Re: version of euphoria ZIP
Juergen Luethje wrote:
>
> Vincent wrote:
>
> > Juergen Luethje wrote:
> >>
> >> Vincent wrote:
>
> [snipped (not necessary to repeat the same stuff over and over again)]
>
> >>> Obviously effeciency is not top priority for you, but that is quite ok.
> >>
> >> This is not "obvious", but false and a simplistic statement.
> >> Writing efficient code does not just mean using some machine specific
> >> routines with comments. You might want to read the Euphoria
> >> documentation carefully, in order to learn more about writing efficient
> >> code.
> >
> > Well excuse me, when did I *ever* imply that using machine specific
> > routines is the only way to improve program efficency?
>
> No, you did not imply that it's the *only* way to improve program
> efficency. But when you write:
> "Obviously effeciency is not top priority for you"
> just because I do not agree with you, that generally using machine_proc()
> is a good idea, then that statement implies that you think it's an
> *important* way to do so. But that is not the case (as Matt also has
> pointed out the day before yesterday).
>
Hi Juergen,
It's not "important" sort of speak, but despite what Matt said, it *is* one of
the *many* things you can do to improve program efficiency, in addition to
improving your algorithms.
It doesn't matter which way you choose as machine_proc/func is used anyway.. But
directly using them eliminates the unnecessary including, scanning/parsing, extra
subroutine calling, type definition (minuscule), variable (when using
machine_proc instead of machine_func), and constant overhead. All this overhead
is realatively small and hardly casts its shadow on Euphoria's blazing execution,
but internally its quite noticable. Add a trace(1) into scanner.e then step
through a few programs with the PD- source... perhaps you will notice all the
expensive internal process of including files and scanning/parsing them.
I'm not against "including" in general, I'm just suggesting not to abuse
the feature if efficiency is really important to you.
> > One of the most important factors in determinding efficency has to do
> > with the algorithms one develops.
> >
> > FYI sir, I have read the Euphoria documentation several times, and
> > overlook perform.htm/doc tips with everything I do in Euphoria.
>
> OK, then you probably know some ways how to really make a program more
> efficient.
>
> >> For instance I use profile and profile_time to examine programs.
> >> Depending on the program, there are several ways to make it more
> >> efficient. Sometimes I write parts of a program in assembler.
> >
> > I suppose peeking & poking machine code is easily readable too, correct?
>
> As I wrote, I do so *sometimes*, in selected cases. I'm not advocating
> for writing all programms in assembler.
> It's like in medicine: We should get a correct diagnosis first, in order
> to be able to choose an appropriate therapy.
>
Thats fine... you have your reasons to use it as I have mine to use machine
routines. Since I dont really know any ASM, this is not an option for me. So I
have to come up with other solutions like using machine routines to compensate.
Of course using machine code offers complete low-level control, so eventually
I'll need to at least partially understand it.
> Regards,
> Juergen
>
> --
> Have you read a good program lately?
>
>
Regards,
Vincent
|
Not Categorized, Please Help
|
|
