Given that the particular hardware setup is detailed here (a GTX 480 achieves the 1 billion keys/sec figure), and the algorithm used (radix sort) has known asymptotic behavior (O(nk) for n keys of length k), 10^9 keys/sec is quite meaningful, particularly since it's a significant implementation challenge (possibly even an algorithmic challenge) to port this algorithm to a GPU.

Furthermore, I think sorting speed is appropriately measured in keys/sec. Big-O does not in fact describe the speed, but rather the upper bound of the growth of an algorithm's asymptotic running time, which needs to be paired with the implementation, architecture, and data set to determine a speed. It turns out the constant factors can actually be quite important in practice.

Channeling Doctor Cox from Scrubs: "People are bastards. Bastard-coated bastards with bastard filling."

Out of curiosity, did your dad try installing Ubuntu on that rig? If so, what problems did he encounter?

I ask because I put together a computer for my mother a few months back, with relatively similar specs, and was pretty impressed with both the Windows 7 and Ubuntu 10 install processes. They were both painless experiences; I can't remember either one being technical at all. I certainly didn't have to muck around with the CLI at all (well, that's not quite true - I did have to do some CLI magic to get a USB wireless stick to work, but in W7 it didn't work because there was simply no driver at all available, so I'd call that one a tie, or in Ubuntu's favor).

Point is, you've clearly had some bad experiences with Ubuntu, and you're entitled to your opinion, but I really think the whole user experience for Ubuntu is vastly, vastly better for the majority of users than your experience indicates.

When you're doing test-driven development, how do you avoid coding too narrowly to the tests? At some point you surely must think about general principles; how does that fit into the framework?

I'm OT, but I wish latex had been developed with this philosophy in mind. I really dread seeing latex spit out screenfuls of unintelligible gobbledygook, followed by a single "?". Oh great, what did I do wrong this time? Latex sure isn't telling me.

C++ is up there (down there?) as well.

This is an interesting comment, because what is "dad age," anyways? To children, parents seem incalculably old, wise, and experienced. Parents, on the other hand, remember well being angst-ridden teenagers themselves, and that it wasn't so long ago.

Then you don't make the change. But then the rationale is: "management didn't approve it," instead of "It was this way when I came on board." There are other valid reasons, too; maybe it would be too expensive, maybe the downtime (or risk of downtime) would be too great. But the reason given was "I inherited it," which isn't a good one, in my opinion.

I couldn't tell you. Even looking at a transcript (which I posted below), I didn't have the patience to cut through all the BS. He says "freedom" or "free" or some permutation thereof pretty much every 15 seconds, which I just couldn't take for more than a few minutes.

For people who hate watching video as much as I do, here's a transcript: http://www.softwarefreedom.org/events/2010/isoc-ny/FreedomInTheCloud-transcript.html

No kidding. The sysadmin who uttered that quote comes across poorly as well. He makes the excuse: "It was like that when I got here, so I inherited the bad design." Seriously?! Your job title is Network Administrator! Administer the damn network! It's what you were hired to do!

In my case, yes, the system was exceedingly well documented, and also made use of standard tools (Makefiles, perl and bash scripts, etc.).

But I don't think documentation is a panacea if the tool used is particularly rarified. Perhaps the DBA in question (this is purely hypothetical now) set something up using Oracle, and then left. Now, maybe it's easy enough to use as the interface for SQL queries and the like, but what happens if there are major reorganizations that really do require specialized knowledge? Can you document all possible contingencies? Without simply giving enough learning materials for the user to become a DBA? (I have no idea, honestly, since I don't have any experience with high-end DBs, I can't say anything about how hard it is to maintain one, so I'm more making a general point rather than a specific one.)

In 5th grade, I always felt sick during math class, which was taught by a teacher that I really didn't like.

In my defense, their understanding of mathematics would have made any thinking person ill :-D.

Er, doesn't that behavior make sense? I wouldn't let a medical doctor to write a compiler for me, and I wouldn't let an engineering Ph.D. remove my appendix. The fact that alternative medicine is bullshit doesn't seem like a reason to disparage M.D.s.

See, if your pictures of your tinfoil hat had geolocation data attached on them, you'd be able to find it!

Depending on the size and stability of the GPs research budget, that may not be practical. I worked on a fairly large academic research team (by EE/CS standards) that had the budget to hire a few full-time staff members for certain things. After the main implementation push the project wound down a bit, and those staff moved on to other jobs, leaving the grad students to maintain the infrastructure. That was fine as it was, but could have been massively not-fine if the staff had used complex tools that required specialized knowledge that the students didn't have, and would have to divert their energies from research to tool-learning.

Basically, if you're hiring a DBA, make sure that you can keep them on staff indefinitely.