Does Nova include a JIT compiler? Or just an interpreter?
replies(1):
1. All data allocated on the JavaScript heap is placed into a type-specific vector. Numbers go into the numbers vector, strings into the strings vector, and so on.
2. All heap references are type-discriminated indexes: A heap number is identified by its discriminant value and the index to which it points to in the numbers vector.
3. Objects are also split up into object kind -specific vectors. Ordinary objects go into one vector, Arrays go into another, DataViews into yet another, and so on.
4. Unordinary objects' heap data does not contain ordinary object data but instead they contain an optional index to the ordinary objects vector.
5. Objects are aggressively split into parts to avoid common use-cases having to reading parts that are known to be unused.
If this sounds interesting, I've written a few blog posts on the internals of Nova over in our blog, you can jump into that here: https://trynova.dev/blog/what-is-the-nova-javascript-engine
The reasoning is that, according to my interpretation of talking with some folks working on JSC and SM, property lookup inline caching is the most important performance optimisation bar none. JIT compiling is an improvement on top, definitely, but it is not an massive step change.
Safari browser has a no-JIT mode that is fairly widely in use, and it is apparently fast enough that you don't really notice the change. Ladybird browser's LibJS has no JIT compiler, yet LibJS isn't really unbearably slow: The browser's biggest performance woes come from the browser around it and especially from having the simplest possible drawing algorithm possible.
From a "personal" experience, while the test262 compliance test set is no performance benchmark, Nova is for some reason consistently at the very top of the runtime list over at https://test262.fyi/#. This is of course partially just because we're really quick to do a controlled panic if an unsupported code path is called, and the remaining part is because the code is run so little that JIT doesn't get to kick in. Still, this meaningless number gives me some measure of hope: We're consistently 3 times as fast as V8 after all :)