A single page load of each cannot tell you the true difference in speed. There are too many variables that affect how fast it will go. Using something like Apache Bench, which I provided my results from before, is much more accurate because it averages the speed based on several hundred requests. This helps to remove many other variables you cannot control, like what else was happening on the server at the time, etc.
So - your 5x slower statement is wrong. According to the AB results I post earlier, a raw startup is nearly 2x slower. With your application, it's impossible for us to diagnose what is making it slower because there are so many variables involved. It could be a slower Query Builder, or cache driver, changes to your code when you made the conversion from 2.x to 3, etc.
While we will definitely look into speed issues, there's many things that aren't answered by your post that would be needed to even begin, like:
- are they running the same PHP version
- are the servers identical
- same DB version and table types?
Then, it would be helpful if you would start the process of diagnosing on your own application the slow parts. You can do this by profiling with XDebug and viewing the results in KCachegrind or Webgrind, or using a different profiler altogether. Let us know when you can narrow down the area that's really slower and that will help us know which areas to look into. And they have to be repeatably slower, not just once.
