I should note that there are two changes I had to make to my MSI wind to tailor it to my liking.

The first is that I wanted more memory. One gigabyte just doesn’t cut it for me these days, and two gigabytes is the low end of useful, normally. Since I’m intending to use this as a portable computer, and not an everyday computer, I don’t think I’ll be placing as much strain on it as I would on my desktop. Since the MSI Wind supports 2 gigabytes, I opted to do the upgrade. It was a relatively simple, but warranty-voiding, matter of removing several screws on the bottom of the case, and putting in a chip. If you’re not comfortable voiding your warranty, you shouldn’t do this. If you’re mildly technically competent, and willing to take the risk, it’s easy to do. Fortunately for me, I had a PC2-6400 1GB SODIMM lying around. This doesn’t exactly match the memory in the MSI Wind, but it does work, and hasn’t been problematic. The default memory is slower than that, and thought he laptop can use PC2-6400, I doubt it’s taking advantage of the faster speed.

Note that the onboard memory on the Wind is actually onboard. You don’t need to remove any memory to add new memory, you just add it. The word out there is that the Wind will only support a maximum of 2GB, so buying anything larger than a 1GB SODIMM is wasted money, I suspect. However, if anyone out there has put in a larger SODIMM and gotten more memory, I’d love to hear about it.

The other configuration change I had to make was to the WiFi network card. The default WiFi card that comes with the Wind is not supported under OSX by default. At the time that I set up my Wind, the acceptable practice was to remove the standard card, and install a 3rd party card with a chipset that is supported. The Dell DW-1390 and DW-1490 both have supported broadcom chips. At the time I was purchasing, the only difference I could find between those two cards was about $15. I got a DW-1390 off of ebay for $12. Installing the card was easy, though there was a small modification to the motherboard that needs to be made: There is a small mounting nut that is soldered onto the motherboard, which doesn’t interfere with the default card, but doesn’t quite fit with the DW-1390. It was very easy to heat the nut with a soldering iron, and pull it off. Other sites suggest that you can just wiggle it off, but I’d recommend against that, especially since you’ve just voided the warranty. Applying a soldering iron to the nut and heating it just enough to loosen the solder made pulling it off the motherboard with a small set of needlenose pliers a trivial move. If you need to go this route, do it carefully and properly.

A bit after I replaced my network card, Realtek provided a drop-in driver for the WiFi card that is in the Wind. By all accounts that I’ve read, this approach “just works” and is an easy install. I believe the Realtek card is 802.11 B/G/N, whereas the Dell cards are just B & G, so there is a bit of an advantage to just adding the driver. I occasionally consider putting the default card back in, but I just haven’t gotten around to it.

Adding the memory and swapping the WiFi card was very quick and easy. I believe that doing both took me less than 10 minutes, and some amount of that 10 minutes involved me getting my soldering iron and waiting for it to warm up. The insides of the MSI wind are very easy to get to, and these specific parts are straightforward and accessible. Some of the other modifications out there, such as modifying the mouse buttons, or replacing the touchpad, will take more time and effort. I may do them at some point, but for now, I just wanted a basic laptop.

With the firmware installed, I’d been gearing up for how to actually install the OS.

First off, I should mention that, for odd reasons, I had a legitimate copy of Leopard that wasn’t going to use. If you’re doing this install, I recommend that you also obtain a legitimate copy of OSX Leopard. Apple sells it for $129, or you can get it here for $109, with free shipping. I’m not sure about the “New and Used” versions for $80, but you may decide to take the risk.

Having acquired a legitimate version of OSX, you’re now going to put it on the shelf. Search your favorite bittorrent site for a distribution called MSIOSX86.iso. This is a version of OSX that has already been modified to include the drivers necessary to make the MSI Wind run nicely on this hardware. The version I installed was based on OS X 10.5.2, and it may have been upgraded by now. There are ways to get your release upgraded to a current version. Right now, it’s all about having a starting point, and I’ll try to cover the upgrade at a later point.

Many people took to burning the OS image onto DVD, and booting from an external DVD drive. Not having one, and not wanting to spend $60 for one, I looked for other options. The best option I found was very similar to what I found for the firmware: Burn the ISO to a USB flashdrive, and boot from there. I found a great set of instructions on how to do that on Dan Cameron’s site.

The short form of what you’re doing to get OSX on to the drive in a bootable manner is to mount the ISO image as a file system. You then use Disk Utility to “restore” that image onto the USB drive. That will result in a USB drive that is (or should be) bootable on Apple Hardware. However, since you’re not using Apple hardware, you need to do a little extra work, which is not mentioned on Dan’s site. To make the drive bootable, you need to use Chameleon, which is one of the tools you can find at the OSx86tools site. Chameleon does all the appropriate magic for non-apple hardware to boot off the USB drive.

Now it’s just a small matter of booting off the USB drive and installing OSX. First, with the machine powered off, install the bootable USB drive. When the Wind powers up, go into the BIOS settings and tell the machine to boot off the USB drive as it’s first option, and save the bios settings. When the machine restarts, it should see the USB drive, boot off of it, and bring you up ready to install. At this point, it’s a normal OSX Leopard installation, and you proceed as you normally would.

I should note that my goal was to have a netbook running OSX. There are people out there who have set up their Wind to triple boot OSX, Ubuntu, and Windows XP. I considered this for a while, and then realized that OSX would really be the only OS to get used. I rarely need a windows box, and I have plenty of other Ubuntu resources generally available to me. It’s nice to have that option available though. If you’re interested in that, a simple websearch should lead you to someplace useful.

Assuming you’ve gotten your MSI Wind, the first thing that you’re going to want to do is to upgrade the firmware. My MSI Wind came with version 1.06 of the firmware, which is a bit older than the current. The current firmware is 1.10 (or 1.0B) and can be found here.

I did things a bit backwards, and just dove in on installing OSX. When I decided that I should upgrade the firmware, I had a bit of a problem. The firmware upgrade procedure involves running a DOS utility that will re-flash the firmware, and in installing OSX on my Wind, I had erased the only real DOS machine I had available. Eventually, I was able to burn a bootable DOS image onto a 128Mb USB thumbdrive that I had around, put the firmware on there, boot off it, and install the firmware. I’m not a windows user at all, but I suspect that there is a way that you can boot into a basic windows command line interface, and do the upgrade from there. I leave it to the reader to figure out the best method for them, but I am now very happy to have a bootable USB drive already setup for the next time I have to do this.

At the time I did my install, I went from firmware 1.06 to 1.09, since that was the most recent at the time.. Some of the changes in the firmware have been noted to give the MSI Wind a ~30% speed boost. Another notable change is that you gain the ability to overclock the Wind anywhere from 8% to 24%, simply by pressing Function + F10.
“Simply” is slightly misleading here. You do have to do a little setup in the BIOS, and pre-determine how fast you want to overclock. You have the option of 8%, 15%, or 24%.
Once you select it in the BIOS, you boot into your OS, and then when you press Function + F10, you get overclocked to the rate you’ve set in the BIOS. This is handy for seeing if your machine can handle the overclock. If it can’t, you can back out of it very easily.

I’m not running my Wind overclocked. I couldn’t tell any notable difference for my computer, and generally I want more memory instead of more CPU. Other people seem happy with the ability though.

For a while, I’ve found netbooks to be interesting items. I’d watched for a while, but they only became interesting when people started making OSX run on them.
Back in December, I purchased an MSI Wind off of Ebay, with the intent of making it a very portable and very inexpensive OSX box. I’d had an extra copy of OSX 10.5 (Tiger) and this was a fine way to put it to good use.

On the whole, the netbook behaves as I’d like, with a few little quirks. There is a webcam, but it’s activated through a hotkey sequence, not directly via software. Bluetooth is much the same. Audio does work, but switching from built-in speakers to the headphone jack (and back) requires a small shell script to be run. The touchpad is overly sensitive, and sometimes problematic. None of these are show stoppers for me, and the machine has many fine features: It’s small and light, yet still usable. It’s very easy to sleep, and it wakes from sleep very very fast. Performance-wise, it seems very similar to my old 1.67Ghz G4 powerbook. Whereas I had been pushing the limits of my powerbook, the MSI Wind doesn’t get as stressed, since I tend to use it in a much more limited way than I’d use the powerbook. One of the best features of the MSI Wind is that it’s cheap, and very hackable. I was surprised at how easy it was to get OSX up and running. Over the next several posts, I’ll detail the bits of knowledge I’ve gleaned from my experience.

I’d been running a beta-release copy of it for a little while.     On the whole, it looks like a nice upgrade for the Wordpress platform.    Specifically of interest is the ability to upgrade your software (including your plugins) “automatically.”    It’s not quite automatic, since you still have to enter in some info, but it does work nicely.   The one drawback is that if you have 5 plugins to upgrade, you have to step through the process individually, and update each one in turn.

Aside from the auto-update feature, most of the changes I’ve seen thusfar have been cosmetic.   The layout is cleaner.    The editor looks like it’s been fixed up a little.  Many things are a lot more customizeable.    It’s hard to point at brand new features, but lots of old features have been polished, with perhaps some added functionality hidden in the fold.

I’ve been really pleased with Wordpress as a blogging platform.   It was very easy to set up (though I’m a technical sort) and very very configurable.   There is a good user base out there, and a lot of people contributing plugins, themes, and features.    In the past, I’ve looked at a few other platforms, notably Movable Type, and didn’t feel the pain was worth return that I got.    Drupal and Plone (which are content management systems, not blogs, but you can do many similar things with them) are also very nice, though often targeted at much larger sites than I’m trying to do here.   Wordpress seems to have the features and flexibilty that I want, and with Wordpress 2.7, it looks as though the amount of time I have to spend administrating the system has dropped a lot.   If you’re looking for a blogging platform, I suspect you don’t need to look any further than Wordpress.

I was recently asked about building a power efficient server, and it seemed like a good idea to share my advice with a wider audience.

Anyone who has to deal with a hosting facility very likely knows the crunch that they’re running into with power. A standard rack may only have 42 rack units in it, but you’ll be lucky to be able to fit and power 15 servers in that rack, on a standard 30 amp circuit. If you’re careful in planning, I can help you get squeeze a bit more into that rack. I won’t advocate any particular server vendor, but I do believe that you’ll have more flexibility with a custom white-box builder, instead of going with one of the (often pricier) big name vendors.

Overall, you want to lower your power draw per server. With some careful planning, you can lower your power consumption by a fair bit, and still get the same power per machine. Here are a few things to consider:

Just to give a real-world example of how I use my kill-a-watt P4400:

The 10+ year old 3com 10/100 network switch I used at home was getting noisy, and I realized that it was probably not a bad time to upgrade it.   Power wasn’t specifically on my mind when I went to do the upgrade, though it did occur to me mid-process when I noted that many of the home network switches were very low on power draw. I’ve ended up replacing  10ish year old 3com 24 port 10/100 switch that was built for enterprise use with a 5 port 10/100/1000 switch that was built for home use.   How this has played out power-wise with measurements from my kill-a-watt:

Now, 41 watts may not seem like much of a difference, but it can add up to real money.   When all is said and done, I pay about 18 cents a kilowatt-hour, including, generation, transmission, and all the assorted fees that go with it.

The basic math breaks down like this:
45 watts (3com) - 4 watts (Trendnet) = a power savings of 41 watts.

41 (watts) * 24 (hours in a day) = 984 watt-hours

984 (watt-hours) * 365 (days in a year) = 359160 watt-hours per year or 359 kilowatt-hours per year.

359 kilowatt-hours * .18 cents per kilowatt-hour = $64.62/year

or, to refer to the above chart: $70.92 - $6.30 = $64.62/year.

So, by investing $19.99 in a newer, better, quieter, switch, I’ll be saving $64.62 a year.   I find myself doing these sorts of calculations on many things around the house.    I’m a big fan of old and reliable, but there is a place for new and cost-effective too.     I’m in middle of a slow and gradual process of replacing all of my old CRT monitors (which can often draw 100-150 watts) with newer, smaller LCD monitors ( the 19″ LCD on my desk draws 30 watts).   If you run the calculations above, you can see that it can add up to real money fairly quickly.  To be fair, monitors are a little trickier, since they may not be on all the time.   My network switch is though, so $64.62 is real savings to me, and the switch itself will be paid off fairly quickly.

I mentioned earlier that I use the Getting Things Done (GTD) methodology for organizing parts of my life. A lot of this is managed on my iMac. I have been using iGTD for a lot of it, but I’ve recently started looking at other GTD software. One tool in particular, is Things. It looks interesting, and a lot more straightforward than iGTD. I haven’t converted to it completely yet, but if you’re considering trying out GTD on the mac, you might want to give Things a look. I’ll report back more when I have a bit more substance (and perhaps screenshots) to add.

Another interesting tool that I looked at was Goal Enforcer , which approaches GTD mind-map style. When all was said and done, that approach didn’t work for me, though I can easily see it working for many people. The walkthrough and tutorial was a great introduction to GTD though, and presented it in a fairly straightforward way. The product is well worth considering, but the tutorial is definitely worth it.

There are a few other packages that I’ve looked at, and I’ll through some more information up here. GTD is big in the geek community, so there are lots of people taking a geek approach to implementing it. I find that for the moment, the tools that I like the best are the ones that I install on my mac, and can sync to my smartphone. There are a large number of web-based tools that I’ve peeked at, but none of them have grabbed me yet, so I haven’t gotten excited about them. If I could have a desktop tool that could publish to the web and sync to my smartphone, I’d get excited.

I’d be delighted to take a look at other GTD packages out there if people have suggestions.

In retrospect, it seems important to note that on the airline trip back, I was wearing airline-issued Bose Quietcomfort 1s, the original model.   The QC1s were the first generation, originally released back in 2000 or so.   On one hand, they’re a bit bulkier, with the electronics contained/powered by an external box that is inline with the wiring to the audio jack.   On the other hand, the headset itself folds up neatly, and the headphones can be somewhat portable.   I used to use a pair of Sony MDR-NC5 foldable noise canceling headphones.   The Sony’s definitely folded better, but the Bose sounded better, and were much better at noise reduction.

With the QC2, the noise canceling is noticeably improved, but the headphones themselves are a bit bullkier to go with it.  Fortunately, the headphones are much more comfortable, and they do not have an external box like the QC1.   The headset sits nicely on the head, with very comfortable earpad, which surround the ear and help to isolate the ear from external noise.

There is also a Quietcomfort 3, which apparently has noise canceling hardware similar to the QC2, but the headset is lighter, and is designed for a more “on the ear” setup.   This is nice, but I didn’t feel like it got quite the same feel as the Quietcomfort 2s.   For short flights, I may not care.  For longer flights, I’d prefer the QC2s to keep my ears from being squashed.   The QC3s are also about $50 more expensive than the QC2.

My take on it is that if space is a problem, I’ll probably use QC1s.  If I’ve got that can be more flexible to deal with it, I’m happier with the QC2s.