AMD

Quest for power efficient servers leads vendors to PC chips

What was once a battle over who could offer the fastest, most powerful server has turned into a competition over which servers can operate on the fewest watts, leading vendors to put PC chips in x86 systems.

"I just can't do it captain, I don't have the power!"

This trend is the result of data center power constraints; IT folks simply don’t have abundant quantities of power to supply their servers any more, but still need to add more compute capacity somehow.

So, server vendors are on the hook to offer systems that operate on very little power, and every vendor wants the right to say they offer “the worlds most efficient” server. But it appears they hit the limit with the level of efficiency today’s x86 server chips can offer, so some vendors have moved on to PC chips.

For example, last week, Dell launched new servers with Taiwan-based Via Technologies Via Nano processors, which the companies say offer the best power efficiency of any processor on the market today. Prior to that, Intel’s Atom processors for notebook computers were launched in SuperMicro servers.

My response was, really? PC chips in servers? Sounds great if you only plan to run Tetris on your servers, because when you choose lower watt chips, you trade off performance.

But it appears certain markets are willing to make that compromise. According to Dell, their new Via Nano-based systems are designed for “hyper-scale customers in the search engine and Web hosting businesses…who typically choose general purpose 1U servers or low-end tower servers, and make compromises around the density, power, and/or manageability aspects associated with these alternatives.”

It will be interesting to see how far these PC-chip servers go, who adopts them and how Intel and AMD respond to the Via Nano, product-wise. Will Intel and AMD try to leapfrog Via’s Nano chip with something that consumes even less power? I’m going to take a guess and say, hell yeah. But really, how low can they go?

Intel breaks another record; biggest anti-trust fine ever

Intel just broke another record - this time for getting the largest anti-trust fine ever.

The European Commission (EC) slapped Intel with a $1.45 billion fine for violating EC Treaty antitrust rules against engaging anti-competitive practices that excludes competitors from the market.

The European Commissioner for Competition Policy stated throughout the period covered by the decision, Intel held at least 70% of the worldwide market in x86 server CPUs, and used anti-competitive practices to hold that position.

“The fact that Intel had such a large market share is not a problem in itself. What is a problem is that Intel abused its dominant position. Specifically, Intel used illegal anti-competitive practices to exclude essentially its only competitor, and thus reduce consumer choice, in the worldwide market for x86 chips,” the commissioner, Neelie Kroes, told press. “Given that Intel has harmed millions of European consumers by deliberately acting to keep competitors out of the market for over five years, the size of the fine should come as no surprise.”

The EC found that Intel gave rebate to computer manufacturers who bought all, or almost all, of their x86 CPUs from Intel. Intel also made direct payments to a major retailer on the condition that it only sell computers with Intel x86 CPUs.

Intel also paid computer manufacturers to halt or delay the launch of specific products containing competitors’ x86 CPUs and to limit the sales channels available to these products, according to the EC.

Intel also faces another anti-trust lawsuit, filed by AMD for similar anti-competitive practices in the U.S. The court date for that trial is in February 2010.

As expected, Intel’s CEO Paul Ottellini denied any wrongdoing and is appealing the decision. In a statement, Ottellini said, “As we go through the appeals process we plan to work with the Commission to ensure we’re in compliance with their decision… there should be no doubt whatsoever that Intel will continue to invest in the products and technologies that provide Europe and the rest of the world the industry’s best performing processors at lower prices.”

And some U.S. based legal pros issued statements today saying the EU’s fine was far too harsh.

“The EC’s use of huge fines against market-leading firms - fines calculated from a firm’s world-wide sales, not from harm to European consumers - discourages aggressive competition that benefits consumers,”  Ronald A. Cass, Chairman, Center for the Rule of Law, said in a statement. “Consumer harm should be the concern for competition law, and here instead consumers saw sharp declines in cost and increases in product quality - even Intel’s complaining rival, AMD, enjoyed historic success during the period it claims Intel’s actions foreclosed competition.”

But the manufacturers concerned by Intel’s conduct in the EC case - Acer, Dell, HP, Lenovo and NEC - aren’t playing the violin for Intel right now, and reports say Intel’s closest competitor, AMD, is celebrating the EC’s decision.

And the EC’s commissioner doesn’t seem to feel bad about the massive fine either. In his closing statement to the press, he drew attention to Intel’s latest global advertising campaign, “Sponsors of Tomorrow,” in which Intel invites visitors to add their ‘vision of tomorrow’ to their website.

“Well, I can give my vision of tomorrow for Intel here and now: “obey the law,” Kroes said.

As large a fine as $1.45 billion is, it’s really a drop in the bucket for Intel; they reported $7.1 billion in revenue for the first quarter of 2009 alone,  so I doubt this will have any affect on their ability to churn out CPUs on the tick-tock cycle. The real issues for Intel is the tarnish the EC’s decision puts on them and that it takes the focus away from their technology - two side effects that are sure to help the competition gain some ground in the CPU market.

Google’s server recipe no longer secret

Looks like Google has finally spilled the beans on its server design, which the company has kept secret for years.

According to a report, Google let everyone see its design at a conference this week. The system was a 3.5 inch thick 2U system with two processors, two hard drives, and eight memory slots mounted on a motherboard built by Gigabyte. Google uses x86 processors from AMD and Intel, and the servers are powered by 12-volt batteries in case there’s a problem with the main source of electricity, according to the report.

We already knew that Google uses 12-volt power supplies for its servers, which the company says is more than 90% efficient compared with typical server efficiencies at or below 70%.

For the most part, Google runs these servers not out of brick and mortar data centers, but from those shipping containers that have become all the rage with vendors like Sun Microsystems , IBM, HP and Rackable Systems in recent years. Google was putting data centers in mobile containers long before those guys, and typically uses standard 1AAA shipping containers to house around 1,160 servers each, according to the report.

Intel set to release “Nehalem” Xeon processors

Intel may launch its next generation multi-core Xeon processors, code-named Nehalem, on Monday.

The company sent out invitations to a live webcast on March 30 “for the launch of a groundbreaking new server architecture.” If that doesn’t give it away, some server vendors have already announced products based on the Nehalem processors, including Cisco, which will use the Intel Xeon CPU’s in their upcoming Unified Computing System’s blade servers.  Rackable Systems already introduced CloudRack systems based on Nehalem, and Dell is expected to introduce Nehalem-based systems this week.

In earlier disclosures about Nehalem chips for x86 servers, Intel said the processor will have two, four or eight processing cores and provide better scalability than previous generations. It will also have scalable cache sizes and simultaneous multithreading, or Hyper-threading, which is already available on other Xeon processors.

While Intel prides itself on introducing multi-core processors at a faster pace than competitor AMD, some of the most significant enhancements to the new Xeon processor have existed in AMD chips for years.

For example, one of the major changes with Nehalem is integration of the memory controller into the CPU. This replaces the legacy Front Side Bus, which is a known culprit in traffic bottlenecks issues. AMD has been offering an integrated memory controller –called Direct Connect Architechture — in its Opteron CPUs for years now.

Another feature in Nehalem is the QuickPath Interconnect (QPI), which will give the chip faster access to a lot more bandwidth. This feature is similar to AMD’s HyperTransport technology, which has been around for a number of years as well.

That said, by adding QPI and an integrated memory controller, Nehalem will have access to a lot more bandwidth than its predecessors without relying on tons of cache memory, according to an ARS Technica report on Nehalem.

More importantly, what all of this means for end users is significantly better performance for applications that can take advantage of multithreading and multiple processing cores.

Gartner warns users of multi-core processing hazards

Gartner, Inc. reported this week that data centers are being attacked by processing cores at a rate their software, operating Systems and applications can’t handle.

“The relentless doubling of processors per microprocessor chip will drive the total processor counts of upcoming server generations to peaks well above the levels for which key software have been engineered,” Gartner reported. “Operating systems, middleware, virtualization tools and applications will all be affected, leaving organizations facing difficult decisions, hurried migrations to new versions and performance challenges as a consequence of this evolution.”

Wow. Sounds serious, huh? Maybe I am simplifying things a bit here, but doesn’t it make sense to upgrade to quad-core chips only if you have applications that can benefit from those chips? Otherwise, why spend the money?

I suppose I am being naive. Perhaps CPU cores are like crack, and once you get a taste of the power in a dual core chip, you want four cores, and then six, and will keep adding more and more cores until your systems are balls to the wall and your software implodes. It’s a vicious cycle, man.

In all seriousness though, people should be aware that throwing cores at applications does not automatically equal better performance; it’s been reported time and time again on SearchDataCenter.com since 2007 that not all your apps can use mutilple cores, because they aren’t written for paralellism.

According to Gartner, “the impact [of putting apps that aren't written for parellelism on multi-core chips] is akin to putting a Ferrari engine in a go-cart; the power may be there, but design mismatches severely limit the ability to exploit it.”

In fact, software developers are doing their best to design products that can take advantage of multiple cores, but find it hard to keep up with the tick-tock advancement model of Intel Corp. and AMD.

Many apps are designed to run on just one core, and work just fine in that one core. In this case, the software doesn’t know what to do with more than one core, and will actually run slower on multi-core chips. Of course, the processor makers don’t advertise this point.

“It’s important to understand that if the software developer doesn’t do something, the majority of software applications will run on a single core. The application will not leverage the multiple cores available and, in fact, the application may even get slower,” said Ray DePaul, president and CEO of RapidMind Inc., in Waterloo, Ont. “There is talk about 80-core processors (from Intel) now and this is scary to software developers. They can’t wrap their head around how that is going to work.”

Meanwhile, organizations get double the number of processors in each chip generation, approximately every two years, according to Gartner. Each generation of microprocessor, with its doubling of processor counts through some combination of more cores and more threads per core, turns the same number of sockets into twice as many processors. “In this way a 32-socket, high-end server with eight core chips in the sockets would deliver 256 processors in 2009. In two years, with 16 processors per socket appearing on the market, the machine swells to 512 processors in total. Four years from now, with 32 processors per socket shipping, that machine would host 1,024 processors,” Gartner reported.

There are apps inherently designed to use multiple cores, like heavy workloads used in virtualization, Java, expansive databases and complex enterprise resource planning (ERP) applications. Apps like these use more than one core and perform up to 50% better on multi-core chips, according to analysts.

So, heed Gartner’s warning and don’t go core-crazy; do your research and make sure the apps you run on multi-core chips before you take money from your tight IT budgets to buy them.

.

Intel lowers Xeon prices on declining CPU sales

Intel Corp. dropped prices up to 40% on some of their Xeon processors this week following the release of some ugly financials for the fourth quarter of 2008 - so if you are in the market for an upgrade (and have any money in your IT budget) now’s the time.

The new prices on server CPUs include a drop on Intel’s quad-core Xeon X3370 from $530 to $316 (40% drop) and the 45nm quad-core X3360 from $316 to $266 (16% price cut).

According to Intel spokesperson Nick Knupffer, the company regularly makes price cuts throughout the year and that is what the recent price change reflects.

But, one would presume that these cuts aren’t just a routine act; Intel’s Q408 profit plummeted 90% and the company predicts even weaker conditions ahead. Intel, which twice lowered its fourth-quarter sales forecast, reported quarterly sales fell 23% from a year earlier to $8.23 billion, the Wall Street Journal reported.

Of course, Intel isn’t alone in the gutter; AMD’s CPU sales also slummped in 2008 and it expects sales to decline further through the first quarter of this year due to poor economic conditions.

Despite AMD’s reported a $1.42 billion fourth-quarter loss, a spokesperson claims there aren’t any price cuts on the docket for its server CPUs. “AMD moves prices on their products, and in this case speaking for Opteron, as required by the market. There is no regular price move schedule in place.”

.

Rackable Systems updates its most popular half depth server

Rackable Systems, Inc. today announced the new C2005 server, designed with more expansion slots and a choice of SAS or SATA II hard drives for greater flexibility and configurability.

The C2005 is an updated version of Rackable’s top selling half depth 2U server product, the C2004, which offers density comparable to industry standard 1U servers.

“The C2005 introduces an unparalleled level of flexibility compared to anything we have ever offered,” said Geoff Noer, VP of Product Management.

The new C2005 includes both Intel and AMD quad-core processor options and offers up to five low-profile expansion slots and a sixth slot available in all configurations (often a x16 PCI-E slot), Noer said.

“A lot of new technologies, like virtualization, benefit from having additional expansion slots for connectivity. In a 1 U server you typically only see one or two expansion slots,” Noer said.

The C2005 also includes up to 10 x 2.5” or 5 x 3.5” SAS or SATA II hard drives – or a mix or both. “An emerging challenge is to address the need to support both 3.5 and 2.5 inch, and providing both in a single platform” adds appeal, Noer said.

“Configurations that would have had to been in a 3U server or larger, for the number of expansion slots, can now be done with a 2U half depth. The density level is much higher and suited for technologies like virtualization,” Noer said.

Noer said the C2005 is well suited for scale-out computing across many markets, especially Internet, Enterprise, and HPC customers.

Up to 44 C2005 servers fit into a single rack. The server is also available in AC or DC power, and Noer said both options are in high demand. “The advantage of DC is largely around increased reliability of the solution. We can also supply AC to the cabinet and DC power to all of the servers inside the cabinet,” through Rackable’s AC-to-DC X86 rack-level rectification technology, which the company has offered since 2003.

The server is available for order immediately. Rackable did not give pricing because the systems are built to order, and pricing depends on configurations Noer said.

AMD holds press conference to bash Intel before IDF

Advanced Micro Devices (AMD) held a Web-based press conference Aug. 15 to formally attack Intel Corp. prior to the Intel Developers Forum (IDF) in San Francisco next week.

When AMD invited me to listen to a webcast regarding the IDF forum, I expected to hear some news; maybe their 45nm processor Shanghai would be coming out earlier than expected to make up for Barcelona’s delay? That would be big and would help AMD make up some ground in the processor product release wars against Intel.

Instead, the conference served as a means for AMD to plant the seeds of skepticism in the minds of IDF attendees before the conference. AMD executives spent the hour call marketing their existing processing and graphics technologies and bashing Intel.

Randy Allen, the SVP and GM for AMD’s computing solutions group, cited benchmarks showing AMD’s Opteron processors in good light, like the SPECweb 2005 benchmark showing Opteron Model 2356 and Model 8356 hold the top x86 Web performance records for two-and four-processor servers.

Allen also touted AMD’s virtualization assist technology, AMD-V with Nested Page Tables, which recieved high praise from a VMware engineer recently, and he noted that quad-core Opteron is being used in a total of seven of the top performing systems in the most recent Top 500 Supercomputers list, including the No. 1 IBM’s RoadRunner.

“We have our swagger back,” Allen said.

He failed to note, however, that AMD’s Opteron chips were used in only 56 systems (11.2%) on the list, which is down from 78 systems six months ago. Intel processors were used in 74.8% of the world’s supercomputers (about 374 systems), up 4% from six months ago.

When a reporter raised this issue during the press conference, Allen said that having Opteron used in the top performing computer and systems high on the list is more notable than the slip in the number of total systems on the top 500.

In addition to hyping AMD products, Allen also spent plenty of time directly attacking Intel, saying the company has an easy time innovating because it simply mimics AMD’s work.

“Intel adopted our power efficiency technology, our multicore technology and you will see them copying the Direct Connect Architecture and HyperTransport technologies we developed five years ago. … Imitation is the sincerest form of flattery, but it is also annoying.”

Surely, IDF conference goers will hear similar hype about Intel products from Intel executives next week and negative remarks about AMD.

The only mention of AMD’s processor on deck, 45-nanometer chip code-named Shanghai, was that it is scheduled for release later this year and will be delivered on time. Shanghai will consume 20% less power at idle than Barcelona and will have 6 MB of L3 cache (compared with Barcelona’s 2MB of L3 cache).

All in all, the press conference simply re-stated old AMD news.

Thanks for the re-cap, AMD. That is an hour of my life I’ll never get back. And if you just finished reading this blog, hopefully it’s only a few minutes of your time that you can never get back.

Facebook relying on Intel Xeon processors in Data Center build-out

The social networking website Facebook is building out its data center infrastructure using Intel Corp. processor-based systems and plans to deploy thousands of Intel Xeon processor-based servers over the next year to help accommodate its rapid growth, the two companies announced last week.

Intel will also collaborate with Facebook to determine the best configurations for its server and software using Intel processors, taking into account energy efficiency and performance.

Over the past several months, Facebook tested and benchmarked a number of server platforms and scenario, and ultimately selected the Intel Xeon 5400 series quad-core processors for its round of new deployments that begin in July.

When Facebook was contacted for more information on the systems and processors it has tested, why it chose Intel over AMD and other questions about data center infrastructure plans, it refused comment.

That said, Intel’s press statement had the following quote from Jonathan Heiliger, vice president of technical operations at Facebook; “Intel has demonstrated that the performance of their systems can help Facebook scale our infrastructure and continue to deliver the best experience to users around the world.”

“When you are responsible for providing a fast, high-quality experience to more than 90 million people worldwide, every ounce of efficiency matters,” Heiliger said in the statement.

Also, since Facebook’s applications are mostly built on open source technologies, the companies stated that some of the insights from this collaboration may be contributed back to the open source community, benefiting other companies that use open source technologies.