Graham,
Thanks, adding to this, I have just viewed test results for various card/card reader combinations, the speed variation is surprising. My kingston cards and reader combination are among the poorer performers but it's not enough to explain the whole story. My ingest is just a simple download and copy, no renaming no IPTC info, 99% of the time a single folder is created, all the cards from one job downloading to this folder. Perhaps you can send me a shots of your recomended settings, though I suspect I'm doing things correctly. Yo can use my email for this.
Here are the results of our tests with a Sandisk Extreme Ducati Edition UDMA CF card and a Lexar FireWire UDMA CF card reader. The top throughput as measured with a Finder copy is 38 MB/sec. Total data size on card was 2.64 GB. Times are in seconds:
Mac OS X 10.5.4 Primary Primary+Secondary
--------------------------------------------------------
PM 4.5.3.2 94 182
PM 4.5.3.1 142 182
Bridge CS3 94 174
Finder 67 134
The Finder is always going to be faster than Photo Mechanic no matter what we try to optimize. The Finder does not parse Exif data, Maker notes, etc. I don't know what kind of parsing Bridge does while it reads the files but the times are similar to PM 4.5.3.2. Notice that PM 4.5.3.1 is very much slower than PM 4.5.3.2 on Mac OS X 10.5.4. We don't know what Apple changed about disk I/O in Mac OS X 10.5.3+ but it affected Photo Mechanic greatly. We have spent several days finding faster I/O routines in the system and now our ingest speeds rival the speeds of Photo Mechanic 4.5.3.1 on trusty Mac OS X 10.4.11.
I would like to see your times. Please note that you must unmount the card and remove it from the reader between each test in order for the test to be fair. Otherwise the OS will have cached many of the files on the card and subsequent tests will go much faster which will skew the results. Also, don't have PM open up the contact sheets during the Ingest. It is not fair to make PM also render contact sheets in realtime along with the copying process.
One other thing that will affect times is that if you have PM set to unmount the card for you when the Ingest completes, then that will also skew the results. Bridge CS3 and the Finder do not offer such a feature, and it can take several seconds in practice to unmount the card. Photo Mechanic's Ingest will not signal ingest completion until the card completes the unmount procedure. So if you really want to compare apples to apples, you need to turn that feature off. We left it on in our tests and despite the additional time spent unmounting the card our speeds compare well against Bridge CS3 on Mac OS X 10.5.4.
Unless you can show otherwise with your own tests, using a stopwatch and unmounting the card between tests, we are considering this a non-issue and will not hold back the release of Photo Mechanic 4.5.3.2.
-Kirk