… because it fees so good when I stop. Or so goes the old joke. A long while ago, I mentioned we have a customer who self-inflicts pain by spending too much time using root for day to day work. We advise against this. No good can possibly come of this, only bad. Like the last time when a key-logger grabbed the root password as some windows user was typing it in.

NSF is funding a 2x10^5 processor monster machine at NCSA. At $208M, each dollar will by you 4.8 MFLOP (4.8x10^6 FLOP). Hmmm…. Assuming a quad core CPU would be able to provide (in theory) 32 GFLOP (4 cores x 8 GFLOP/core), you would need 31,250 units to provide this … (125000 cores). There are some interesting things about this machine. Very interesting … not just the price tag or the estimated sustainable performance

New unit being built for another customer [JackRabbit orders are multiplying like bunnies] [root@jackrabbitm ~]# dd if=/big/big.file of=/dev/null ... 10000+0 records in 10000+0 records out 83886080000 bytes (84 GB) copied, 51.8149 seconds, 1.6 GB/s [root@jackrabbitm ~]# cat /proc/meminfo | grep MemTotal MemTotal: 33011556 kB Yes. We did just stream a file more than 2x the size of ram (32 GB) from disk. Yes, it sustained 1.62 GB/s. Yes. We did this with 24 disks.

Again, on the excellent InsideHPC.com blog, John West points to a blog at Intel with an interesting observation: I would add to this that we have research computer users who prefer the expressiveness of languages such as Matlab, often ignoring the huge performance penalty for using such languages. The value to them is the ease of writing/maintaining their “code”. Tools such as ISC’s StarP attempt to build compiled code from the Matlab code.

Again, John West reads more than I, and notes at InsideHPC.com blog, an article from Doug Eadline on Linux Magazine, all about really big clusters. These are subjects I have explored a number of times. Doug points to nature and how nature scales and isolates failure. This also reminds me of the multiple types of networks that can be formed for computation/processing. One that I deal with every now and then are the spammers, and their bot-nets.

John West at the excellent InsideHPC.com blog points out in a short note that A number of obvious points about this … the economy has been under pressure. Customers have been buying less. SGI reports that revenue has increased to $93.9M from $79.1M in the third quarter. How much of this is normal seasonal variance (government purchase cycles) versus an actual increase in bookings (e.g. new deals that haven’t been worked on for a while).

I don’t mind tearing into bad implementations of an idea or product. If the thing isn’t good, criticism can help focus where it needs to improve. The trouble with this is when the criticism arrives too late, or is rejected out of hand by those who would benefit most. No, not being self righteous. I am just as critical of our stuff, our products and mis-steps as I am of others.

[optical communications/.](http://slashdot.org) reports ona wired story which covers the demise of fundamental physics research at Bell labs. For those who aren’t aware, your ability to read this on your electronic device is directly as a result of fundamental physics research at Bell Labs. The vast majority of computers these days are based upon transistors. Which was invented at Bell Labs. Yeah, you might say “so what”. Curiousity driven research can sometimes pay back in a big way.

So we had loaded two pretty darn nearly identical JackRabbitsландшафт for delivery to a customer tomorrow. As part of the load, we want serial consoles available in case we need emergency access. Plug it in and solve problems. Great. Remember, these are virtually identical machines. Same MB, same CPU, same rev (one has more cores/twice the memory of the other used for disk to disk backups). Same bios, same bios settings.

Running NCBI BLAST on the JackRabbit we are preparing for shipment. Used the nt database from last july (21 GB in size, 5+M sequences). Our a. thaliana had 1164 sequences, and about 500k letters. Took 8m 44s to BLAST these sequences against this database. This means about 2.1838e+13 cell updates per second. This is the product of the number of letters in the database and the sequence under test divided by the total wall clock time.