Is there any Revit versions performance comparisons. (R2025-R2022)?

[MALIN00152]

Is there a comparison of the performance of different versions of Revit, specially between R2022-R2025?

Some benchmarks? Article guide? I cannot find any hard data of that matter.

Currently my team is working on huge* project and I am curious if there would be any profits in performance (synchro time, opening view). If we would force to upgrade from 2022 to any newer version. On the other hand I am afraid of compatibility/plugin issues. So if there is no more like 10-20%** then upgrade make no real profits for us in terms of performance ... And can provide unknown problems to existing now.

We are MEP company, so Revit versions of model is more up to architects decision but I believe we can agree with architects with arguments.

*Shopping center, area size something between 200000-400000 square meters area, currently 32GB RAM seems problematic for fluent work with all linked models loaded and I expect that even 64-96GB would be required for full load all linked models after release execution design stage.

**less opening view time, requirements of RAM etc. Somebody maybe had done some test, like "after upgrade from Revit 2016 up to R2020, RAM usage got reduced/enlarged by XXX, opening views were more/less smooth".

I ask, because frankly speaking ... I don't know why, but I expect that after each version Revit, use more RAM memory not less xDD. (that might be, a bad feeling is that the models simply contain more and more data, when I started it happened that 15-40% of the objects were not modeled in Revit but in 2D AutoCAD as attach by various subcontractors, now, not counting the schematics, almost all the geometry are in the models.)

[RSomppi]

I would look at splitting up the project before upgrading. Everyone has excuses for not splitting up projects but, let's face it, it's already being done if you are using links and worksets (properly). Any performance improvements you might get from upgrading would be negligible compared to splitting up the model resulting in a lot of wasted time and effort doing the upgrade and troubleshooting anything that might/will come up during the upgrade.

Also, things like strategically opening worksets and loading links does help.

[HVAC-Novice]

I think you should start with a better PC. Computers are cheap compared to humans. And then set up the model the way it is convenient to humans. If you work on projects that size, 32GB isn't enough no matter how you split the model. There also is RAM generation, speed number of channels. Like an old dual-channel DDR 4 3200MT RAM will be much slower than a DDR5 6000MT quad or octa-channel. 

I've used every version you mentioned on same or similar projects and didn't notice a difference in performance. The newer versions are a bit more mature and fluent. And obviously new features (like tabbed project browser) make life easer. They do claim better multi-core use with every version. I don't know if that is true or not. I currently have a pretty outdated PC at work, so I may not be able to take advantage of some improvements. 

If you collaborate, you all have to be on the same version. that may be a more important reason than some features or performance. I personally try to upgrade to the new version end of the year it was released. But that is based on me wanting the new features, and hoping by that time all bugs are resolved. 

There are ways to model more efficiently. But that should be done regardless of version or hardware. Like don't model families to too much detail. 

It is extremely hard to benchmark a design software. You could model a garden shed with no electricity etc. Or you could model the World Trade Center with all mechanical and electrical systems. Plus all the way to model inefficiently, or efficiently.