Hello everyone,
I recently got my computer running FME Form replaced and made some observations regarding performance that I wanted to lay out since I am not sure that everything runs as it should.
I am testing with a big workspace with close to 1000 transformers.
Coming from an aging Intel i5 6th Gen with 32 gigs of RAM and now sporting an Intel i7 14700-K with 128GB of RAM I was expecting some substantial performance uplifts.
The observation I made is this:
Yes, when I only read and write locally I am seeing a good boost in performance (a little over 2 times faster). However, in production this workspace is meant to write to a PostgreSQL database over the network. In this scenario my new computer is very close in performance to the old machine, even getting beaten by a few % in overall processing time (writing to the same database over the network).
Here is the numbers.
| New Computer | Old Computer | |
|---|---|---|
| Writing around 430.000 Features to database over network | 1hr 39m | 1hr 29m |
| Writing around 430.000 Features to GeoPackage locally | 12m 41s | 26m 52s |
I realize that writing over the network can bottleneck overall performance but considering the large amounts of processing I was expecting some kind of uplift, certainly not degradation.
Does this seem plausible or did I miss something obvious? Can the difference in writing speeds over the network attributed to variations in network speeds (both computers are tethered to the same network via cable but the sockets may be handled differently on the backend I am not sure on this)
Since I just installed FME on the new computer I was wondering if I maybe missed some setting that I may have switched in the past.
Any suggestions what I can try out or what changes I can make?
Thanks you all
