Re: What I suspect : [PATCH] sysdat driver for faster gettimeofday()
John Wright (jwright@cthulhu.engr.sgi.com)
Thu, 9 Dec 1999 19:30:17 -0800
On Thu, Dec 09, 1999 at 08:36:31AM +0000, Tigran Aivazian wrote:
> Hi,
>
> About your /dev/sysdat patch - have you actually benchmarked and proved
> that it does indeed speed up gettimeofday(2) (or more precisely new
> gettimeofday(3) vs old gettimeofday(2))?
>
> Yes, conceptually your approach is obviously correct but seeing benchmark
> numbers is somewhat more pleasing to the eye...
>
> Regards,
> ------
> Tigran A. Aivazian | http://www.sco.com
> Escalations Research Group | tel: +44-(0)1923-813796
> Santa Cruz Operation Ltd | http://www.ocston.org/~tigran
It gives a large speedup to a loop test as I showed. Our database tests
show about a 5% speedup. There are other issues that we haven't quite
figured out... Depending on the disk driver, you may not get the speedup.
I'm kind of baffled by this... since it is page aligned, etc.
So in short, the concept looks like it has benefit, the implementation
is the question. This one works, but since Linus doesn't like this method,
we'll take a look at trying it another way.
--
John Wright (jwright@engr.sgi.com)
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/