Posts Tagged ‘work’

HPM interrupt handling

Wednesday, February 21st, 2007

Getting exact process information is tough, especially when the tool one uses is not behaving as one expects. To ease the matter somewhat, I drew a graph that show exactly how perfctr calls the interrupt handler and sets the offending process’ signal.


Perfctr HPM interrupt handling

The idea is to assemble correct information on the event counts per fixed intruction count interval of e.g., 100M instructions. However, due to the non-deterministic nature of a multitasking kernel, sometimes we see IPC values that are … unexpected. Hopefully this graph sets us on the right track once more. In normal circumstances, having a few counts off track, is not too bad, but for the work we’re currently doing, we require correct counts.

Spam@work

Thursday, February 8th, 2007

For the past months, all Ghent university employees and students have been spammed on a regular basis, to ask them to cast their vote in favour of a Ghent University building that is competing in a TV show, to get refurbished. The show will select one monument in Flanders for getting fixed back to its original state, as the government seems to have too little money to keep our monuments intact.

I can see that the upper echelons would want to let us know that we can actually vote, should we not have watched that TV show (which I didn’t). But to keep sending out these emails on a regular basis, to all employees and students, just seems quite unethical to me. By now, we know! And asking us to ask our friends and family to vote for this particular building seems to take it a step further.

And it’s not even a cool building. To put it simple: it’s butt ugly (YMMV). During WWII, the Germans had occupied the university library, and mounted anti-aircraft defenses on its top. Meanwhile, they apparently had decadent parties on the upper floors. If there was one building the English should have levelled with the ground, this was probably it. A pity for the books, but IMO, there are far better looking libraries, e.g., in Seattle.

So please, stop spamming us. I’m not casting any vote, and should I cast one, I’d certainly not vote for the university library. I like the King’s Stables in Ostend far better.

Alcohol at work

Tuesday, January 9th, 2007

I received a flyer today in which the university encourages its employees to think about alcohol usage at work. I know of no people in my immediate vicinity who have alcohol problems, but higher echelons seem to take this matter very serious, as they should. However, upon reading the flyer, I noticed a page on which the reasonable amount of alcohol a healthy person may drink, is discussed. Three rules are presented:

  • On average, drink no more than three glasses a day (two for women).
  • At most, drink five glasses a time (three for women).
  • Don’t drink any alcohol at least two days per week.

So, basically, don’t drink over the weekend, as you probably will want to remember the fun you had. And spread the 21 drinks you’re allowed to have accross the five other days. That’s four drinks per day, except on Friday, when you can start the weekend by having an extra drink.

It seems to me that the rules stated above are well beyond what’s reasonable. In my opinion, somebody who even drinks but one glass a day, can get addicted, though it may not seem too bad. At first.

Getting work done.

Friday, December 8th, 2006

There is an excellent post by Jason Fried on how to actually get some work done in an invasive work-environment. I was pointed to it here.

It seems funny that, while email is asynchronous, most people tend to answer their mails immediately upon receiving them. As Andy Smith indicated (point 3), IM can be used asynchronously too. But people seem to be even less inclined to use it that way. The bells ans whistles in the email, IM, and IRC applications are not very helpful. We should really try harder to use these tools in a way that benefits our workflow.

So, if I don’t answer your email/IM/IRC message as soon as you’d like, I’m most likely not dead, but simply trying to get a continuous workflow. Yes, ironically, I’m writing about this instead of continuing the work I was doing.

ACES

Tuesday, October 17th, 2006

On October 3rd and 4th, I attended the ACES symposium. It is an opportunity for the PhD students of the participating research groups to present their ongoing work or wild ideas they would like to try out by means of a poster and a small introductory presentation. Obviously, getting practice for giving presentations is always a Good Thing ™. Additionally, it is a chance to socialize during the recurring bowling evening, and to exchange ideas. All in all, a good time is ensured, though the food – while excellent – can use larger amounts :-) Pictures can be found on flickr.