Visit additional Tabor Communication Publications
April 21, 2006
With the popularity of the 64-bit x86 architecture in the high performance computing market now established, a lot of us in the HPC community closely follow the rivalry between the two chip vendors, AMD and Intel. Rivalries are fun, especially when it's a "David and Goliath" story. But a lot is on the line. At a time when the demand for commodity clusters and blade servers is rapidly growing, these two companies have much to gain and just as much to lose.
For the past two years, AMD's Opteron has clearly dominated in both performance and power efficiency when compared to Intel's Xeon processor. But things may be about to change. Last month at the Intel Developer Forum in San Francisco, Intel introduced their new Core microarchitecture, which will be the basis for all x86 microprocessors for the next several years. The new microarchitecture will incorporate a variety of technologies to deliver higher levels of energy-efficient performance. With Intel's dual-core Woodcrest processor due to be released in the third quarter of this year, most analysts believe it will represent the first serious competition for the dual-core Opteron.
You have to give credit to Intel for remodeling the x86 microarchitecture so thoroughly. The fact that the company was willing to take that risk now says a lot about how seriously it perceives the competition from its smaller rival. In my recent conversation with the Intel CTO, Justin Rattner, he explained that the move to this new design started about four years ago when they realized that the power requirements for its processors was going to be intolerable in a few years. As Rattner told to me: "We didn't wake up six months ago or last year and say 'Wow, it's energy efficiency!' We knew we had to make this turn for some time."
I have no doubt that this was the original motivation to redesign the core, but if AMD hadn't led with its own success, Intel might not have felt compelled to make the architectural transition so quickly. With Intel's x86 market being eroded by its competitor, the company probably realized it needed to come out swinging in 2006, before any more damage was done.
With the introduction of the new Intel microarchitecture, the competition for x86 microprocessor superiority is going to be a lot closer, especially at the high end. Woodcrest will be built with the 65nm process, giving Intel an edge over the 90nm Opterons. On the other hand, the Direct Connect Architecture on the Opterons is still widely regarded as superior to the older front-side bus technology that was retained in the new Intel processors. Looking at power usage, the dual-core Opteron burns about 95 watts, while Woodcrest is expected to use around 80 watts. But when you factor in the external extra memory controller needed by Woodcrest (on-chip with Opterons), it's basically a wash. Overall, the perception seems to be that Woodcrest will at least break even with the current Opterons on energy efficiency and have slightly better overall performance.
If that's the case, it's not good news for AMD. The Opteron line is one of the company's most successful product lines. Last week AMD reported outstanding first quarter results, realizing year-over-year sales growth of 71 percent, driven in part by record Opteron sales. If the chip has to go head-to-head with Woodcrest, AMD is less likely to have great quarters like this one.
Of course AMD knows all this so is unlikely to just sit still. The company will probably ramp up its 65nm process as quickly as possible at their Dresden Fab36. And they'll continue to make other incremental improvements to their processor line to keep up the pressure on Intel.
Aside from that, AMD has built a certain amount of momentum with its larger Opteron-friendly OEM partners, such as Sun, HP and IBM. Cray, though not a volume user, is committed to AMD processors through the end of the decade. A number of smaller OEMs are also on board with the Opterons. So an Intel processor that just "catches up" to the Opteron is not likely to upset things too much in the short-term.
One thing that could alter the balance more quickly would be a price war. And in fact there have been rumors of Intel instigating a price war within the past few weeks. This is emphatically denied by AMD. A few days ago, I spoke with the AMD communication director, Dave Kroll, and he told me that Intel was lowering pricing on some of its older, lower performing products in order to move inventory, but no price war is taking place. Nor is there any expectation that one will occur. Says Kroll: "The price war is a complete myth manufactured by some folks, I think, on Wall Street. No executive here can actually ever recall us having a price war."
If true, that's good news. Ultimately, the race to the bottom is never very satisfying for either the vendors or their customers.
I started out by saying rivalries were fun. In fact, they're also useful. Among other things, the competition between AMD and Intel has fueled the rise of supercomputers based on commodity chips, making high performance computing accessible to a much wider community. It doesn't take a lot of imagination to see that the products offered by both of these vendors are better because of the rivalry. So if the the competition between the two chipmakers is about to intensify, I think it bodes well for the future of the industry.
Also in this issue:
* When chipmakers like Intel and AMD increase microprocessor energy efficiency, system designers use that extra efficiency to pack more chips into their machines. The data center gets more performance per square foot, but power usage and heat continue to rise. Recently I spoke with Richard Sawyer, Data Center Technology Director at American Power Conversion. He told me about some new solutions that are enabling data centers to deal with the increasing thermal loads. To give you an idea of where things are headed, one of Sawyer's favorite lines to his customers is: "If you've got hydrophobia, get over it." For those of you who spend a fair amount of your time in that wind-tunnel that passes for a data center, you might want to read this article.
* Unless they're producing 'Lord of the Rings' movies for Hollywood, companies in New Zealand typically can't afford multi-million dollar supercomputers. But with the help of its mild temperate climate, New Zealand has built a world-class agriculture and forestry sector that increasingly requires the latest bioinformatics tools to remain competitive. Kiwi entrepreneur, Leonard Bloksberg has developed a new genomics search engine which he says is 10,000 times faster than Blast, the industry standard. According to him, this enables industrial-strength genomic searches to be performed using a PC instead of a supercomputer. I recently got a chance to ask Bloksberg about this new technology and his startup company, Cartesian Gridspeed.
* Graphene. You may want to remember that name. It's a material that is being researched at Georgia Tech to explore its suitability for nanoelectronics. Researchers have measured electron transport properties in graphene that are comparable to those seen in carbon nanotubes. But unlike carbon nanotubes, graphene circuitry can be produced using standard lithography techniques, which allows a clear path to commercialization. Says researcher Walt de Heer: "We are at the proof-of principle stage, comparable to where transistors were in the late 1940s. We have a lot to do, but I believe this technology will advance rapidly." For the whole story, read the article in this week's issue.
-- Michael Feldman
Posted by Michael Feldman - April 20, 2006 @ 9:00 PM, Pacific Daylight Time
Michael Feldman is the editor of HPCwire.
No Recent Blog Comments
During a conversation this week with Cray CEO, Peter Ungaro, we learned that the company has managed to extend its reach into the enterprise HPC market quite dramatically--at least in supercomputing business terms. With steady growth into these markets, however, the focus on hardware versus the software side of certain problems for such users is....
Contributing commentator, Andrew Jones, offers a break in the news cycle with an assessment of what the national "size matters" contest means for the U.S. and other nations...
Today at the International Supercomputing Conference in Leipzing, Germany, Jack Dongarra presented on a proposed benchmark that could carry a bit more weight than its older Linpack companion. The high performance conjugate gradient (HPCG) concept takes into account new architectures for new applications, while shedding the floating point....
Jun 19, 2013 |
Supercomputer architectures have evolved considerably over the last 20 years, particularly in the number of processors that are linked together. One aspect of HPC architecture that hasn't changed is the MPI programming model.
Jun 18, 2013 |
The world's largest supercomputers, like Tianhe-2, are great at traditional, compute-intensive HPC workloads, such as simulating atomic decay or modeling tornados. But data-intensive applications--such as mining big data sets for connections--is a different sort of workload, and runs best on a different sort of computer.
Jun 18, 2013 |
Researchers are finding innovative uses for Gordon, the 285 teraflop supercomputer housed at the San Diego Supercomputer Center (SDSC) that has a unique Flash-based storage system. Since going online, researchers have put the incredibly fast I/O to use on a wide variety of workloads, ranging from chemistry to political science.
Jun 17, 2013 |
The advent of low-power mobile processors and cloud delivery models is changing the economics of computing. But just as an economy car is good at different things than a full size truck, an HPC workload still has certain computing demands that neither the fastest smartphone nor the most elastic cloud cluster can fulfill.
Jun 14, 2013 |
For all the progress we've made in IT over the last 50 years, there's one area of life that has steadfastly eluded the grasp of computers: understanding human language. Now, researchers at the Texas Advanced Computing Center (TACC) are utilizing a Hadoop cluster on its Longhorn supercomputer to move the state of the art of language processing a little bit further.
05/10/2013 | Cleversafe, Cray, DDN, NetApp, & Panasas | From Wall Street to Hollywood, drug discovery to homeland security, companies and organizations of all sizes and stripes are coming face to face with the challenges – and opportunities – afforded by Big Data. Before anyone can utilize these extraordinary data repositories, however, they must first harness and manage their data stores, and do so utilizing technologies that underscore affordability, security, and scalability.
04/15/2013 | Bull | “50% of HPC users say their largest jobs scale to 120 cores or less.” How about yours? Are your codes ready to take advantage of today’s and tomorrow’s ultra-parallel HPC systems? Download this White Paper by Analysts Intersect360 Research to see what Bull and Intel’s Center for Excellence in Parallel Programming can do for your codes.
Join HPCwire Editor Nicole Hemsoth and Dr. David Bader from Georgia Tech as they take center stage on opening night at Atlanta's first Big Data Kick Off Week, filmed in front of a live audience. Nicole and David look at the evolution of HPC, today's big data challenges, discuss real world solutions, and reveal their predictions. Exactly what does the future holds for HPC?
Join our webinar to learn how IT managers can migrate to a more resilient, flexible and scalable solution that grows with the data center. Mellanox VMS is future-proof, efficient and brings significant CAPEX and OPEX savings. The VMS is available today.