Question

Does anyone out there have hard data from profiling the iPhone/iPod filesystem? I’m more interested in reading than writing.

I was thinking of running a few tests, but thought I’d check to see if someone smarter and/or more time-rich had already done so.

Specifically I’m interested in speed difference between many small files versus fewer big files, and the differences between the various generations of iP* devices.

Another question is whether or not ZIP compression of read files is worth the tradeoff of decompression, or at what data size there is a “break-even” point.

Answer 1

From NSBlog, written my Mike Ash:

Name                        Iterations      Total time (sec)       Time per (ns)
Write 16-byte file (atomic)      10000               51.5             5153943.7
Write 16-byte file               10000               80.9             8089726.2
Read 16MB file                     100              137.6          1376092573.3
Write 16MB file (atomic)            30              143.8          4793527088.9
Write 16MB file                     30              151.2          5038515361.1

Here’s the URL:
http://www.mikeash.com/?page=pyblog/performance-comparisons-of-common-operations-iphone-edition.html