As has already been pointed out, KASY0 has better price/performance on Linpack. Also, the Space Simulator at Los Alamos was built over a year ago from Intel P4 hardware, and is currently listed at #88 on the TOP500 list, so it certainly qualifies as a supercomputer. It acheived 757.1 Gflops on Linpack, and cost $484,000 in September 2002. So, that's works out to a Linpack Tflop per 1.56 million, vs the current (improved) result of 8.2 Tflops/5.2 million, or 1.58. So, the cluster price/performance is about the same
between the brand new G5 Big Mac hardware, and year-old Intel P4 hardware.
The cluster is certainly an important milestone, though. The days of being locked in to buying a
commercial architecture that was designed three years ago and costs 10x too much are over.
If you have an existing conditional offer in the RHAT directed shares program (i.e. you got the letter), you can still re-confirm interest, at least as of 11:00 Eastern Time. You will have to get a "broker" on the phone who will take the re-confirmation manually. I was on hold for about 15 minutes. If you were not part of the directed shares arrangement, you are SOL at this point.
Paul Ginsparg's e-print server began as an automated e-mail archive of high-energy physics theory papers in 1991. xxx.lanl.gov (now also known as arXiv.org is now the primary source of online literature in Physics and Mathematics.
So, where is MILO for the 21264? How about porting glibc2 to DEC Unix? Then we can use those statically linked binaries compiled with DEC's fast compilers on our Linux boxes without violating the copyright on their C library.
The people who have built Linux/Alpha clusters have been asking for a few simple things for two years now. The message from above has always been "We're thinking about it." I'll take Compaq's commitment to Linux seriously when I see a continuous contribution to Linux software technology. The release of the math library was a giant step in the right direction. Let's see more statements like that, not stuff that was cooked up in the marketing department.
Have a meeting and invite the people managing the systems to contribute ideas. Nothing will cause more trouble than to tell people who are managing machines in such an environment that "they don't know what they are doing." If you actually take some time to learn the reasons why the problems you are seeing exist, you will learn that there are probably more skilled and resourceful people managing the machines than there are people who don't care about security and don't know what they are doing.
As one anecdote, I was told my machines were insecure after a remote probe. It turns out the software used to probe the machines was interpreting the message from tcp_wrappers "You are not allowed to connect from IP www.xxx.yyy.zzz" as a successful login. The people running the probe were so sure that they were checking machines run by idiots, they didn't even look at the logs themselves to check what was really happening.
Slashdot Mirror
The cluster is certainly an important milestone, though. The days of being locked in to buying a commercial architecture that was designed three years ago and costs 10x too much are over.
Mike
Paul Ginsparg's e-print server began as an automated e-mail archive of high-energy physics theory papers in 1991. xxx.lanl.gov (now also known as arXiv.org is now the primary source of online literature in Physics and Mathematics.
The people who have built Linux/Alpha clusters have been asking for a few simple things for two years now. The message from above has always been "We're thinking about it." I'll take Compaq's commitment to Linux seriously when I see a continuous contribution to Linux software technology. The release of the math library was a giant step in the right direction. Let's see more statements like that, not stuff that was cooked up in the marketing department.
As one anecdote, I was told my machines were insecure after a remote probe. It turns out the software used to probe the machines was interpreting the message from tcp_wrappers "You are not allowed to connect from IP www.xxx.yyy.zzz" as a successful login. The people running the probe were so sure that they were checking machines run by idiots, they didn't even look at the logs themselves to check what was really happening.