My laptop, a Fujitsu Lifebook P-2046, has a built-in wireless ethernet card. It’s only 802.11b, which is slower than most these days at 10Mbit/sec. It works fine, but I see E using her iBook at 54Mbit/sec and can’t believe how fast things appear on her screen. (Part of this is because her Web browser doesn’t go thru a proxy, while mine uses privoxy to weed out the crap, but it’s still noticably faster.)
I got a Netgear WG511 card, but it doesn’t work out-of-the-box with SuSE 9.2 and its 2.6.8 Linux kernel. I followed the steps with SuSE’s YaST2 program to configure the card so everything ended up being put into the file /etc/sysconfig/network/ifcfg-wlan-bus-pci-0000:01:00.0. (Even though it’s showing up as eth1, not a wlan* device.) Trying to activate the card failed. A little Google-digging, and I found the answer: use a driver from www.prism54.org. No luck, the server seemed to be down for days.
The Web server over at prism54.org finally came back up, so I was able to download the newer driver for the card to make it actually work. Part of making the prism54 driver work includes setting up a firmware update, which was made available for easy download to be installed in /usr/lib/hotplug/firmware as the file isl3890. I did also have to make one tweak so the ‘make modules‘ command would work.
The combination worked like a charm. Now it’s zooming away happy as can be. 🙂
One catch: like most other PCMCIA cards for this laptop, they suck the battery dry like a purist enjoying his pint during a heated conversation in a pub. On the road, the internal one has far less impact on the battery, but within proximity to a power source the WG511 card helps convince you there’s no need to upgrade the laptop just yet.
Besides, the next generation of the Lifebook is using Intel’s chip instead of Transmeta’s, which is a real shame. For the religious, the next step is the Actius MP30 from Sharp. With their ongoing work with Sharp, I wonder if Transmeta’s going to do a chip similar to the ARM so the successor to my Sharp C860 described recently could grow to use a Transmeta chip instead? The basics of its Code Morphing software seem to me to imply the ability to emulate any architecture, not just the x86. So why not shoot for a chip that’s buried inside an awful lot of stuff these days?