Re: Linux 2.6.6 appears to be 3 to 4 times slower than 2.6.5

Re: Linux 2.6.6 appears to be 3 to 4 times slower than 2.6.5

[Daniel Blueman]

Alastair,

nForce2 chipset by any chance? There were changes for the C1 disconnect
issue in 2.6.6.

-- 
Daniel J Blueman

"Sie haben neue Mails!" - Die GMX Toolbar informiert Sie beim Surfen!
Jetzt aktivieren unter http://www.gmx.net/info

RE: Linux 2.6.6 appears to be 3 to 4 times slower than 2.6.5.

[Piszcz, Justin Michael]

It may not dause -data corruption- but it deleted a whole bunch of files
that were in use before the previous reboot.


-----Original Message-----
From: linux-kernel-owner@vger.kernel.org
[mailto:linux-kernel-owner@vger.kernel.org] On Behalf Of Bartlomiej
Zolnierkiewicz
Sent: Tuesday, May 18, 2004 9:23 AM
To: Justin Piszcz; baldrick@free.fr; gene.heskett@verizon.net
Cc: linux-kernel@vger.kernel.org
Subject: Re: Linux 2.6.6 appears to be 3 to 4 times slower than 2.6.5.

On Monday 17 of May 2004 17:06, Justin Piszcz wrote:
> Sorry to all, it turns out (in two separate cases I had two different
> problems that affected the results).
>
> Case 1: No SMP turned on for CPU w/HT after fix (~4.78 seconds compile
time
> (2.6GHZ w/HT))
> Case 2: Box had 4GB of NON-ECC memory in it, only recognized 2.56GB,
took
> out (2) 1GB DDR DIMM's, and the speed returned what it should be.
(~4.3
> seconds compile time (3.0GHZ w/HT))
>
> The control box was a 2.53GHZ (533MHZ BUS w/NO HT) = ~5.3seconds
>
> I have not tested 2.6.6 recently, but in one of my tests I believe it
> worked OK, ever since 2.6.6 removed my /etc/lilo.conf and /etc/mtab
and
> several other files, I do not wish to touch that kernel with a 10 foot
poll
> :-P due to the IDE disk flush/cache issue.

I told you this already: 2.6.6 IDE changes don't cause data corruption
- but fixes some instead (that's why there were merged so quickly!)
so stop spreading FUD and see
http://bugme.osdl.org/show_bug.cgi?id=2672.

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

Re: Linux 2.6.6 appears to be 3 to 4 times slower than 2.6.5.

[Bartlomiej Zolnierkiewicz]

On Wednesday 19 of May 2004 14:28, Piszcz, Justin Michael wrote:
> It may not dause -data corruption- but it deleted a whole bunch of files
> that were in use before the previous reboot.

deleting a bunch of files is a data corruption

You picked IDE cache fixes as a cause of your problems without any
verification or proof - 2.6.6 contains also a lot of VM and FS changes.

Stop spreading FUD or verify that IDE changes are to blame!
[ I will happily fix it if it turns out to be IDE. ]

> -----Original Message-----
> From: linux-kernel-owner@vger.kernel.org
> [mailto:linux-kernel-owner@vger.kernel.org] On Behalf Of Bartlomiej
> Zolnierkiewicz
> Sent: Tuesday, May 18, 2004 9:23 AM
> To: Justin Piszcz; baldrick@free.fr; gene.heskett@verizon.net
> Cc: linux-kernel@vger.kernel.org
> Subject: Re: Linux 2.6.6 appears to be 3 to 4 times slower than 2.6.5.
>
> On Monday 17 of May 2004 17:06, Justin Piszcz wrote:
> > Sorry to all, it turns out (in two separate cases I had two different
> > problems that affected the results).
> >
> > Case 1: No SMP turned on for CPU w/HT after fix (~4.78 seconds compile
>
> time
>
> > (2.6GHZ w/HT))
> > Case 2: Box had 4GB of NON-ECC memory in it, only recognized 2.56GB,
>
> took
>
> > out (2) 1GB DDR DIMM's, and the speed returned what it should be.
>
> (~4.3
>
> > seconds compile time (3.0GHZ w/HT))
> >
> > The control box was a 2.53GHZ (533MHZ BUS w/NO HT) = ~5.3seconds
> >
> > I have not tested 2.6.6 recently, but in one of my tests I believe it
> > worked OK, ever since 2.6.6 removed my /etc/lilo.conf and /etc/mtab
>
> and
>
> > several other files, I do not wish to touch that kernel with a 10 foot
>
> poll
>
> > :-P due to the IDE disk flush/cache issue.
>
> I told you this already: 2.6.6 IDE changes don't cause data corruption
> - but fixes some instead (that's why there were merged so quickly!)
> so stop spreading FUD and see
> http://bugme.osdl.org/show_bug.cgi?id=2672.

Re: Linux 2.6.6 appears to be 3 to 4 times slower than 2.6.5.

[Justin Piszcz]

Sorry to all, it turns out (in two separate cases I had two different 
problems that affected the results).

Case 1: No SMP turned on for CPU w/HT after fix (~4.78 seconds compile time 
(2.6GHZ w/HT))
Case 2: Box had 4GB of NON-ECC memory in it, only recognized 2.56GB, took 
out (2) 1GB DDR DIMM's, and the speed returned what it should be. (~4.3 
seconds compile time (3.0GHZ w/HT))

The control box was a 2.53GHZ (533MHZ BUS w/NO HT) = ~5.3seconds

I have not tested 2.6.6 recently, but in one of my tests I believe it worked 
OK, ever since 2.6.6 removed my /etc/lilo.conf and /etc/mtab and several 
other files, I do not wish to touch that kernel with a 10 foot poll :-P due 
to the IDE disk flush/cache issue.

>From: Duncan Sands <baldrick@free.fr>
>To: gene.heskett@verizon.net, "Justin Piszcz" <jpiszcz@hotmail.com>
>CC: linux-kernel@vger.kernel.org
>Subject: Re: Linux 2.6.6 appears to be 3 to 4 times slower than 2.6.5.
>Date: Mon, 17 May 2004 10:07:17 +0200
>
> > I noted that my epson C82 usb printer was running about 25% of its
> > normal speed last night, it took gimp-print several hours to do half
> > a dozen 8x10's in 720dpi.  gkrellm's display looks normal though,
> > with setiathome currently taking 90+% of the cpu, which is normal.
>
>2.6.6 or 2.6.6 plus patches from BK or -mm?
>
>Thanks,
>
>Duncan.

_________________________________________________________________
Best Restaurant Giveaway Ever! Vote for your favorites for a chance to win 
$1 million! http://local.msn.com/special/giveaway.asp

Re: Linux 2.6.6 appears to be 3 to 4 times slower than 2.6.5.

[Bartlomiej Zolnierkiewicz]

On Monday 17 of May 2004 17:06, Justin Piszcz wrote:
> Sorry to all, it turns out (in two separate cases I had two different
> problems that affected the results).
>
> Case 1: No SMP turned on for CPU w/HT after fix (~4.78 seconds compile time
> (2.6GHZ w/HT))
> Case 2: Box had 4GB of NON-ECC memory in it, only recognized 2.56GB, took
> out (2) 1GB DDR DIMM's, and the speed returned what it should be. (~4.3
> seconds compile time (3.0GHZ w/HT))
>
> The control box was a 2.53GHZ (533MHZ BUS w/NO HT) = ~5.3seconds
>
> I have not tested 2.6.6 recently, but in one of my tests I believe it
> worked OK, ever since 2.6.6 removed my /etc/lilo.conf and /etc/mtab and
> several other files, I do not wish to touch that kernel with a 10 foot poll
> :-P due to the IDE disk flush/cache issue.

I told you this already: 2.6.6 IDE changes don't cause data corruption
- but fixes some instead (that's why there were merged so quickly!)
so stop spreading FUD and see http://bugme.osdl.org/show_bug.cgi?id=2672.

Re: Linux 2.6.6 appears to be 3 to 4 times slower than 2.6.5

[Alastair Stevens]

Another "me too" I'm afraid.  I'm running 2.6.6 on an ordinary UP Athlon 
system under Gentoo, and it behaves very strangely.  About 3 out of 4 times, 
it hangs during boot-up, at the "Freeing unused kernel memory" stage, and 
seems to not get as far as calling init....

Yesterday, it got past that stage, and the various startup scripts began to 
run, but *achingly* slowly.  It took about 10 minutes to get most of the 
network services started, with no login prompt in sight - I got fed up and 
rebooted then, because I needed to get some work done!

I rebooted with 2.6.6-rc3, which has always been absolutely fine.

Let me know if any other info would be useful....

Cheers
Alastair

-- 
  \\ ........................................   www.mrc-bsu.cam.ac.uk
   \\  Alastair Stevens, SysAdmin Team       \       01223 330383
   //  MRC Biostatistics Unit, Cambridge UK   \.......................
  --

Re: Linux 2.6.6 appears to be 3 to 4 times slower than 2.6.5

[Andrew Morton]

Alastair Stevens <alastair.stevens@mrc-bsu.cam.ac.uk> wrote:
>
> Another "me too" I'm afraid.  I'm running 2.6.6 on an ordinary UP Athlon 
> system under Gentoo, and it behaves very strangely.  About 3 out of 4 times, 
> it hangs during boot-up, at the "Freeing unused kernel memory" stage, and 
> seems to not get as far as calling init....
> 
> Yesterday, it got past that stage, and the various startup scripts began to 
> run, but *achingly* slowly.  It took about 10 minutes to get most of the 
> network services started, with no login prompt in sight - I got fed up and 
> rebooted then, because I needed to get some work done!

Please try again and if it gets all the way to a login prompt, compare the
contents of /proc/mtrr with 2.6.6-rc3.  Also play around with `top', etc. 
See if there's anything unusual happening.

Re: Linux 2.6.6 appears to be 3 to 4 times slower than 2.6.5

[Alastair Stevens]

 >   Alastair,
 >
 >   nForce2 chipset by any chance? There were changes for the C1 disconnect
 >   issue in 2.6.6.
 >
 >   --
 >   Daniel J Blueman

Nope - Via KT400, so that ought to be fine.  I'll try to get it going again 
tonight and do some more testing if I get a chance....

-- 
  \\ ........................................   www.mrc-bsu.cam.ac.uk
   \\  Alastair Stevens, SysAdmin Team       \       01223 330383
   //  MRC Biostatistics Unit, Cambridge UK   \.......................
  --

Linux 2.6.6 appears to be 3 to 4 times slower than 2.6.5.

[Justin Piszcz]

I have memory benchmarks and compile benchmarks,

essentially

compiling takes ~5 sec on a 2.53ghz (533mhz bus/2GB ram box) for lilgp - 
genetic program
ram  =ddr 333
compiling takes ~15-20 sec on a 3ghz (not sure on bus/4gb ram box) for lilgp 
- genetic program
ram = ddr 400

dd if=/dev/zero of=/mnt/ramdisk/file bs=409e size=(64 or 256mb)
it is 12% faster on the slower (2.53ghz box) vs the box w/DDR 400mhz ram

Is anyone else having SERIOUS PROBLEMS with the 2.6.6 kernel as well?

_________________________________________________________________
Watch LIVE baseball games on your computer with MLB.TV, included with MSN 
Premium! http://join.msn.click-url.com/go/onm00200439ave/direct/01/

Re: Linux 2.6.6 appears to be 3 to 4 times slower than 2.6.5.

[Gene Heskett]

On Saturday 15 May 2004 08:04, Justin Piszcz wrote:
>I have memory benchmarks and compile benchmarks,
>
>essentially
>
>compiling takes ~5 sec on a 2.53ghz (533mhz bus/2GB ram box) for
> lilgp - genetic program
>ram  =ddr 333
>compiling takes ~15-20 sec on a 3ghz (not sure on bus/4gb ram box)
> for lilgp - genetic program
>ram = ddr 400
>
>dd if=/dev/zero of=/mnt/ramdisk/file bs=409e size=(64 or 256mb)
>it is 12% faster on the slower (2.53ghz box) vs the box w/DDR 400mhz
> ram
>
>Is anyone else having SERIOUS PROBLEMS with the 2.6.6 kernel as
> well?

I noted that my epson C82 usb printer was running about 25% of its 
normal speed last night, it took gimp-print several hours to do half 
a dozen 8x10's in 720dpi.  gkrellm's display looks normal though, 
with setiathome currently taking 90+% of the cpu, which is normal.

-- 
Cheers, Gene
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
99.22% setiathome rank, not too shabby for a WV hillbilly
Yahoo.com attorneys please note, additions to this message
by Gene Heskett are:
Copyright 2004 by Maurice Eugene Heskett, all rights reserved.

Re: Linux 2.6.6 appears to be 3 to 4 times slower than 2.6.5.

[Duncan Sands]

> I noted that my epson C82 usb printer was running about 25% of its
> normal speed last night, it took gimp-print several hours to do half
> a dozen 8x10's in 720dpi.  gkrellm's display looks normal though,
> with setiathome currently taking 90+% of the cpu, which is normal.

2.6.6 or 2.6.6 plus patches from BK or -mm?

Thanks,

Duncan.