lua-users home
lua-l archive

Re: bit.lshift and performance - luabitop v.s. lua-5.2.0-work4

[Date Prev][Date Next][Thread Prev][Thread Next] [Date Index] [Thread Index]


A lot of strange assumptions there..

First, "all languages that do not compile to native machine code have some overhead"
As opposed to JIT compiled languages that can take stuff into account discovered at runtime, which a static compilation cant?
Having a jit compiler can improve cache coherency for you, which is tedious to do otherwise, among other things.
Static compilers have had 20+ years headstart on research, which makes jit compilers downright impressive to me.
The fact that -one- person (mike pall) alone can produce a jit compiler that takes a relatively 'simple' language like lua,
and push it to the levels he has, says alot too.
Seriously, which compiler do you think will be more prevalent in say, 5-10 years time, when research -really- catches up?


Second, "any slight overhead is multiplied a huge number of times"
That's just weird and strange, and begs for a description of what you think "huge" means, and where that multiplication comes from?



On 2010年10月13日 15:32, HyperHacker wrote: [On 2010年10月13日 7:18 AM, "KHMan" <keinhong@gmail.com> wrote:

On 10/13/2010 8:52 PM, Bogdan Marinescu wrote:


>
> On Wed, Oct 13, 2010 at 3:40 PM, Enrico Colombini wrote:
>>
>> On 13/10/2010 14.12, KHMan wrote...

I agree with that. If there is a lot of bit ops, I think there is a good case to bite the bullet and use bitwise operators. I would want that too on an embedded system.



--
Cheers,
Kein-Hong Man (esq.)
Kuala Lumpur, Malaysia



AltStyle によって変換されたページ (->オリジナル) /