lkml.org 
[lkml]   [2001]   [Jan]   [10]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRe: * 4 converted to << 2 for networking code
Date
> 
> On Wed, Jan 10, 2001 at 06:03:22PM +0100, antirez wrote:
> > On Wed, Jan 10, 2001 at 09:54:04AM -0500, Brian Gerst wrote:
> > > This patch isn't really necessary, because GCC will automatically
> > > convert multiplications and divisions by powers of two to use shifts.
> >
> > Sure, but since many << 2 already exists in the net kernel code
> > I feel it's better to use just a format, and it seems more appropriate
> > to write << 2, just to reflect what we want.
> > Also some piece of kernel code may be used with compilers that does not
> > optimize power of two.
>
> On most processors <<2 is slower than *4. It's outright stupid to
> write <<2 when we mean *4 in order to optimize for one out of a
> gazillion supported architectures - even more so when the compiler
> for the one CPU where <<2 is faster, will actually generate a shift
> instead of a multiply as a part of the standard optimization.
>
> One question for the GCC people: Will gcc change <<2 to *4 on other
> architectures ? If so, then my case is not quite as strong of course.

Be careful. *4 is not a simple <<2 substitution (by the compiler) if
the variable is signed. *4 translates to 3 instructions (on x86) if
it's an int.

My feeling is that it shouldn't matter if you use <<2 or *4 even if the
compiler optimises - one would hope that the compiler would optimise to
the fastest in both directions.

Regards,

/Mike
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:28    [W:0.065 / U:1.736 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site