Re:Have Tracing JIT Compilers Won?
by silentbicycle on Wednesday March 10, 2010 @11:48AM
(#1181300)
One important issue is that compilation often needs to happen in bounded time. A JIT-compiler that does overly elaborate optimizations at runtime may lead to excessive pauses, and sometimes that's a dealbreaker. (Some of the same trade-offs appear with non-incremental garbage collection.)
[ Parent
| on HN ]
Re:Have Tracing JIT Compilers Won?
by Xichekolas on Wednesday March 10, 2010 @12:29PM
(#1181430)
Many tracing JIT implementations do the compilation in a separate thread from the running VM, only providing the compiled code when it's ready. This reduces the overhead considerably, and should pretty much prevent any pauses in execution (assuming a CPU that can reasonably handle multiple threads).
There is always the risk of unbounded compilation time as you say... or at least compilation time that runs long enough that the compiled code isn't ready in time to be useful, but empirical evidence of tracing has shown it to be a net win (over interpretation) in every implementation I have seen, so this risk must be small or at least manageable.
Re:Have Tracing JIT Compilers Won?
by silentbicycle on Wednesday March 10, 2010 @01:21PM
(#1181557)
I'm not in any way saying that it rules out JIT compilation* , just that it's a a constraint that may not have occurred to people, and one that gives it a different (often complementary) focus than static analysis.
* I'm delighted that the LuaJIT port to amd64 is almost done, BTW. I've submitted two threads about it that haven't made it to the front page.