Re: Small but representative self-contained Lua benchmark?

[Alexander Gladysh <agladysh@...>]

On Fri, Apr 15, 2011 at 05:03, Geoff Leyland <geoff_leyland@fastmail.fm> wrote:
> On 15/04/2011, at 12:40 PM, Alexander Gladysh wrote:
>> On Fri, Apr 15, 2011 at 04:30, Geoff Leyland <geoff_leyland@fastmail.fm> wrote:
>>> On 15/04/2011, at 12:25 PM, Alexander Gladysh wrote:

>>>> For illustrative reasons I need a Lua VM benchmark script (note: not a
>>>> benchmarking tool, but a script).

>>> There's a raytracer here: http://www.incremental.co.nz/projects/lua.html
>>> (It's not intended for drawing pictures, it was just to compare Lua and LuaJIT)

>> Thanks.

>> You suggest that it should be modified that its output is suppressed
>> and execution time measured with os.clock(), right?

> Um, yes, that sounds good, why not?  I'm not sure what you want to measure :-)

Relative performance.

No, that way would work of course, no problem.

>> (Or os.time()? Not sure if os.clock() is available in exotic VMs.)

> Or just 'time lua ray.lua 9 512 > /dev/null' if you're on Unix?

Would be a bit hard to do with LuaAlchemy, I'm afraid. (And, I guess,
with some Java implementations too.)

>> This would do, but I would prefer something intended as a benchmark,
>> so it would be more illustrative...

> Fair enough.  I can't even argue it's a typical workload because I don't often trace rays.  It's also a bit odd because I deliberately avoided creating tables if I could.

Being illustrative beats being typical in this particular case.

Alexander.