| 17 Jun, 2010
Everybody likes to talk about processor power when it comes to new computing gear. Whether it's the hot new server at the office or the sleek machine in the den at home, the first question most people ask is: "What's it got in it? Xeons? Itaniums? Dual-core? Quad-core?"
In a way, it's the computer geek's answer to the hot-rodders of my youth. While they talked about horse power and overhead cams, we have clock speed and megs (now gigs) of RAM.
But now that processors have gotten so powerful and the ability to pool them has been simplified, is it time to start wondering whether the old ways of placing value on new equipment are no longer valid? InformationWeek's Alexander Wolfe posed that very question this week with a look at the new x86 devices. For one thing, does the ubiquity of commodity servers shift the value proposition away from the processor to more specific functions like virtualization, management and, naturally, power and cooling efficiency? It seems that a good number of enterprise executives are thinking along these lines. That magazine's most recent survey pegged consolidation and virtualization as the top two drivers for server updates.
Server designers are, in fact, looking beyond the Xeon/Opteron box to devise new lines optimized for specific enterprise applications. A company called SeaMicro just released a low-power, small form-factor server-on-motherboard design using 512 Intel Atom processors. The company says it can provide a more efficient platform for such in-demand functions as cloud-based services and Web transactions, allowing enterprises to concentrate their higher-end x86 machines on heavier workloads like database processing. The system also features a virtual I/O component that removes a good 90 percent of hardware from the board.
Of course, there will still be a market for devices that push the processing envelope. Intel is getting ready to ship a 32-core chip that combines x86 designs with specialized cores with enhanced parallel processing capabilities. And AMD is already in the channel with a dual hexacore die for server applications along with a single hex device for CPUs, with each core hitting 3.2 GHz.
It would seem, though, that the only users still focusing on clock speeds and overall processing capability are in the HPC market, where that x86/RISC/Itanium gap is growing ever-narrower. Increasingly, chip-level specs are becoming irrelevent for mainstream enterprise applications. And if some cloud proponents have their way, it won't be long before enterprise users care as little about computing hardware as they care about the kind of generator that provides their electricity.
Processors, then, truly have become the victims of their own success.
Nice article. There's an interesting follow-up blog post from the Dell HPC Team about the HPC industry's focus on CPU performance & clock speeds - worth a read: http://dld.bz/mveu
ReplyOopps - I posted the wrong link to the Dell blog post I reference. Here is the correct link: http://dld.bz/mvtd Sorry for the confusion!
ReplyPost a comment


Business IntelligenceBusiness performance information for strategic and operational decision-making
SOASOA uses interoperable services grouped around business processes to ease data integration
Data WarehousingData warehousing helps companies make sense of their operational data
Although processor speeds are still increasing rapidly, the push needs to be on the energy consumed by these power hungry processors. Just as the hot rods of the past are now being replaced by hyper-miling hybrids, processors must also adjust. This is especially true when you start to see how much it costs to power and cool some of these new machines. With power costs increasing and data centers running out of available power and cooling, it is more important to make your processors efficient than powerful.
Reply