I have discovered what appears to be a 5x-10x slowdown in FMS19 compared to FMS18. The symptoms are:
- very slow calcuation update when one table is updating stored calculated fields fields using data from related table(s)
- Using FMS18, client and server CPU usage is near 100% on both sides, suggesting it's going as fast as possible.
- Using FMS19, by comparison, CPU usage on the server and client never gets much above 10%-20%.
- the low CPU usage would suggest client and server are starved for data (e.g. waiting for disk or network)
- however the same behavior happens when running the test using FMP on the server itself (via localhost) or when running in PSOS mode
- this is on a M1 Mac Mini, which has blazing fast SSD, so it's very unlikely the CPU could be starved for data.
In any case, my ask:
- is there anyone out there who is set up to do benchmarking using FM Bench | 24U Software who could run the full suite of tests on FMP18 vs. 19 and report back if there is anything showing up?
- I'm seeing a 5x to 10x slowdown on FMS19 vs. 18, it seems like one or more of the benchmarks should reveal this problem.
I would love to submit a bug report, but my ~2GB database is full of PII information, so I can't easily submit it to anyone to repeat my results.
Edit: @HOnza - tagging this for you.