2008

Thu, Jan 31

<20:41 EEST> Wrestling with the 2.6.24 NFS issue, I've even moved to NFSv4 using a couple of guides. This was a sensible thing to do anyway, for increased security and other features. Alas, the network mounts don't perform any better, so a more general problem outside the NFS stack seems more likely.

Strangely enough, no similar reports are showing up via Google. Perhaps this is an isolated case with my particular hardware and config? Not counting network hardware though, as I've tried an alternative NIC on the server, and a Firewire link as well :-/

<21:09 EEST> At least two other people have noticed general I/O slowdown :-j

Wed, Jan 30

<23:14 EEST> Since upgrading to Linux 2.6.24, I've experienced an annoying slowdown of NFS. Specifically, my /usr/portage resides on my server, and it's mounted via NFS on my laptop. I often emerge --sync on the laptop, since it has more packages to deal with, so it's more convenient to proceed that way.

With 2.6.24 on both machines, the Portage tree has become slower to access, though bulk file accesses via NFS (like videos) go just fine. So it's a latency issue, rather than throughput, as Portage deals with lots of small files. I've tried all possible combinations of client/server kernels between 2.6.23.14 and 2.6.24, and the only consistently good performance comes with the old kernel on both machines. This is somewhat daunting, as I can't pinpoint the problem to either server or client issues. I'm also suspecting a general scheduling problem with the new kernel.

Mon, Jan 28

<14:01 EEST> I guess we all experience the problems with human multitasking, so this article won't provide much new insight, but I guess it's important that the issue is actively discussed. People are making themselves miserable with their lifestyles, and they need to be reminded there are alternatives.

Incidentally, I have a minor software issue where multitasking would actually help. On my new machine, MPlayer cannot play a 1280x720 H.264 video with AC3 sound smoothly. But video or audio alone play just fine -- in fact, adding up to less than 95% of CPU load. It's even possible to run these with separate MPlayer processes to get an apparently smooth performance, albeit with the obvious loss of sync.

This is pretty much the reason why there's MPlayerXP, a multithreaded version of MPlayer. Unfortunately the project is not very active and the releases not very stable.

Sun, Jan 20

<18:09 EEST> The new Hoo is up and running! I received my order on Friday, including a nice and sleek Firewire/USB SATA hard drive enclosure. That night I only had time for testing the latter, as I was off to Huoneteatteri to see Some Explicit Polaroids. Firewire was no trouble except for the new, experimental drivers that turned out faster but slightly buggy. Of course the old drivers are still available.

Switching to the new motherboard on Saturday was a breeze. It was almost boring, an incremental upgrade that booted up fine with the existing installation. Naturally I'm satisfied with some of the incremental stuff as well, which makes video playback and other things smoother. The only major glitch for now is with HDMI output, which is stuck at 1632x1026. I was particularly interested in driving my monitor via DVI, like the deities intended, but this resolution glitch kind of moots the point. Fortunately, VGA is slighly better than before, and Xvideo is no longer pixelated.

Fri, Jan 18

<00:26 EEST> My "prime obsession" that started last summer with Project Euler and some pop-sci books, has gained a new foothold with a similarly titled volume by John Derbyshire. It's a shaggy dog story around Riemann hypothesis, traversing an intriguing path (or perhaps a contour) in the history of math and science.

This could even be one of the problems where mathematics turns out an experimental science. Vaguely related to this idea, I've joined yet another distributed computing scheme: the search for Mersenne primes.

It also happens that I've completed one third of this spring's training lessons. My schedule for the rest of the spring looks pretty bearable in this respect.

Thu, Jan 10

<23:52 EEST> Watched the movie Valo on Wednesday at a teacher training seminar. For a children's movie and a school assignment, it turned out wonderfully captivating. Probably in part due to the real-life basis of Aleksanteri Ahola-Valo.

Wed, Jan 9

<00:04 EEST> Dr. Mr. Kalle Haatanen, I salute thee! Pitkäveteisyyden filosofiaa (~ philosophy of boredom) is my latest favourite book. Incidentally it's not so much about boredom itself, but of our present time of work and leisure. The person of this time in Finland is called 'tavis', somebody who takes pride in being a somewhat boring everyman. The book kept me nodding in agreement, not only with the themes but with the quirky style as well. I'm sure Haatanen would be very much at home with the SBF of the RDKH. This all makes me wonder if I might be a tavis as well, but given the self-proclamation in the definition, I'd quote Herman Melville's Bartleby in saying I prefer not to.

This book recommendation comes squeezing in between the new year's load of study and teaching. I started my share of the latter on Monday, experimenting with an open-ended practical project in electromagnetism.

---

Risto A. Paju