Re: mysql scaling questions

> Does anyone have a theory why syscalls are so expensive
in FreeBSD? Here
> are the results of unixbench 4.1 on two machines. First
is the machine
> running FreeBSD HEAD (debugging disabled) on a
dual-core Athlon 64 (i386
> mode), 2 GHz:

I ran the syscall benchmark from UnixBench on the same
hardware (Intel
Q6600  3.2 GHz) for both FreeBSD 7.0-RC1 (amd64) and
Knoppix 5.1
(Linux 2.6.9 PREEMPT kernel, i386).

Per the comments in this thread, i386 syscalls should be
less
expensive than amd64 syscalls, so the results are
interesting:

FreeBSD 7.0-RC1 (amd64):  System Call Overhead
15000.0  1103233.7      735.5
Knoppix Linux 2.6.9 (i686):   System Call Overhead
    15000.0  1108869.7      739.2

I can try to find a Linux live CD with a 2.6.x x86-64 kernel
for a
valid apples-to-apples comparison, but thought these results
might
provide some parity/insight.

Doesn't look like the syscalls are slower at all on
FreeBSD.

Josh
_______________________________________________
freebsd-performancefreebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-p
erformance
To unsubscribe, send any mail to
"freebsd-performance-unsubscribefreebsd.org"

Josh Carroll wrote:
>> Does anyone have a theory why syscalls are so
expensive in FreeBSD? Here
>> are the results of unixbench 4.1 on two machines.
First is the machine
>> running FreeBSD HEAD (debugging disabled) on a
dual-core Athlon 64 (i386
>> mode), 2 GHz:
> 
> I ran the syscall benchmark from UnixBench on the same
hardware (Intel
> Q6600  3.2 GHz) for both FreeBSD 7.0-RC1 (amd64) and
Knoppix 5.1
> (Linux 2.6.9 PREEMPT kernel, i386).
> 
> Per the comments in this thread, i386 syscalls should
be less
> expensive than amd64 syscalls, so the results are
interesting:
> 
> FreeBSD 7.0-RC1 (amd64):  System Call Overhead
> 15000.0  1103233.7      735.5
> Knoppix Linux 2.6.9 (i686):   System Call Overhead
>     15000.0  1108869.7      739.2
> 
> I can try to find a Linux live CD with a 2.6.x x86-64
kernel for a
> valid apples-to-apples comparison, but thought these
results might
> provide some parity/insight.
> 
> Doesn't look like the syscalls are slower at all on
FreeBSD.
> 
> Josh
> _______________________________________________
> freebsd-performancefreebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-p
erformance
> To unsubscribe, send any mail to
"freebsd-performance-unsubscribefreebsd.org"
> 
> 

Per later discussion you will also need to either comment
out the 
syscalls that are (might be) being cached by glibc to
artificially 
inflate its reported rate, or verify that it is not doing
so.

Kris
_______________________________________________
freebsd-performancefreebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-p
erformance
To unsubscribe, send any mail to
"freebsd-performance-unsubscribefreebsd.org"

On 02/01/2008, Josh Carroll <josh.carrollgmail.com> wrote:
> > Does anyone have a theory why syscalls are so
expensive in FreeBSD? Here
> > are the results of unixbench 4.1 on two machines.
First is the machine
> > running FreeBSD HEAD (debugging disabled) on a
dual-core Athlon 64 (i386
> > mode), 2 GHz:
>
> I ran the syscall benchmark from UnixBench on the same
hardware (Intel
> Q6600  3.2 GHz) for both FreeBSD 7.0-RC1 (amd64) and
Knoppix 5.1
> (Linux 2.6.9 PREEMPT kernel, i386).
>
> Per the comments in this thread, i386 syscalls should
be less
> expensive than amd64 syscalls, so the results are
interesting:
>
> FreeBSD 7.0-RC1 (amd64):  System Call Overhead
> 15000.0  1103233.7      735.5
> Knoppix Linux 2.6.9 (i686):   System Call Overhead
>     15000.0  1108869.7      739.2

> Doesn't look like the syscalls are slower at all on
FreeBSD.

Thanks, I'll have to look further to see where the slowdowns
are.
_______________________________________________
freebsd-performancefreebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-p
erformance
To unsubscribe, send any mail to
"freebsd-performance-unsubscribefreebsd.org"