If anyone has the resources to check, I'd love to see load timings. For example, I have groov running on a VM hosted in New York (we're in California) specifically to test longer-latency connections. (I should probably move it over to Amsterdam or something eventually.)
With the cache enabled, I'm down to 3-4 seconds to load the page. The single largest thing that gets loaded is my page definitions: this project has 1,449 gadgets spread across 84 pages, so it's on the larger side.
There's not much we can do about the cache-disabled case at the moment: the application is just large to begin with. That said, no one should be running with the cache disabled.
For the cache-enabled case, the biggest hit we take is loading all the pages. I've been looking at changing that recently: instead of loading them all at once, I would load the page index (just enough to render the page menu fully) and load the rest of the page definitions on demand. Prototype tests show it helps a lot, but it does mean it takes a bit longer to render a page for the first time.
There are also a few things we can collapse down into a single request: that's on the roadmap as well.
If anyone else wants to record how long it takes to load, open up your browser's development tools (on Windows, most browsers will pop them open when you press F12), look for something that allows you to record the Network timings, go to your groov instance and log in, then hit Refresh to record how long it takes to load and render. It should be fully rendered when you see a call to the