jits

Tracing JITs like LuaJIT and PyPY as fast to warm up compared to method JITs and PEs.

This is already a problem for tracing JITs, but is supposedly worse for partial evaluation JITs.

The opposite is also true, smart tracing JITs won't do much for Common Lisp, but shine for JavaScript.

Tracing JITs' advantages are that they're simpler to write than partial evaluation JITs, and produce decent code faster, which is why they're used for JavaScript.

So IonMonkey, Crankshaft, SpeculativeJIT/FTL are all JITs that live inside a larger engine that switches between them as needed.

That should put our discussion, and the warmup performance of tracing JITs, on a firmer basis.

Andreas Gal was instrumental in getting tracing JS JITs of the ground, and until recently served as CTO at Mozilla.

State of the art in JITs has moved on so much that intent's fairly crude binary translation's not worth much.

He mentions that the demands of tracing JITs and current Intel CPUs are well aligned.

JITs can do gradual refinement, and for this they may implement multiple different compilation techniques.

You aren't burning all the cycles you think you are if you are applying typical arguments about the JVM or JavaScript JITs.

> JITs are free to inline beyond that as much as they want JITs are primitive and do not have much time to do anything interesting.

I don't think that tracing JITs are at all like "programming in a straight jacket." Look at the LuaJIT performance optimization guide.

In JS land we usually call the individual modes "JITs" even though they may (or may not) share a bunch of code.

For anyone interested in learning more about Tracing JITs and the problems involved in implementing them should read the paper "Trace-based Just-in-time Compilation for Haskell" by Thomas Schilling [1].