Re: LuaJIT performance

[Alexander Gladysh <agladysh@...>]

On Mon, Aug 10, 2009 at 05:33, Mike Pall<mikelu-0908@mike.de> wrote:
> Alexander Gladysh wrote:
>> You've said there is no ETA yet for x86, but, perhaps, you may share
>> some information on the amount of work left to do?

> Well, I'm already cutting corners everywhere wrt. features for the
> first alpha. But issues with correctness and completeness keep me
> busy (the coordination between the JIT code and the GC is currently
> a minefield).

Do you consider to release a test suite along with LJ2?

> And the code needs to be cleaned up a lot before it's
> ready for public consumption. Then I'll need to work on the
> packaging, the docs, the web site reorganization and so on ...

If the community may help you anyhow, please do not hesitate to ask.
I'm sure we'll be able to lend a hand with routine tasks at least.

> Thankfully I've recently removed the last major stumbling block
> (better trace linking) and the benchmark results demonstrate that
> going for a trace compiler was a sound design decision after all.

That is good to hear!

> But I have to say it was an expensive decision: I've considerably
> underestimated the amount of research and trial-and-error which
> was needed to convert a research toy into a production compiler.
> There are some important implementation details which the few
> papers about trace compilers completely fail to mention ... :-|

Perhaps that is the reason to write one yourself someday? ;-)

Alexander.