Re: A89: Re: "Transfer" of values between C and ASM

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

Re: A89: Re: "Transfer" of values between C and ASM




>
> Ok, I can see your point in some of this but generally, ASM programs will
> run faster then C programs even if they take much longer to make. You have
> made good points and I was being too general.
>
> -'FuZeD
>
> b.t.w. a 400mhz x86 computer would most likely run more then 40x the speed
> of the 89. This is because desktop "computers" have more RAM, higher bus
> speeds, and better instruction sets. I'm not saying that a 400mhz 68k
                    ^^

This is not true. The 68k series has a far better ISA than the x86. Its
simply
more modern. Also the "desktop" versions of the 68k have the same
instruction set.

But anyways. A 400 Mhz P2/p3/Celeron does run up to 300-400 times faster
than the 10Mhz 68000 in your TI89, since it simply is able to do more
per cycle. Of course this is again limited by memory speed etc, as you
mentioned.

> processor wouldn't run 40x as fast but you catch my drift.
>
> -----Original Message-----
> From: Adam Davis <adavis@baladyne.com>
> To: assembly-89@lists.ticalc.org <assembly-89@lists.ticalc.org>
> Date: Monday, August 09, 1999 12:35 PM
> Subject: Re: A89: Re: "Transfer" of values between C and ASM
>
>
> >
> >}InFuZeD{ wrote:
> >>
> >> Ummm....No they cant =P
> >
> >Ummm?  You sound unsure.  You should be unsure, since a good C programmer
> can
> >tweak their program to compile to the same optomized asm code an
assembler
> would
> >produce.  Hello, let's think about this for a minute.  You know that C
> compiles
> >to assembly, which is then assembled to machine code.  With a good two or
> even
> >three pass compiler, made by those who know the processor inside and out,
> the
> >assembly produced by the compiler rivals that which is hand produced.
> There is
> >only a finite amount of tweaking which can be done to a program produced
in
> >assembly, and a good compiler used by a good programmer will produce code
> that
> >is that good.
> >
> >> Note that C is a "High Level" language and that Asm
> >> is a "Low Level" language =P
> >
> >So noted.  It is a common misconception that all high level languages are
> >inefficient and bloated.  Of course, that is the same as saying "All
> >generalities are false (Including this one)"  Please don't generalize.
> >
> >> You'd have to modify your C program ALOT w/
> >> bits of asm code to make it run as fast.
> >
> >That is not correct as long as you have a good programmer and a good
> compiler.
> >Secondly, you would not have to modify it "ALOT".  A decent programmer
with
> a
> >decent compiler will produce code where, while not efficient, and perhaps
a
> >teeny bit bloated, there would be little or no speed difference between
it
> and
> >asm code(to the user, on a 10MHz 68k).  Except, of course, the time it
took
> to
> >develop that program.
> >
> >> I'm not talking about on the
> >> computer because computers are fast enough either way not to really
show
> >> much difference.
> >
> >Ok, so what you are saying is that the 68k processor is not a
'computer'???
> >Please enlighten me as to your definition of 'computer'.  I suppose you
> mean
> >modern computers which run at 400+MHz.  Only 40 times faster than our
calc,
> I
> >might add.
> >
> >> The fact is that Asm programs are faster then those coded
> >> in C (if the asm programmer knows what he's doing =)
> >
> >Sorry, that is not a fact.  It is a generality.  Given two decent
> programmers,
> >one programming in assembly, the other in C, both using free programming
> >utilities on a familiar platform, the program from the first will be
> slightly
> >smaller, and slightly faster.  The second programmer will spend at least
> half as
> >much time coding and debugging.
> >
> >Of course, given two great programmers, each using professional
development
> >tools, and each tweaking for maximum effect, the output will be exactly
the
> >same... But the ASM programmer will still have spent twice as much time
as
> the C
> >programmer.  But then we're talking about programming tools in the 1-20
> grand
> >price range.  Everything is a trade off.
> >
> >>
> >> -'FuZeD
> >
> >I do not mean to sound like a jerk, but I program professionally, as well
> as a
> >hobbyist on both computers and embedded systems.  A good one, not a great
> one.
> >If there were a clear advantage to using ASM to make my programs, I would
> be
> >doing it.  The slight performance decrease in programs which I do not
tweak
> is
> >worth the gained time.  I can see though, that for those for whom it is a
> hobby,
> >it is thrilling to gain that slight increase - at the cost of time.  For
a
> >hobbyist, time is not directly proportional to money.
> >
> >-Adam
> >
> >P.S.  Now here's the funny part - I know C and C++, and would rather
> program in
> >it, but, get this - My employer has me program in Visual Basic.  His
> reasoning?
> >The development cycle is faster, and he and the other embedded engineers
> can
> >program it themselves if need be.  <sigh>  (that's one tradeoff I would
> rather
> >not make... But hey, They tell me what to do, I get a paycheck.  It's
> pretty
> >simple...)
> >
>
>
Follow-Ups: References: