Question on DEVONthink memory usage

I have a quick question for the DTP engineers:

I went to this webpage:

soschildrensvillages.org.uk/ … chools.htm

And downloaded their custom Wikipedia extraction. On disk, it takes up about 3.7G.

I think indexed this into a DTP database. It took about 5.5G of RAM to do so (swap was the highest I’ve ever seen). After quitting and restarting, DTP is now taking 808MB of resident memory, 792M RPRVT and 1.18G VSIZE.

My question is, what is it keeping in memory that needs to be so large? Here are the database statistics:

Groups: 978
Images: 84,255
HTML Pages: 43,224
Links: 6
Plain Texts: 27
QuickTime: 9
Words: 522,400 unique, 29,331,450 total

That really doesn’t seem like it should demand so much memory. I have some databases with over 62 million words, and they don’t take up as much resident memory on start as this!

Thanks,
John

Quick answer: DEVONthink 1.x stores HTML (including images) in the monolithic database, which must be loaded into memory when the database is opened. That’s also true of images in WebArchive and RTFD documents.

DEVONthink 2 significantly reduces memory required to load the database, as the HTML pages, incuding images, will be stored in the Finder. Images will not be loaded until the page is displayed in the database.

It is safe to say that DTP 2 has been my most awaited app upgrade ever. :slight_smile:

John

SUCH a tease :wink:

Removing the huge number of images from the database should reduce memory usage as each image requires some characteristics for See Also/Classify.

BTW:
Indexing/importing by V2 will require the same amount of memory (and less than indexing by V1.x).

Yes, when I clicked “See Also” DTP immediately consumed another 250 Mb on top of the 808 it was already using. Yikes! Guess I won’t be browsing Wikipedia articles while doing much else with the machine.

John