After reading all the comments and doing some more research about the M1 and M2, I think going with the M1 with more RAM is the smarter choice.
I decided to sell my Pro Display XDR so I can afford the most RAM. If anyone is interested in a brand new Pro Display XDR feel free to contact me for more information. It’s the nano-texture version and comes with the VESA mount adapter. Everything is still brand new and in its original packaging. I never had a chance to use it yet. The one year warranty has expired though but you can always get a third-party warranty. It’s located in Spain near Granada. With the VESA mount Apple sells it for €6718, but of course I’m prepared to go much lower than that.
I think you may find that the DEVONthink database depends on the file system (each file resides in a file in the file system). I think your issue may be having so many Groups which is unusual and not using Groups as intended in DEVONthink (as others have mentioned). Maybe not an issue, dunno. Just seems very unusual. There are so many ways to do automation (Hazel is outstanding as a first step) with files in the file system. But without knowing and thinking more about your situation I can’t say. But as it is just doesn’t feel right. Just my two-bits.
I’m very interested in improving my current system and your thoughts give me something to think about. At some point in the future I might share a video of my setup and workflow here in the forum to learn more about how things could be improved. Like you said without knowing more about my current setup it’s difficult to think about it.
Just in case anyone is interested, Howard Oakley has written an extremely informative series of articles regarding M1 (and now, M2) Macs, including discussions of M1 memory management with comparison to Intel Macs’ management, etc. The link below connects to an index of articles he’s maintaining. Might help in terms of planning for future systems and purchases.
One important difference between web pages and PDFs is the very extensive use of scripts in modern web pages. Web pages are in general much less efficient than PDFs for a given amount of user-visible content.
The one I specifically had in mind was computational load. The rendering device has to parse any scripts even if it’s ultimately going to discard them. PDFs simply don’t include the elements that tend to make web pages slow.
PDFs in my world tend to be big (mb) and the whole file has to be downloaded before it becomes readable. A server computing HTML files to client regardless of the server’s computational load seems to send smaller files and I suspect browsers can display partially downloaded pages to make it seem to go fast for the reader. Just my view.
Above, rfog: “I have about 500 GB mostly of PDF across 18 databases (and billions of words), all synchronized between 3 Macs, 2 iPad and one iPhone.” Clearly, all these are somehow networked. It seemed to me that his existing data base may be distributed over the two machines. It might be possible to do searches, say, over the two machines and then merged. Back in my working days, I used to set up 3 or 4 asynchronous queries, and process the results after all were finished, sometimes setting more asychon sessions down the production line.
I have all in same Mac. In fact, I have the same in 3 Macs and 3 iOS devices (only on-demand), synchronized across a WebDAV server in my NAS.
Each Mac has its own complete local DT databases set in its 1 TB disk. Difference is search speed.
Having two Macs without enough disk space, you can put databases in external disk (I had that in my old MacBook Pro 2012, external disk glued with Velcro in the lid and attached with a normal USB short cable), or if you are patience enough and have very good connection, in a NAS.
Or put half of your databases in one Mac and the other half in the other, but they will be completely independent.
The timing of this question was quite great as I too am overdue for a hardware upgrade, and running practically the same current hardware as @cubicray (in my case; a late 2013 MBP w/ 4 core i7 (2.66 GhZ), 16GB RAM).
My goal is to merge two DT databases into one. These databases were split only because of hardware limitations, so I usually only run one at a time–which is not ideal at all because I constantly forget to search both of them.
162-million words; 1.7 million unique
89-milloin words; 881,000 unique
all items are old books and journals OCRed through DT (so, ABBYY API)
Glancing on the Apple Store website, I see the 16" MBP with the new M1 Max processor starts out w/ 32 GB RAM as the baseline. This can be doubled for $400 USD, and $400 bucks seems reasonable in comparison to the overarching cost of a new machine.
I like saving a buck though, so my $400 question is; is this much RAM truly worth it? Or is this extreme overkill given the crazy-impressive benchmark metrics on these new unified M1 chips?
My DT databases are comprised of only textual documents, but some are quite large files in word-count and file size (i.e. old books and journals that I re-OCRed using DT/ABBYY). I’ve been known to try crazy things before though, like combining an entire multivolume set of books or journal issues into one ginormous PDF (sometimes RTF) for convenience in using DT’s amazing NEAR operator across the entire publication, without having to jump to a new document. …but when I try putting this grand idea into practice on this old dog of a machine, I realize instead I must go reach for a fire extinguisher (beachballing, fan on high, etc.).
So what do you think, fellas? Is 64 GB RAM completely insane?