears to be the latter-day equivalent.

The theory behind the slogan is that humans move around a lot, and the only result of all this motion is that the humans end up in a different place.

According to some, this isn't good enough.

The MIT News reports that two MIT graduate students in architecture have proposed to extract energy from the motion of humans through ... Read more

When I posted my review of Belkin's TuneTalk Stereo, a representative of APC commented on the story to suggest that I might want to use APC's Mobile Power Pack to extend the iPod's battery life when recording.

I started to reply to describe the gizmo I designed to power my iPod from a spare PowerBook battery-- a little clip that connects the battery terminals to an iPod dock cable. This is what I used to use to power up my old 4th-generation iPod while recording long conference sessions with the Belkin Voice Recorder for iPod.

But even as I was typing, I realized ... Read more