Here's a build of Pyrit 0.2.5-dev-r193 for Windows, including the Nvidia CUDA add-on :)

Since WinUSB is not working all that well under Windows XP, I've re-added support for LibUSB. You can grab the package containing both LibUSB and WinUSB versions of the module plus the required drivers for various platforms here. Also, you can check out a quick preview of Mwifi, a wireless tool I've been working on for some time :)

The module is now using WinUSB instead of libusb-win32 so running it on a 64bit platform is now possible (XP/Vista/7). I've also synced the code with the one in the latest Linux kernel source (2.6.29-rc6 at this time). A build of Aircrack-ng suite r1431 is available here.

Here are some SunSpider JavaScript benchmark results for a bunch of web browsers.

Holy shit, yet another year has passed! Well, this time I bring you (who?) Superfetch Filter. If you want to hear the story behind it, read on.

So after a few failed attempts, I've finally decided to switch over to Vista. Got my stuff installed and working but after a while something didn't seem right. It looked like the OS was doing some strange disk IO when it really shouldn't. After taking a closer look with Mark's tools, Process Explorer and Process Monitor, it turned out that Superfetch was constantly fetching parts of some torrent file that I was seeding at the time. Why use precious memory for a damn torrent? I'd rather have more important stuff cached in memory than torrents and files that I don't care about.

Searching the web for Superfetch turned up no useful information on how to control its behaviour, except maybe for an interview on Channel9 with Michael Fortin, the Windows Performance Team leader at Microsoft. In this interview, he states that the folks at Microsoft did not want to offer any user control over Superfetch in its first incarnation and just let it auto-calibrate itself. That's nice and all but I want MY computer to do what I want, like ignore that damn torrent file and go fetch something else.

Well anyway... Poking around in the registry and in various other dark places turned out that Superfetch supports some third party control after all. Internally, Superfetch uses event tracing to gather information about accessed files, running applications, etc. Then it passes these events to its internal agents that use databases to store this information and decide what files to load into memory and when. But before passing the events to its internal agents, it passes them to a third party agent that can be specified in the registry. This third party agent can decide if the event is passed along to the Superfetch internal agents or if it is ignored. This is basically what Superfetch Filter does. It checks the events for certain filtered locations that you can specify and prevents them for being processed internally. Unfortunately, there is no publicly available information on how to build a third party Superfetch agent, so this is some kind of an ugly hack.

So until Microsoft decides to release some official information about controlling Superfetch, I'm stuck with this crap.

Update: I got used to the default behaviour so I'm not using this anymore.

Since I got a ColorVision Spyder 2 colorimeter to calibrate my 24" Dell LCD, I've calibrated a few other displays as well. So what you see here is all the data I've gathered so far. If you have one of these displays, you can use the brightness, contrast and RGB values from the table to obtain a white level of approximately 100cd/sqm, a minimum black level and a white point of 6500K. The software used to calibrate the displays and obtain the ΔE₉₄ values was basICColor Display 4.

ΔE is an international standard measurement of color difference, corresponding to the perceptual color sensitivity of the human visual system. Generally, a ΔE value of 1 corresponds to a color difference between two samples that would be noticed only by an expert under strictly controlled lighting, while a ΔE of 3 can just barely be seen by most viewers with good color vision in a brightly lit room.

I've decided to release my Descent port to the PSP even though its not 100% stable and may still contain zillions of bugs. In my opinion, the Descent source is a mess and not really PSP-port friendly. Check the psp-readme.txt file for install instructions and details. Also note that ad-hoc play will most likely not work in this release, since I haven't got a chance to test it yet, but infrastructure mode should work just fine. Please remember that you need updated PIG and HOG files.

Since I got my MX518 a few days ago I thought I should google around and see if I can find some tweaking goodness for it. To my surprise I stumbled upon this thread and this thread on the Razer Blueprints forums. You may already know about the first one, it's a patch that allows the USB polling rate to be changed from 125Hz to 250Hz, 500Hz or 1000Hz. I got mine set at 500Hz. The second thread surprised me the most (in a very good way). It's a patch that removes that very annoying acceleration that may still be active in some games, even if enhanced precision is disabled in the mouse control panel.

Today I managed to login to Google Talk thru Trillian. All you have to do is get this updated Jabber plugin and then follow these instructions from Google. I'll be at magikboy at gmail dot com so holla at me dogg. Peace!

And they're off! To a standing start...