DT uses about 78 MB real memory when opened. If I do a search it consumes a couple hundred MB real memory, which I understand. What I don’t understand is why it doesn’t release the memory when the search is concluded and cleared. Is this going to be addressed in the Pro version? Even with a gig of memory in my powerbook having a single program grab this much causes slowdowns with all the programs open that I usually run.
ChemBob
ChemBob:
Memory usage is the same in DT Pro. At the moment, DT Pro is showing use of 166 MB real RAM and 466 MB virtual memory.
I’ve got 1 GB RAM also. I’m thinking about getting a 20 inch iMac and putting in the maximum 2 GB RAM, but I’m not certain that would solve the RAM usage either.
I don’t know what the developers options are for release of real and/or virtual memory, but I do wish there were a “magic button” that I could press to release memory! 
I have been noticing the same huge slowdown after doing a search - DT becomes essentially unuseable. But this only happens when I use the actual Search Panel. The toolbar search does not cause this slowdown. I just noticed this tonight as I have not used the search panel in quite some time – I had just become accustomed to using the toolbar search and had not had any need to use the added search features. Then tonight I did, and began to notice this problem. For the record, I quit and reopened DT several times and each time once I initiated a search panel search I got these horrible slowdowns and DT would hang on almost every action, no matter how simple, for about 1-3 minutes. If I opened DT and only used the toolbar search things worked just fine, which is how I ended up doing the work I needed to do. Has anyone else noticed this difference in the search panel and the toolbar search?
Memory usage is for me going to decide whether I use Spotlight or DT. If I can’t have DT permanently open without it hogging lots of RAM, I’ll forsake its additional features for the more frugal combination of the Finder and Spotlight.
Spotlight is neat, but it doesn’t even begin to do for me what DEVONthink does. Spotlight makes its own demands on CPU, RAM and disk space. I’ll use both under TIger – and continue to envy those with faster CPUs, more RAM and bigger and faster drives!
Real soon now, Apple will kick up the specs to another generation level. I’ll get a new Mac.
And then DEVONtechnologies will be able to program for still more CPU and memory intensive artificial intelligence features in DEVONthink.
The cycle will continue.
I’m glad that DEVONthink pushes the envelope, because I like the things it lets me do.
But there are a lot of sophisticated mac programs that release the memory after completing a process. It’s OK to hog it when something is happening, but not when the program is just idling in the background, IMHO. I’ve got the latest 17 in PowerBook with a gig of memory and I don’t like the slowdowns in overall processing of all my other programs when DT wants 200 or more MB of the total. I often have 15 or more programs running at the same time and all the others are playing nice comparatively. I think the next worst is Word that sucks up about 0.1 gig. Adobe Photoshop is a big hog, of course, but I don’t use it much. I just think that if memory usage can be cut back during idle periods that this would be a big plus for DT.
ChemBob
A quick update on the issue I mentioned above. This morning, after having let my computer sit off all night and rebooting I opened DT and tried the same actions and searches as last night. DT worked much better, even from the Search Panel – not real zippy, but at first there were no beachballs. Then after about 5 minutes of using DT the bachballs began to appear again. Only for 15-20 seconds, but still… all I was doing was copying text from found results in the search panel and pasting them into notes in a DT RTF note. Switching windows would cause a beachball, pasting the text another… and so on. Of course this makes DT very frustrating to use. Don’t have time to research this problem more, but wanted to report on the improved behaviour after startup and then slowly degrading speed thereafter. Any suggestions are welcome… (I am using a PowerBook G4, 768MB RAM, 1 GHZ)
I had similar problems on my machine, a G5 2x1.8 GHz, and I think this should not happen. Everytime I had my database checked and made a back up, things were fine for some minutes. I wondered whether DT would always update pages from the internet, whether the db was too large, I unchecked automatic wiki links.
After all, I started to divide my almost 4GB db into several dbs again (of course, no solution for PE). Now I have one week off university to clean my stuff and hopefully, I can reunite all the smaller dbases again.
The small db (1.9 GB incl file folder and backups; some 3 mio words) I am working with mainly works fine now.
I am sure Christian is working hard on that problem.
Best,
Maria
Does this still happen using v1.9.1 and the same (!) search settings in the toolbar and the panel? Because the memory usage of both methods is identical over here (just checked this). Talking of memory usage - if you’re storing huge amounts of data, it’s definitely recommended to index static material instead of importing it.
By the way, version 2.0 will work that way - all contents will remain files (to be compatible to Spotlight) and DT will just use its own metadata & index necessary for searching, classifying, see also etc. Will reduce memory usage (depending on the nature of the contents) 1.5-4 times (and I know that won’t be enough because people will probably store 4 times more data then )
If this is the case, then I’m definitely much keener on DT. Is the 2.0 release dependent on Tiger, then?
No, DT 2.0 will remain OS X 10.3.x compatible (but we’ll probably drop support for 10.2.x as soon as Tiger will be available).
I am using v1.9.1. Like I said earlier these were some observations but I have not had time to run tests on this to try to isolate the issue. Regarding Indexing vs Importing: I have a collection of very large static documents. First I imported them all into DT RTF’s. Then after reading the same suggestion to Index in the manual, I got rid of the imported docs and brought them all back into DT as Indexed. Recently I went back to the old way because it seemed that the Indexed documents took longer to “load” onto my screen and I thought the actually Imported documents were being “loaded” faster… is that just my imagination? or does DT work a little slower with Imported documents since it still has to reference a file outside the DT file structure?
Displaying indexed files should be (more or less) as fast as displaying imported ones. But as indexed documents need of course less memory and database space, indexing does improve the overall performance (especially if there’s not enough real memory available).
In addition, the indexing process is faster than the importing process (up to 2.5 times) - if you want to manage more than one GB of data within DT, indexing is the way to go.
Note: DT 2.0 will also be able to edit indexed documents.
I’ve considered getting the same machine and maxing it out. Great value for many people, but the 2GB RAM limit is bad for people like us, and will probably stop me from doing it. Other problem is Firewire 400, not 800. If they changed those two things about the 20-inch G5 iMac, I’d buy it in a second.
These sound like significant changes and improvements. I look forward to it…
Maybe I’m not quite understanding what’s meant here by “indexing,” but when the developer says it’s “the way to go,” I listen. (And, yes, I will be doing big dbs.)
The DT user’s manual p. 21 clearly implies that “Copy files into database” is generally better than “Link to originals” or “Copy files to database folder.” So that’s how I set my db up.
When Pro and 2.0 come along, how are these embedded files going to be handled?
Will there be automatic conversion of PE dbs to Pro dbs, and 1.9 dbs to 2.0 dbs?
Also, plain and rich text files in DT are never part of the OS file system in PE 1.9, are they? Will this change in Pro and 2.0?
Version 2.0 will store all contents as files (including text documents) to be compatible to Spotlight and there will be no more “Copy into database option” therefore. Instead, importing will always copy the files to DT PE’s “Files” folder (or inside DT Pro’s file packages) whereas indexing will work the way the current indexing does.
Therefore all those currently available import options (copy/don’t copy etc.) and commands (import/index/link) will be reduced to the two commands “Import” or “Index” without sacrificing functionality (e.g. the “Link” command is more or less identical to “index unknown files” and therefore redundant).
Finally, DT Pro will be able to use DT PE database. But version 2.0 might need a reduild (can’t tell for sure, v2.0 will need a few more months).
That really got my attention! I also understood from an earlier comment in this thread that the 2.0 database will move to storing individual files (is that correct)?
I can see some major, exciting potentials in this. Emphasizes DT as an information manager rather than a file manager. Clears the decks for more efficient use of CPU and RAM resources to do the things that DT is best at, such as its AI features.
One factor that has restricted my use of index imports of PDF files is that I can’t edit such imports to add notes, comments or links (“value added” information). Another limitation in earlier betas of DT Pro was that “phrase” searches could not be run on index-imported documents (I think that limitation is now gone). Still another limitation is that search terms are not highlighted in the search view – a limitation that, I hope, will disappear under Tiger.
I’m hopeful that index imports of Word documents will go much further in rendering the original layout under the improved Tiger Cocoa text engine. Again, it would be great to edit such files with one’s ‘metadata’ about the file.
Christian, although you have emphasized the usefulness of index import for large ‘static’ PDF and Word files, users often do want to edit/update Word documents, and have the changes reflected in their DT database. But if I’ve been given the opportunity to add notes and hyperlinks to my DT version of a Word file, I would like to retain that information if the original Word file is updated and then the DT version is updated. Just deleting and replacing the DT index import copy would lose my notes and links. How about this: For editing such index imported Word files (and other file types), give me a ‘metadata’ pane (larger than the Comments window) in which I can comfortably write, and in which I can make links. Then set up a synchronization technique so that if I edit the original Word file, my DT index import is updated, but my ‘metadata’ pane is retained.
Might be not as powerful as requested but it’s already possible to add comments and at least one link to indexed material. And synchronizing indexed files won’t modify this meta data.