Re: Compiler inserts redundant comparison against zero
- From: Waldek Hebisch <spamtrap@xxxxxxxxxx>
- Date: Mon, 14 Nov 2005 21:19:36 +0000 (UTC)
spamtrap@xxxxxxxxxx wrote:
> I've been led to believe that modern optimising
> compilers are so sharp that it would be a waste
> of time writing my own assembly language code
> for general purposes.
>
> So I was surprised to find out how the C compiler
> I use, described as having "world-class
> optimization, both for speed and size", deals
> with the fragment:
>
> if( x >= 0 && x < 14 )...
>
> which is often encountered, especially in validation
> code (I mean the zero is often met with, the 14 could
> be anything).
>
> The compiler issues these instructions:
>
> test eax, eax ; value of variable x already in eax
> jl condition_false
> cmp eax, 0Eh
> jge condition_false
> condition_true:
> etc
>
> instead of the economical:
>
> cmp eax, 0Eh
> jae condition_false
> condition_true:
> etc
>
> Is this optimisation not well-known?
>
It is well-known. From C code:
extern int y;
void
foo(int x)
{
if( x >= 0 && x < 14 )
y = 1;
}
gcc -O gives:
foo:
pushl %ebp
movl %esp, %ebp
cmpl $13, 8(%ebp)
ja .L1
movl $1, y
..L1:
popl %ebp
ret
this assumes that the target is i386. Works both with new gcc (3.4.3 and
3.3.6) and quite old (2.8.1 and 2.95.3).
If I use -O -march=athlon-xp I get jumpless code:
foo:
pushl %ebp
movl %esp, %ebp
cmpl $13, 8(%ebp)
movl $1, %eax
cmova y, %eax
movl %eax, y
leave
ret
--
Waldek Hebisch
hebisch@xxxxxxxxxxxxxxxx
.
- References:
- Compiler inserts redundant comparison against zero
- From: spamtrap
- Compiler inserts redundant comparison against zero
- Prev by Date: Get the FAQs
- Next by Date: Re: Combining two MMX registers into one SSE register?
- Previous by thread: Re: Compiler inserts redundant comparison against zero
- Next by thread: Re: Compiler inserts redundant comparison against zero
- Index(es):
Relevant Pages
|
