Re: Cracking DES with C++ is faster than Java?

From: Claudio Puviani (puviani_at_hotmail.com)
Date: 04/30/04 

Date: Fri, 30 Apr 2004 00:18:06 GMT

"Julie":
> > > I think that increases in speed will mitigate the need for most
optimizations,
> > > so my feeling is that hand optimization will die mainly due to lack of
need,
> > > rather than lack of ability.
> >

"Claudio":
> > That naively fallacious argument has existed since the early days of
computers,
> > and it's usually uttered by academics who have little contact with the real
> > world. No matter how fast a computer is, there will always be problems that
take
> > extremely long to process and the volume of operations that need to be done
per
> > unit time is growing faster than the hardware can keep up. A compiler that
> > performs worse optimizations (let alone none at all) will not be competitive
with
> > one that performs better ones. Certainly, a company that uses slower code
will be
> > at a disadvantage with respect to one that strives for efficiency. You can't
> > think in absolutes; it's relative performance that matters.

"Julie":
> Claudio, please follow the context of the thread before making such responses.
>
> The topic of discussion relates to HAND-OPTIMIZED ASSEMBLY, not optimizations
> in general.
>
> My feeling is that the need for hand-optimized assembly will diminish over time
> as baseline performance increases over that same period of time. That is what
> I was communicating, nothing more, nothing less.

I apologize if I misinterpreted your intent, but even in context, the statement,
"I think that increases in speed will mitigate the need for most optimizations",
seemed to apply to compiler optimizations as well.

Claudio Puviani

Relevant Pages

  • Re: Cracking DES with C++ is faster than Java?
    ... "Julie": ... > Claudio, please follow the context of the thread before making such responses. ... > The topic of discussion relates to HAND-OPTIMIZED ASSEMBLY, not optimizations ...
    (comp.lang.cpp)
  • Re: Cracking DES with C++ is faster than Java?
    ... "Julie": ... > Claudio, please follow the context of the thread before making such responses. ... > The topic of discussion relates to HAND-OPTIMIZED ASSEMBLY, not optimizations ...
    (sci.crypt)