RE: Lua and Neko comparison

["Grellier, Thierry" <t-grellier@...>]

"The difference with Lua is that Neko has value pointers, so no copying
is needed when passing values into the program. That actually means that
Neko is more efficient"

I'm not sure I agree with such a statement. Because locality is a very
important efficiency factor (due to CPU cache). And copying may be more
efficient than pointers to dereference or shifting integers. Larger
values could be used for value too in FFI. Although then they can't be
transmitted through registers as easily which might not be the case
anyway. Regarding FFI, I guess lua is relying upon macro too also, or at
least inline function then it shouldn't be an overhead. 
But do you really perform a malloc in alloc_float for example? Because
then return has also a terrible cost, although you may hope most of the
time it would be an integer.

You don't show how you call necko code from C. Is it possible?

Well I don't know actually if lua allocates with a malloc a stack for
each call...

-----Original Message-----
From: lua-bounces@bazar2.conectiva.com.br
[mailto:lua-bounces@bazar2.conectiva.com.br] On Behalf Of Nicolas
Cannasse
Sent: Tuesday, September 19, 2006 7:28 PM
To: lua@bazar2.conectiva.com.br
Subject: Lua and Neko comparison

Hi list,

I have been working in the past years on the Neko VM (http://nekovm.org)
which share similar goals with Lua. I was interested in comparing the
different choices made by the two VM and how they can impact the
performances, so I wrote a comparison report available there :

http://nekovm.org/lua

I would like to get reviews and feedbacks from people that have been
working on Lua VM or have knowledge about its internals.

Best,

Nicolas