Switching from Mac to Windows (addendum)
Addendum – Performance comparison
Ok, I've got my PowerBook back now, so thought I would do some quick performance comparisons. I didn't put a lot of effort into this, so the results shouldn't be taken as being very scientific.
First off, I had to copy a bunch of files I'd been working on back to the Mac from my PC, so I tried networking them together by Ethernet. My work folder contained approximately 120MB in 3000 files. I managed to set up file sharing in both directions quite easily (though I wasn't able to share the whole of C:\, despite confirming I knew the security risks involved). Copying the files from PC to Mac using Windows Explorer took 125 seconds, with time estimates fluctuating wildly (2min 55s, 50s, 2min, 90s, 60s) as the copy progressed. When I looked at the result on the Mac I noticed that the timestamp of files was preserved but directories lost their timestamp, and files were all copied with mode rwxr-xrw-x. For comparison, I performed the same copy from PC to Mac but this time driving the process from the Mac. This time the copy took 33 seconds, with the Mac Finder also providing better feedback on the copy (not just a progress bar and time estimate but also telling you many files and megabytes were being copied). Directories had the correct timestamp, but files had mode rwx------ (I would have preferred making them world readable by default).
I then wondered how much of this time was due to networking and how much due to writing the files. I therefore duplicated the test directory on each of the machines. Under Windows this took 73 seconds, on the Mac 33 seconds, so it seems like the Mac is significantly better at reading and writing files.
I'm not terribly well organised and something I do often is use the operating system to search for a file by name. Although difficult to compare like for like, so any timings would be meaningless, searching was much faster using the Mac's Finder than in Windows Explorer, a matter of seconds rather than minutes.
Next I tried another test I knew I would be able to do easily on both platforms, import a CD into iTunes. I used the same version and settings on both machines (iTunes 4.6, 128kbps AAC) and a CD lasting 50min with 13 tracks. My HP imported it in 4m45s, with a peak rate of 13x shown. On the PowerBook this was slower: 5m40s with 11x shown. I don't know if ripping CDs is more a test of processor power or of drive speed, if the latter then the fact that the PowerBook has a superdrive while the HP has a combo-drive might help explain the slightly disappointing difference.
Next I tried opening a Word document. On my PC it took 6s for the Word window to appear the first time, then a further 4s for the document to appear, giving a total of 10s. After quitting Word and repeating the test the whole thing took under 3s. Again, the Mac was slightly slower, taking 10s for Word to open, and another 5s for the document to appear. Subsequent launches were also very quick, however, under 3s. One possible explanation for the difference is that some of the required components may already be loaded under Windows (e.g. if Internet Explorer is open) but either way, the performance on both platforms seems very acceptable to me.
Finally, I ran one of my own applications, CatDV Pro, which is written in Java. Importing and analyzing the same directories in each case, containing a mix of media file types, showed that the two machines were very similar, though with some fluctuation depending on the precise test.
These timings are summarised in a table, with the best results in each for each test marked in bold:
|Rip audio CD:||285s||340s|
|Open Word document:||10s / 3s||15s / 3s|
|CatDV launch (first time/subsequently):||24s / 4s||11s / 4s|
|CatDV import/analysis (various folders):||19s / 22s / 12s||13s / 24s / 18s|
On balance then, across a range of typical operations (none of which were unduly processor intensive), it seems as though the two machines were quite well matched in terms of raw performance. Given that the HP was bought as a budget second machine its overall performance was therefore very respectable, even if it was inferior in other respects.
Copyright (C) Rolf Howarth 2004. All rights reserved.