It might be possible. But you'd have to solve the following problems: - OSR exits at the rate of roughly one per opcode. I'm not sure if LLVM supports them at all and if it does, I'm not sure that it can efficiently support as many of them as our backend. We can provide meta-data for how to reify baseline state at almost any point, which allows us to do aggressive profile driven speculations. I don't know if LLVM would allow us to speculate quite so aggressively. And remember, LLVM's main source languages are all strongly typed, so speculation and OSR exit may be underexplored territory.
- Idiom tuning. How you tune a compiler that mainly sees C style code is very different from how you would tune a compiler for JS. This is due to (1) what idioms a programmer is likely to use and (2) what idioms are likely to be introduced by the frontend. If you wanted to get LLVM to compete with our backend you'd have to probably do some tuning. - Code patching. We rely heavily on self modifying code. I'm not sure if LLVM would give you quite enough flexibility to do that. If you removed code patching, btw, you'd get destroyed on every benchmark. - Speed of compilation. We optimize for this with great zeal. Often we avoid certain otherwise popular optimization strategies because we don't want to hurt page load times. That said, you could avoid this with the right combo of concurrent JIT and tiering. In my experience the first two points can kill you if you try to use an optimized-for-C compiler since C code requires only minimal code patching (often it's just an idiom that the linker and the C compiler agree on) and usually only supports side exits for exceptions, which are a lot simpler than OSR. -Filip On Feb 17, 2012, at 10:28 PM, "Michael.Kang" <[email protected]> wrote: > Hello WebKitten: > > I am a newbie here. Now I am investigating if we can use LLVM > as our JIT system of Javascript engine to get a faster Javascript > engine? I had > searched webkit-dev maillist (probably I missed some previous related > discussion ), and it seems there is only some discussions about > translating > llvm IR to javascript. > I know the speed of javascript engine is a fight among V8, SFX > etc. Several famous companies have already done a lot of thing on it. > But I will > think that why we can not translate the javascript to the IR of llvm > since LLVM have a good support for JIT system. Is it feasible? Could > we get a > faster JIT engine based on LLVM? I like to get your suggestion or > advice about that. Is there some special considerations or obstacle > for such way. > Any discussion about my stupid thinking is very appreciate. > > Thanks > MK > > -- > www.skyeye.org > _______________________________________________ > webkit-dev mailing list > [email protected] > http://lists.webkit.org/mailman/listinfo.cgi/webkit-dev _______________________________________________ webkit-dev mailing list [email protected] http://lists.webkit.org/mailman/listinfo.cgi/webkit-dev
