Re: [patch 1/13] Qsort

[Matt Mackall]

On Sun, Jan 23, 2005 at 03:03:32AM +0100, Felipe Alfaro Solana wrote:
> On 22 Jan 2005, at 22:00, vlobanov wrote:
> 
> >Hi,
> >
> >I was just reading over the patch, and had a quick question/comment 
> >upon
> >the SWAP macro defined below. I think it's possible to do a tiny bit
> >better (better, of course, being subjective), as follows:
> >
> >#define SWAP(a, b, size)			\
> >    do {					\
> >	register size_t __size = (size);	\
> >	register char * __a = (a), * __b = (b);	\
> >	do {					\
> >	    *__a ^= *__b;			\
> >	    *__b ^= *__a;			\
> >	    *__a ^= *__b;			\
> >	    __a++;				\
> >	    __b++;				\
> >	} while ((--__size) > 0);		\
> >    } while (0)
> >
> >What do you think? :)
> 
> AFAIK, XOR is quite expensive on IA32 when compared to simple MOV 
> operatings. Also, since the original patch uses 3 MOVs to perform the 
> swapping, and your version uses 3 XOR operations, I don't see any 
> gains.
> 
> Am I missing something?

No temporary variable needed in the xor version. mov and xor are
roughly the same speed, but xor modifies flags.

-- 
Mathematics is the supreme nostalgia of our time.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/