Another Mindcradt case?

Another Mindcradt case?

[Jean Francois Martinez]

Moshe Barr is claiming on Byte that sendmail and mysql are 30% faster on
FreeBSD than Linux 2.4.   Now given that I don't think that mysql is spending
30% of its time in kernel mode there are not many ways FreeBSD can be 30%
faster.

1) Compiler issues.  Moshe Barr does not tell what distrib ha was using so we
don't know what distribution was used and with what 

2) Driver problems, specally those related to enabling UltraDma since without
UltraDMA most disks are both very slow and CPU hogs.

3) Options not enabled.  This a 2.2 Linux distrib with a 2.4 keernel plastered
on it and thus software has not been compiled to take advantage of 2.4bfeatures
and the boot sequence does not do a good job of tuning the kernel through
sysctl

4) Memory management.  If FreeBSD is smarter than Linux about which page to
thow out it will be _much_ faster when memory is tight.


Now what about a Mindcraft-style reaction?  Check what was wrong in the test
protocol, write an answer if it was due to the test and in case it was not the
test but a perfomance bottleneck fix it.

Moshe (and a few other people will do the same) was wondering why to stay with
Linux so better fix the problems if we want Linux and not BSD reaching world
domination. 


									JF
Martinez
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/

Re: Another Mindcradt case?

[Christoph Hellwig]

On Tue, Feb 06, 2001 at 09:25:10PM +0100, Jean Francois Martinez wrote:
> Moshe Barr is claiming on Byte that sendmail and mysql are 30% faster on
> FreeBSD than Linux 2.4.   Now given that I don't think that mysql is spending
> 30% of its time in kernel mode there are not many ways FreeBSD can be 30%
> faster.

As he has not even documented all the parameters:  Ignore him.

	Christoph

-- 
Whip me.  Beat me.  Make me maintain AIX.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/

Re: Another Mindcradt case?

[Roger Larsson]

Hi,

Please note that Moshe is (usually) very pro Linux (read his earlier columns).
I guess that he had expected Linux 2.4 to really beat FreeBSD - it did not...

/RogerL 

On Wednesday 07 February 2001 01:25, Jean Francois Martinez wrote:
> Moshe Barr is claiming on Byte that sendmail and mysql are 30% faster on
> FreeBSD than Linux 2.4.   Now given that I don't think that mysql is
> spending 30% of its time in kernel mode there are not many ways FreeBSD can
> be 30% faster.
>
> 1) Compiler issues.  Moshe Barr does not tell what distrib ha was using so
> we don't know what distribution was used and with what
>
> 2) Driver problems, specally those related to enabling UltraDma since
> without UltraDMA most disks are both very slow and CPU hogs.
>
> 3) Options not enabled.  This a 2.2 Linux distrib with a 2.4 keernel
> plastered on it and thus software has not been compiled to take advantage
> of 2.4bfeatures and the boot sequence does not do a good job of tuning the
> kernel through sysctl
>
> 4) Memory management.  If FreeBSD is smarter than Linux about which page to
> thow out it will be _much_ faster when memory is tight.
>
>
> Now what about a Mindcraft-style reaction?  Check what was wrong in the
> test protocol, write an answer if it was due to the test and in case it was
> not the test but a perfomance bottleneck fix it.
>
> Moshe (and a few other people will do the same) was wondering why to stay
> with Linux so better fix the problems if we want Linux and not BSD reaching
> world domination.
>
>
> 									JF
> Martinez
> --
> To unsubscribe, send a message with 'unsubscribe linux-mm' in
> the body to majordomo@kvack.org.  For more info on Linux MM,
> see: http://www.linux.eu.org/Linux-MM/

-- 
Home page:
  none currently

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/

Re: Another Mindcradt case?

[Jean Francois Martinez]

On Tue, 6 Feb 2001 23:02:01 +0100, Roger Larsson said:

> Hi,
>  
>  Please note that Moshe is (usually) very pro Linux (read his earlier columns).
>  I guess that he had expected Linux 2.4 to really beat FreeBSD - it did not...
>  

I was speaking about the reaction of pride Linux developpers had after the
Mindcraft "benchamrks" pointed to design flaws in Linux.  Here it is a BSD who
could be beating the pants of Linux

>  /RogerL 
>  
>  On Wednesday 07 February 2001 01:25, Jean Francois Martinez wrote:
>  > Moshe Barr is claiming on Byte that sendmail and mysql are 30% faster on
>  > FreeBSD than Linux 2.4.   Now given that I don't think that mysql is
>  > spending 30% of its time in kernel mode there are not many ways FreeBSD can
>  > be 30% faster.
>  >
>  > 1) Compiler issues.  Moshe Barr does not tell what distrib ha was using so
>  > we don't know what distribution was used and with what
>  >

Here should have read "we don't know what compiler was used and with what
compiler flags (like -O2 versus -O3)"

							JFM
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/

Re: Another Mindcradt case?

[Rik van Riel]

On 6 Feb 2001, Jean Francois Martinez wrote:
> On Tue, 6 Feb 2001 23:02:01 +0100, Roger Larsson said:
> 
> >  Please note that Moshe is (usually) very pro Linux (read his earlier columns).
> >  I guess that he had expected Linux 2.4 to really beat FreeBSD - it did not...
> 
> I was speaking about the reaction of pride Linux developpers had
> after the Mindcraft "benchamrks" pointed to design flaws in
> Linux.  Here it is a BSD who could be beating the pants of Linux

In some areas FreeBSD is certainly better than Linux.
In other areas, Linux is better than FreeBSD.

At the moment both systems mostly seem to be on par
with each other and in some subsystems there's even a
friendly coopetition between the developers on both
OSes.

(and yes, I'm sure they can do better on MySQL benchmarks
than Linux at the moment ... but this is something we are
working on)

regards,

Rik
--
Linux MM bugzilla: http://linux-mm.org/bugzilla.shtml

Virtual memory is like a game you can't win;
However, without VM there's truly nothing to lose...

		http://www.surriel.com/
http://www.conectiva.com/	http://distro.conectiva.com/

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/

Re: Another Mindcradt case?

[Jean Francois Martinez]

On Tue, 6 Feb 2001 21:27:04 +0100, Christoph Hellwig said:

> On Tue, Feb 06, 2001 at 09:25:10PM +0100, Jean Francois Martinez wrote:
>  > Moshe Barr is claiming on Byte that sendmail and mysql are 30% faster on
>  > FreeBSD than Linux 2.4.   Now given that I don't think that mysql is spending
>  > 30% of its time in kernel mode there are not many ways FreeBSD can be 30%
>  > faster.
>  
>  As he has not even documented all the parameters:  Ignore him.
>  

The fact he doesn't give all parms doesn't mean Linux could not have a
performance problem.	And  one thing is sure we have a propaganda problem
since the BSD people will mention this for ten years.  So better to solve both
problems than ignoring them.

							JF Martinez

                        
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/

Re: Another Mindcradt case?

[Kip Macy]

Jean Francois Martinez wrote:

> On Tue, 6 Feb 2001 21:27:04 +0100, Christoph Hellwig said:
>
> > On Tue, Feb 06, 2001 at 09:25:10PM +0100, Jean Francois Martinez wrote:
> >  > Moshe Barr is claiming on Byte that sendmail and mysql are 30% faster on
> >  > FreeBSD than Linux 2.4.   Now given that I don't think that mysql is spending
> >  > 30% of its time in kernel mode there are not many ways FreeBSD can be 30%
> >  > faster.
> >
> >  As he has not even documented all the parameters:  Ignore him.
> >
>
> The fact he doesn't give all parms doesn't mean Linux could not have a
> performance problem.    And  one thing is sure we have a propaganda problem
> since the BSD people will mention this for ten years.  So better to solve both
> problems than ignoring them.
>
>                                                         JF Martinez
>
>
> --
> To unsubscribe, send a message with 'unsubscribe linux-mm' in
> the body to majordomo@kvack.org.  For more info on Linux MM,
> see: http://www.linux.eu.org/Linux-MM/

In all due fairness the propaganda out there is overwhelmingly in favor of Linux.
Not to mention that its user base is close to 10x that of *BSD.

                                        -Kip

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/

Re: Another Mindcradt case?

[Rik van Riel]

On 7 Feb 2001, Jean Francois Martinez wrote:

> The fact he doesn't give all parms doesn't mean Linux could not
> have a performance problem. And one thing is sure we have a
> propaganda problem since the BSD people will mention this for
> ten years.  So better to solve both problems than ignoring them.

If you can reproduce this test and tell us exactly how we
can reproduce it here, then I'll take a look at the problem.

If you can get more details, like vmstat output, ..., could
you file a bug in the linux-mm bugzilla ?

(if you keep it on mail, you can be sure we'll all have
forgotten about this within 3 weeks)

regards,

Rik
--
Linux MM bugzilla: http://linux-mm.org/bugzilla.shtml

Virtual memory is like a game you can't win;
However, without VM there's truly nothing to lose...

		http://www.surriel.com/
http://www.conectiva.com/	http://distro.conectiva.com/



--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/

Re: Another Mindcradt case?

[Gerrit Huizenga]

It sure seems like this was a roughly fair comparison by friendlies,
as opposed to a de fecto, closed-source, self-perpetuating propoganda
machine out to steam roll possible competition.  Seems like it should
be responded to in the spirit of open source:  COOL!  So what was
broken in Linux/BSD and what technology did BSD/Linux use to fix it
that we can share?

End result could be to make them both better than any closed, proprietary,
over-large corporate product offering.

And, keep in mind that comparisons usually indicate that there are some
flaws - even if the flaw is that the user's weren't able to set up a
system that worked for them.  The key is to find and fix the flaws as
they are exposed.

Just my $0.03.

gerrit

> Jean Francois Martinez wrote:
> 
> > On Tue, 6 Feb 2001 21:27:04 +0100, Christoph Hellwig said:
> >
> > > On Tue, Feb 06, 2001 at 09:25:10PM +0100, Jean Francois Martinez wrote:
> > >  > Moshe Barr is claiming on Byte that sendmail and mysql are 30%
> faster on
> > >  > FreeBSD than Linux 2.4.   Now given that I don't think that mysql is
> spending
> > >  > 30% of its time in kernel mode there are not many ways FreeBSD can
> be 30%
> > >  > faster.
> > >
> > >  As he has not even documented all the parameters:  Ignore him.
> > >
> >
> > The fact he doesn't give all parms doesn't mean Linux could not have a
> > performance problem.    And  one thing is sure we have a propaganda
> problem
> > since the BSD people will mention this for ten years.  So better to solve
> both
> > problems than ignoring them.
> >
> >                                                         JF Martinez
> >
> >
> > --
> > To unsubscribe, send a message with 'unsubscribe linux-mm' in
> > the body to majordomo@kvack.org.  For more info on Linux MM,
> > see: http://www.linux.eu.org/Linux-MM/
> 
> In all due fairness the propaganda out there is overwhelmingly in favor of
> Linux.
> Not to mention that its user base is close to 10x that of *BSD.
> 
>                                         -Kip
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/

Re: Another Mindcradt case?

[Rik van Riel]

On Tue, 6 Feb 2001, Gerrit Huizenga wrote:

> It sure seems like this was a roughly fair comparison by friendlies,
> as opposed to a de fecto, closed-source, self-perpetuating propoganda
> machine out to steam roll possible competition.  Seems like it should
> be responded to in the spirit of open source:  COOL!  So what was
> broken in Linux/BSD and what technology did BSD/Linux use to fix it
> that we can share?
> 
> End result could be to make them both better than any closed,
> proprietary, over-large corporate product offering.

Exactly!

Now if the guys who submitted this thing to linux-mm
could take the time to help us find out what was going
on I'm sure we'll have a fix by 2.5.1 ...

(and maybe earlier)

regards,

Rik
--
Linux MM bugzilla: http://linux-mm.org/bugzilla.shtml

Virtual memory is like a game you can't win;
However, without VM there's truly nothing to lose...

		http://www.surriel.com/
http://www.conectiva.com/	http://distro.conectiva.com/

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/