Posted on 12/03/2003 6:21:59 PM PST by Swordmaker
22nd Edition of TOP500 List of World’s Fastest Supercomputers Released
MANNHEIM, Germany; KNOXVILLE, Tenn.; & BERKELEY, Calif. – In what has become a much-anticipated event in the world of high-performance computing, the 22nd edition of the “TOP500” list of the world’s fastest supercomputers was released today (November 16, 2003).
The Earth Simulator supercomputer retains the number one position with its Linpack benchmark performance of 35.86 Tflop/s (“teraflops” or trillions of calculations per second). It was built by NEC and installed last year at the Earth Simulator Center in Yokohama, Japan.
The list of cluster systems in the TOP10 has grown impressively to seven systems. These systems are built with workstations or PCs as building blocks and often connected by special high-speed internal networks. The number of clusters in the full TOP500 grew also again strongly, now totaling 208 systems up from 149 six months ago. This makes clustered systems the most common computer architecture seen in the TOP500. The importance of this market can also be seen by the fact that most manufacturers are now active in this market segment.
The new TOP500 list, as well as the former lists, can be found on the Web at http://www.top500.org/.
The number two position is again held by the ASCI Q system at the U.S. Department of Energy’s Los Alamos National Laboratory. ASCI Q was built by Hewlett-Packard and is based on the AlphaServer SC computer system. With 13.88 Tflop/s, it was the second system ever to exceed the 10 Tflop/s mark.
The third system ever to exceed the 10 TFflop/s mark is Virgina Tech's X Cluster Institute measured at 10.28 TFlop/s. This cluster is built with the Apple G5 as building blocks. It uses a Mellanox network based on the new Infinband technology as interconnect.
The fourth system is also a cluster. The Tungsten cluster at NCSA is based on the Dell PowerEdge system with its Pentium4 Xeon processor and uses a Myrinet interconnect. It missed the 10 TFlop/s mark by only a tiny margin with a measured 9.82 TFlop/s.
The list of clusters in the TOP10 continues with the upgraded Itanium2-based Hewlett-Packard system, located at DOE's Pacific Northwest National Laboratory, which uses a Quadrics interconnect. The sixth largest system is the first system in the TOP500 based on AMD's Opteron chip. It was installed by Linux Networx at the Los Alamos National Laboratory and also uses a Myrinet interconnect.
The TOP10 finishes with the IBM SP systems at two other DOE national laboratories (Lawrence Livermore National Laboratory and the National Energy Research Scientific Computing Center (NERSC) at Lawrence Berkeley National Laboratory) ahead of another Pentium4 Xeon-based cluster also at the Lawrence Livermore National Laboratory. Gaining entry into the top 10 positions on the new list now requires achieving a Linpack performance of at least 6.6 Tflop/s.
Already, 131 systems report a Linpack performance exceeding 1 Tflop/s. Performance levels at all positions of the TOP500 list keep growing impressively, driven by Moore’s Law. The total combined performance of all 500 computers on the list exceeded the level of half a “petaflop/s” (or quadrillion of calculations per second). It is now 528 TFlop/s, compared to 375 Tflop/s just six months ago. The “slowest” system on the newest listing is now 403.4 Gflop/s (“gigaflops” or billions of calculations per second), compared to 245.1 Gflop/s six months ago.
The number of systems in the TOP500 list using Intel processors grew in the last six months from 119 to 189, signifying a major shift in this marketplace. With this increase, the Intel processor family is now the most dominant processor used in HPC systems. It is followed by Hewlett-Packard’s PA-RISC chips and IBM’s Power architecture.
In terms of total performance of all the installed systems, the latest TOP500 edition still shows IBM as the clear leader with 35.4 percent, ahead of HP with 22.7 percent and NEC with 8.7 percent.
Hewlett-Packard was able to hold on to its lead in terms of the number of systems on the list by a small margin. A total of 165 systems on the TOP500 list were installed by Hewlett-Packard, compared to 159 systems by IBM. This time SGI is third in this category with 41 systems. No other company was able to gain more than 6 percent in either of these two categories.
The Cray X1 also appears on the list with 10 systems listed. It is the only computer system currently being built in the U.S. that uses vector processors, as the Earth simulator does. Therefore, the Cray X1 is sometimes seen as a possible U.S. answer to the Earth Simulator. The highest ranked Cray X1 appears on rank 19.
Nine of the TOP10 systems, 50 percent of all 500 systems, and 57 percent of the total performance are installed in the United States. Also, 90 percent of all 500 systems are produced in the United States. These market shares demonstrate the continued and clear leadership of the U.S. in both the production and use of high performance computing.
The TOP500 list is compiled by Hans Meuer of the University of Mannheim, Germany; Erich Strohmaier and Horst Simon of NERSC/Lawrence Berkeley National Laboratory; and Jack Dongarra of the University of Tennessee. The list was released today and will be presented in detail at the SC2003 Conference in Phoenix, AZ, which started Saturday, November 15, and continues through Friday, November 21.
# # #
Hmmm... PC based Pentium4 Xeon placed 4th, and Itanium2 based clusted placed 5th.
I guess that answers that question.
Wow...my Apple //c+ was faster than 2MHz :-)
Moore's law can still be applied. :D
I owned a NEC computer. This can't be the same company. -Tom
Oops... 2Ghz!! gadzooks!
For most of the machines, and all of the cluster ones, the idea is to break down the problem into smaller pieces, run each smaller piece on one of the CPUs, then combine the results together to come up with the answer.
Examples include: simulation of a nuclear bomb explosion (you can calculate each portion of the 3D space separately); finite-element analysis of the aerodynamic properties of a car or airplane's wing (split the surface into 1 inch sections and compute them separately); that sort of thing.
It doesn't make Freeping any faster, however :-)
What was the question? ;)
Actually, I'll be the first to pee in the pool here. I still have yet to see any neutral benchmarks, SPEC in particular, that show the 2 GHz G5 outperforming top of the line Xeons or I2s. At best so far, performance appears to be on par with Xeons on a clock-for-clock basis, which is what gives the edge to Intel for the moment - higher clock rates.
So I tend to discount the notion that the G5 is inherently faster than the 3 GHz Xeons per se, and I'll guess that the real difference is in the networking technology - many of the Intel systems are built on older Myrinet interconnects, and it seems likely that the newer Infiniband stuff is faster and/or more efficient than Myrinet or the Quadrics interconnects. That being said, what's most impressive about the VT cluster, to my mind, is the bang for the buck they achieved - #3 is pretty dang impressive considering how much (little) they spent to get there.
BTW, my alma mater ranks in the top 40. How about yours? ;)
I've got to upgrade my machine, it just seems so slow now.
Newegg has a nice special on an AMD 64 3200+ with ASUS board and CoolerMaster Heat Sink for $525 after a small rebate. That could get me started ! LOL!!!
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.