$ kill -15 sleep
Installation of development tools: unfortunately (though not in the least bit surprisingly) the ROM shipped with the computer does not come with a distribution of XCode. But, that means I got to test out the WiFi based remote disc installation ability. Turns out, it’s quite simple: you just put the disc into the host computer, and enable CD and DVD Sharing. The Air can then find the disc on the network, mount it, and about 40 minutes later XCode is ready to go.
Apache tests: Compiling for 64-bit Snow Leopard; Apache 2.2.17; basic configuration layouts for OS X 10.6 (merely to overwrite the Darwin supplied Apache installation). Tests run with power not battery, of course.
configure : 1m05s
make : 3m01s
CPU utilisation: both cores at about 65–75% the entire time. I was also interested in heat during this test. Up until this point, under normal usage, the machine has run as cold and silent as a brick. Doesn’t even get slightly warm. After 4m of reasonably saturated CPU usage, there was only mild amount of heat if you hold you hand against the bottom of it. This thing is amazingly efficient with power. Fans never spun up—I’m not even sure if it has fans (okay, installed iStat Pro and yes there is one exhaust fan that hovers around 1,400 to 2,000 RPM. I have to hold the thing up to my head to even hear it though). CPU temperature stable at around 37º; enclosure units and heat sinks at 27º. So yeah… like I said, it runs cold. For comparison my MBP runs at about 65º CPU and 70º for the GPU when basically idling.
So, 3m to compile is not shabby. To compare, here are the stats for the MacBook Pro.
configure : 1m18s
make : 4m50s
Given that the MBP is way more machine than the 11”, I think my earlier hypothesis on disk I/O bottlenecking hold true. It would be interesting to see what a more involved compile like Firefox would result in.
Machine comparisons:
MBA11 : 4GB DDR3 RAM; 1.6Ghz C2D w/ 3MB of L2; Sys bus 800Mhz
MBP : 4GB DDR3 RAM; 2.8Ghz C2D w/ 6MB of L2; Sys bus 1.07Ghz
External monitor tests:
These are more difficult to do, since I don’t want to actually test for hours on end, and the estimation system must be taken with a grain of salt. But, preliminary results look quite good. To reiterate the above post, with all antennas off and screen brightness at 1, battery time fluctuates at 8:30 to 9:30 for a full charge. With the antenna on, that drops to 6:00 to 7:00. With an external 27” monitor plugged in, estimation dropped to 5:30 for a bit, and then went back up to 7:50. With the WiFi antenna on, it stayed at around 6:00. Now for the external tests, I operated the machine in clamshell mode to conserve more power, I didn’t test with both monitors on, but I suspect at 1 brightness it wouldn’t make a huge impact.
So unless you have all-day meetings, you might not need a power supply. The big question I have is whether or not this will apply to a non-Apple DVI monitor. Unfortunately my old Dell 24” crapped out a few months ago and I sent it off to get recycled so I can’t provide a control.
All in all, very impressive results for something that is in all but name, a netbook.
I’ll have to run tests with Parallels later. I’m guessing stripped down w7 would do fine though, especially with the extra RAM.