The difference in CPU use will be a combination of how the drive is being accessed and the speed of the CPU.
How it is being accessed is unlikely to be a big differentiator in this case - if it was the difference between (U)DMA and PIO you would see much lower rates on the PIO example.
I suspect that the CPU in your laptop is simply running slower, either because it is a slower CPU or because it has been throttled down to conserve power as it doesn't need to be running at full speed during the test. It might be worth finding a clock-frequency meter to run on the laptop to check if this it the case. Must desktop Linux variants come with one in the standard repositories (Ubuntu 9.04 UNR installs one in the default install), and you will not doubt be able to find free examples for Windows too. You may also be able to turn off CPU scaling in your BIOS settings temporarily and rerun the test - I don;t expect it will make a difference to the speed but if the CPU was clocked down in the last test you will see a drop in %used during the test as you'll be using, say, "12.5% of a CPU at 2Ghz" instead of "50% if one running at 0.5Ghz".