file write performance drop between 2.5.60 and 2.5.70

file write performance drop between 2.5.60 and 2.5.70

Post by Vladimir Savelie » Thu, 05 Jun 2003 18:20:24



Hi

It looks like file write performance dropped somewhere between 2.5.60 and
2.5.70.
Doing
time dd if=/dev/zero of=file bs=4096 count=60000

on a box with Xeon(TM) CPU 2.40GHz and 1gb of RAM
I get for ext2
2.5.60:         real    1.42 sys 0.77
2.5.70:         real 1.73 sys 1.23
for reiserfs
2.5.60:         real 1.62 sys 1.56
2.5.70:         real 1.90 sys 1.86

Any ideas of what could cause this drop?

Thanks,
vs

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in

More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

 
 
 

file write performance drop between 2.5.60 and 2.5.70

Post by Felipe Alfaro Solan » Thu, 05 Jun 2003 20:10:09



> Hi

> It looks like file write performance dropped somewhere between 2.5.60 and
> 2.5.70.
> Doing
> time dd if=/dev/zero of=file bs=4096 count=60000

> on a box with Xeon(TM) CPU 2.40GHz and 1gb of RAM
> I get for ext2
> 2.5.60:    real    1.42 sys 0.77
> 2.5.70:    real 1.73 sys 1.23
> for reiserfs
> 2.5.60:    real 1.62 sys 1.56
> 2.5.70:    real 1.90 sys 1.86

> Any ideas of what could cause this drop?

What filesystem are you using?

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in

More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

 
 
 

file write performance drop between 2.5.60 and 2.5.70

Post by Andrew Morto » Fri, 06 Jun 2003 04:00:14



> Hi

> It looks like file write performance dropped somewhere between 2.5.60 and
> 2.5.70.
> Doing
> time dd if=/dev/zero of=file bs=4096 count=60000

> on a box with Xeon(TM) CPU 2.40GHz and 1gb of RAM
> I get for ext2
> 2.5.60:    real    1.42 sys 0.77
> 2.5.70:    real 1.73 sys 1.23
> for reiserfs
> 2.5.60:    real 1.62 sys 1.56
> 2.5.70:    real 1.90 sys 1.86

> Any ideas of what could cause this drop?

hm, 2.5.60 was a long time ago.  The best way to tell would be comparative
oprofiling.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in

More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

 
 
 

file write performance drop between 2.5.60 and 2.5.70

Post by Mike Fedy » Fri, 06 Jun 2003 20:40:16




> > Hi

> > It looks like file write performance dropped somewhere between 2.5.60 and
> > 2.5.70.
> > Doing
> > time dd if=/dev/zero of=file bs=4096 count=60000

> > on a box with Xeon(TM) CPU 2.40GHz and 1gb of RAM
> > I get for ext2
> > 2.5.60:       real    1.42 sys 0.77
> > 2.5.70:       real 1.73 sys 1.23
> > for reiserfs
> > 2.5.60:       real 1.62 sys 1.56
> > 2.5.70:       real 1.90 sys 1.86

> > Any ideas of what could cause this drop?

> What filesystem are you using?

Good one.

ext2 and reiserfs.

Read the origional message again.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in

More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

 
 
 

file write performance drop between 2.5.60 and 2.5.70

Post by Felipe Alfaro Solan » Fri, 06 Jun 2003 21:30:30





> > > Hi

> > > It looks like file write performance dropped somewhere between 2.5.60 and
> > > 2.5.70.
> > > Doing
> > > time dd if=/dev/zero of=file bs=4096 count=60000

> > > on a box with Xeon(TM) CPU 2.40GHz and 1gb of RAM
> > > I get for ext2
> > > 2.5.60:  real    1.42 sys 0.77
> > > 2.5.70:  real 1.73 sys 1.23
> > > for reiserfs
> > > 2.5.60:  real 1.62 sys 1.56
> > > 2.5.70:  real 1.90 sys 1.86

> > > Any ideas of what could cause this drop?

> > What filesystem are you using?

> Good one.

> ext2 and reiserfs.

> Read the origional message again.

Hmmm... What filesystem was being used?
Arghh!!! I'm blind!!!!!!!

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in

More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

 
 
 

file write performance drop between 2.5.60 and 2.5.70

Post by u.. » Fri, 06 Jun 2003 21:30:35


[...]




> > > > I get for ext2
> > > > 2.5.60:     real    1.42 sys 0.77
> > > > 2.5.70:     real 1.73 sys 1.23
> > > > for reiserfs
> > > > 2.5.60:     real 1.62 sys 1.56
> > > > 2.5.70:     real 1.90 sys 1.86
> Hmmm... What filesystem was being used?
> Arghh!!! I'm blind!!!!!!!

you should be going to sleep :-)

        Ulisses

                Debian GNU/Linux: a dream come true
-----------------------------------------------------------------------------
"Computers are useless. They can only give answers."            Pablo Picasso

---> Visita http://www.valux.org/ para saber acerca de la  <---
---> Asociacin Valenciana de Usuarios de Linux             <---

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in

More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

 
 
 

1. 2.5.70-mm2 causes performance drop of random read O_DIRECT

Starting in 2.5.70-mm2 and continuing in the mm tree, there is a
significant degrade in random read for block devices using O_DIRECT.  
 The drop occurs for all block sizes and ranges from 30%-40.  CPU usage
is also lower although it may already be so low as to be irrelavent.

                                 tolerance = 0.00 + 3.00% of 2.5.70-mm1
             2.5.70-mm1   2.5.70-mm2
 Blocksize      KBs/sec      KBs/sec    %diff         diff    tolerance
---------- ------------ ------------ -------- ------------ ------------
      4096         1567          924   -41.03      -643.00        47.01  *
      8192         3057         1815   -40.63     -1242.00        91.71  *
     16384         5745         3509   -38.92     -2236.00       172.35  *
     65536        17357        11283   -34.99     -6074.00       520.71  *
    262144        37537        27302   -27.27    -10235.00      1126.11  *

Full results can be found at:
http://www-124.ibm.com/developerworks/oss/linuxperf/regression/2.5.70...

Steve

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in

More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

2. Does anyone still use Diskless Clients today?

3. [2.5.70][ANNOUNCE] kexec for 2.5.70 available

4. mounting linux procfs...

5. 2.5.70 and 2.5.70-mm3 hang on bootup

6. SCO 5.0.2 NSF is slow - VERY slow What am I doing wrong??

7. ALSA Segmentation fault load modules 2.5.70 and 2.5.70mm1

8. Occam

9. TIObench 2.5.60 performance

10. 2.5.60 Lmbench performance

11. No /proc/*/wchan files in 2.5.60 - dbench does nothing

12. 2.5.60 - drivers/char/esp.c vs include/linux/serialP.h