Computer Ressurection and Elastic Cloud Experimentation

I was home on Thanksgiving Break with Sharvil, and we decided to revive some old computers. Partly I’d like to experiment with some clustering stuff without incurring CPU time at the AMATH department or Teragrid stuff I’m likely gonna be working on soon with Shea-Brown’s neuroscience research. So, it turns out I resurrected about 5-6 old computers(final tally is still waiting on the number of successful Xubuntu installs on them, among other practical issues(where the hell am I going to put six computers…?): The very first computer I built(a P3 450), P3 700, Dual P2 266, a couple of AMD64 3200’s, and a Sony Vaio P3 733. The cool thing is that the neuron spiking models are basically embarassingly parallel(well, each run isn’t necesarily, but from what I’ve gathered so far, we’re looking for averages over a bunch of them. So, sweet! Again, this would be terrible for actual research, especially against something like TG or even Amazon’s EC2–which is another thing I really need to check out.

Sharvil and I also managed to restore functionality in a set of other computers, but the details of that have to remain somewhat quiet, apparently. CIA-types like their privacy, I suppose. Probably I’ll drop in a plug about it on some future project. On the way home from that I picked up a little light bulb for the dome lamp in my old F250 Diesel so that Dad won’t have to search though his cab in the dark anymore.

Above I mentioned Xubuntu… but currently I’m installing regular Ubuntu. It’s taking FOREVER. Which is why I think it’ll end up being Xubuntu. Either way, I guess I shouldn’t have started with the second-slowest computer on some of the slowest hard drives I’ve got. I’m probably going to wait for it to finish, then install Xubuntu on another drive or try imaging this one or something to see how long it takes.

Incidentally, I decided it’d be a great time to run some Amazon EC2 tests(since I happened across some tutorials on YouTube. It’s amazing… You load up the Firefox interface, find the Amazon machine image you want, click ‘boot’, and in short order you can ssh over to it directly. I decided I’d run a little python Sieving program to find prime numbers. I haven’t got the final numbers yet, because I want to get a pretty graph…. but it’s pretty sweet so far. The gist is: A P3 450 machine takes 8 times longer to run the program than my MacBook Pro, a small-image Amazon EC2 instance runs in just a little bit less time, and a extra-large EC2 High-CPU node runs it in just about 60% of the time it takes my MacBook Pro. Granted–this is a terrible benchmark, and it’s straight Python, and I’m comparing across OSX vs Linux, i586 vs Core2Duo vs ??, etc, but these are some good rough numbers for how long this stuff is going to be taking. I’m a little impressed my MacBook Pro was keeping up that well… but it has been a great little machine. And it explains why it runs so hot, probably. In any case, I’ve certainly had less fun with $.92!

UPDATE: Here’s that follow-up with the graph. It’s pretty revealing:

Incidentally, after a bit more sleep I realized that obviously this has nothing to do with float point ops, so this isn’t entirely indicative of how these machines would fare on scientific computations. But probably still interesting. Honestly I’m not quite sure why the AMD64 did so well. The machine feels way slower than the MBPo, but they are fairly close in actuality.