Re: sometimes match or gmatch are very slow under luajit (compared to raw lua)
- From: Mike Pall <mike-1209@xxxxxxxxxx>
- To: luajit@xxxxxxxxxxxxx
- Date: Wed, 19 Sep 2012 10:05:58 +0200
Ronan Collobert wrote:
> I try to generate random files where this would happen, with no
> success: I really can't find the "trigger" which makes the thing
> (match or gmatch) slow.
The problem is the string interning and the partial hash used for
strings.
Actually, you can easily come up with an example where Lua 5.1
would hit the same problem, since it uses a (different) partial
hash, too. There are some postings on the Lua mailing list that
show examples for the weaknesses of the various partial hashes.
One solution for this problem would be partial string interning
and delayed hashing (like Lua 5.2.1). I can think of a couple more
options. Alas, this requires pretty extensive changes and lots of
experimentation to come up with the best compromise.
IMHO it's too late in the LuaJIT 2.0 release cycle to change that.
But I'll have to tackle this problem in LuaJIT 2.1, anyway (since
everything about the internal object layout is going to change).
> I noticed someone posted a somewhat related problem previously:
> //www.freelists.org/post/luajit/Luajit-string-split-much-much-slower-that-Lua-string-split
I don't think this is the same problem. The relevant strings are
rather short and fully hashed.
--Mike
Other related posts:
