Computer Hardware FAQs
What is the most powerful super computer?
As of this edition, the largest planned supercomputer in the world is the International Business Machine’s Sequoia supercomputer. The statistics on the piece are boggling, even if you spend the time to understand the lingo. As the numbers go, the supercomputer is expected to run 20 petaflops. That means nothing unless you’re a computer science graduate, to the layperson that is more computing power than the rest of the top 500 supercomputers in the world combined. For a human reference, a common calculator needs 10 flops in order to seem instantaneous to humans. 20 petaflops is 20,000,000,000,000,000 flops. In reference to other supercomputers, it would be 20 times faster than the reigning champion. The whole system will consume 6MW of power, or the equivalent to a good sized business. Still, that is considered surprisingly efficient when you consider the power divided by the computing output, measured by performance per watt. It will deliver 3000Mflops/watt. Since the computer is supposed to drastically realign the computing landscape by bursting through the petaflop barrier, it is hailed by academia and funded by various bureaucracies and made by IBM. It is scheduled to go online in 2011 at the Laurence Livermore National Laboratory in Cal.
What are the world’s most powerful supercomputers calculating?
One category of academic research that is being uncovered with the advent of these highly capable machines is research into protein folding. Protein folding is a meshing of chemistry, biology and physics to understand the basic framework that causes protein strings to behave the way they do when introduced to another chemical, all of which takes place in a fraction of a second. The folding has been researched and is found to correspond with a protein hiding the most solvent portion of the string surrounded by the less solvent portions of the chain. The folding takes place in a millionth of a second, but the way that it does so is purely hypothetical. The chain folds according to the atomic layout of the string corresponding to the substance it encounters. Since the folding occurs in 3d in a fraction of a second, there are a nearly infinite number of plausible ways and steps that the chain moves in order to protect its most solvent portion. Laying out the accurate steps in a brain is also plausible with this technology and indeed, some supercomputers have been used to measure half the brain capacity of a mouse for one second, the computation was done 1/10 of the brain function for 10 seconds. Almost every field of science would like to utilize these computers for various purposes to test hypothetical models of their work. For that reason the U.S. Department of Energy is partially funding the research along with the Laurence Livermore National Laboratory and a $100 grant from Intel.
Is there a CPU configuration larger than a quad core?
Yes. Intel experimented with a 6 core processor configuration. The dilemma is that most programs are not tuned to run on 6 separate cores, complicating its work load by divvying up the workload instead of having it come pre separated into the caching system. Even though the Gulftown microprocessor has been shown to be about 50% faster than the equivalently clocked quad core systems, it uses less power. Over all the micro processor is scheduled for 130 thermal design power systems. For the multi core systems one of the most important indications of potential computing speed is the availability of memory upgrades and the communication between the nodes. The L3 communication between the hubs in the Gulftown processor is at 12 megabytes. The processors use a newly redesigned 32 nanometer wafer design in the manufacturing process which results in a smaller chip. There are variants of the chip that are specifically made for server applications. The two potential models are for the Xeon processors and come by the name of the Xeon 36xx or Xeon 56xx, the normal personal computer variant is titled the Core i7 9XX. The chip launched in the first quarter of 2010.