845GE Chipset severe performance problems

845GE Chipset severe performance problems

[Craig Robinson]

Hello there, I would appreciate any help any of you guys can give..

Having built 3 identical machines based on:

MSI 845GE Max-L mainboard
P4 (533) 2.4GHz CPU
512Mb PC2700 DDR
Maxtor DiamondMax9 80GB

... we're experiencing very strange and severe performance problems with the systems.

They all have RH8 installed, we have tried both the default kernel and the latest 2.4.20. The problem manifests itself in overall slow system performance. I'll give an example below:

When running 'make menuconfig', this normally takes about a second to make and load, however on these machines it takes about 20-30 seconds! We've got a Celeron which does it faster.

A bit of background on what we've tried:

-HD's are running 32bit I/O and UDMA, infact hdparm tests report some very nice figures.
-There is no excessive HD activity, infact HD activity seems normal.
-Onboard Sound and all USB is disabled, and on the 2.4.20 we're using, these have been removed from the kernel.
-We've tried various combinations of just about every BIOS setting which could effect this, so we're resigned to thinking this is some sort of OS/Kernel/Support issue.

The best way to describe the problem is; it's like trying to use the system while it's under excessive load. Obviously there is no load on the system however, the problem occurs when using the system from idle.

We have tried a number of different kernels and also tried fresh
installs of Redhat8 and Redhat 7.3, both make the same problems.

Kernels tried:
2.5.66
2.4.21-grsec-xfs
2.4.20
 
All bring up the same results.

We have run tests on the memory and cpu and each seem to
running fine (CPU is very quick actually..)

If anyone needs any more info feel free to ask, we're beginning to tear our hair out on this one, it's very frustrating... ANY help at all is really appreciated.

Here is some info:


cat /proc/cpuinfo
processor       : 0
vendor_id       : GenuineIntel
cpu family      : 15
model           : 2
model name      : Intel(R) Pentium(R) 4 CPU 2.40GHz
stepping        : 7
cpu MHz         : 2405.506
cache size      : 512 KB
fdiv_bug        : no
hlt_bug         : no
f00f_bug        : no
coma_bug        : no
fpu             : yes
fpu_exception   : yes
cpuid level     : 2
wp              : yes
flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm
bogomips        : 4784.26

cat /proc/meminfo
        total:    used:    free:  shared: buffers:  cached:
Mem:  525225984 215515136 309710848        0 93630464 74813440
Swap: 1069244416        0 1069244416
MemTotal:       512916 kB
MemFree:        302452 kB
MemShared:           0 kB
Buffers:         91436 kB
Cached:          73060 kB
SwapCached:          0 kB
Active:         190012 kB
Inact_dirty:      2200 kB
Inact_clean:      4036 kB
Inact_target:    39248 kB
HighTotal:           0 kB
HighFree:            0 kB
LowTotal:       512916 kB
LowFree:        302452 kB
SwapTotal:     1044184 kB
SwapFree:      1044184 kB
Committed_AS:    63376 kB

fdisk -l

Disk /dev/hda: 255 heads, 63 sectors, 9964 cylinders
Units = cylinders of 16065 * 512 bytes

   Device Boot    Start       End    Blocks   Id  System
/dev/hda1   *         1        13    104391   83  Linux
/dev/hda2            14      1543  12289725   83  Linux
/dev/hda3          1544      2563   8193150   83  Linux
/dev/hda4          2564      9964  59448532+   f  Win95 Ext'd (LBA)
/dev/hda5          2564      2690   1020096   83  Linux
/dev/hda6          2691      2820   1044193+  82  Linux swap
/dev/hda7          2821      9964  57384148+  83  Linux

Re: 845GE Chipset severe performance problems

[Alan Cox]

Check the mtrr ranges. The odd junk bios gets this wrong and leaves
you with parts of main memory uncached. The symptoms you see might
be from that if they are all identical boxes.

Alan

Re[2]: 845GE Chipset severe performance problems

[Craig Robinson]

Hello Alan,


Many thanks for the reply..

cat /proc/mtrr shows things, but not sure what it SHOULD be...(output
below)

Running 2.5.66 now seems to have speeded up things considerably,
however, I still consider this slow for a 2.4Ghz machine (we have some
2Ghz under alot of load that will still compile things quicker..)

So any suggestions and of course the help you have already provided
are appreciated.

Craig

--

 [/usr/src]# cat /proc/mtrr
reg00: base=0x00000000 (   0MB), size= 256MB: write-back, count=1
reg01: base=0x10000000 ( 256MB), size= 128MB: write-back, count=1
reg02: base=0x18000000 ( 384MB), size=  64MB: write-back, count=1
reg03: base=0x1c000000 ( 448MB), size=  32MB: write-back, count=1
reg04: base=0x1e000000 ( 480MB), size=  16MB: write-back, count=1
reg05: base=0x1f000000 ( 496MB), size=   8MB: write-back, count=1
reg06: base=0xe0000000 (3584MB), size=  64MB: write-combining, count=1

-
Saturday, March 29, 2003, 2:45:48 PM, you wrote:

AC> Check the mtrr ranges. The odd junk bios gets this wrong and leaves
AC> you with parts of main memory uncached. The symptoms you see might
AC> be from that if they are all identical boxes.

AC> Alan



-- 
Best regards,
 Craig                            mailto:craig.robinson@btinternet.com

Re: Re[2]: 845GE Chipset severe performance problems

[Alan Cox]

On Sat, 2003-03-29 at 17:04, Craig Robinson wrote:
>  [/usr/src]# cat /proc/mtrr
> reg00: base=0x00000000 (   0MB), size= 256MB: write-back, count=1
> reg01: base=0x10000000 ( 256MB), size= 128MB: write-back, count=1
> reg02: base=0x18000000 ( 384MB), size=  64MB: write-back, count=1
> reg03: base=0x1c000000 ( 448MB), size=  32MB: write-back, count=1
> reg04: base=0x1e000000 ( 480MB), size=  16MB: write-back, count=1
> reg05: base=0x1f000000 ( 496MB), size=   8MB: write-back, count=1

Looks right for a machine with 504Mb of RAM. How does your box run
if you boot with mem=504M ?

Re: Re[2]: 845GE Chipset severe performance problems

[Rogier Wolff]

On Sun, Mar 30, 2003 at 04:59:46PM +0100, Alan Cox wrote:
> On Sat, 2003-03-29 at 17:04, Craig Robinson wrote:
> >  [/usr/src]# cat /proc/mtrr
> > reg00: base=0x00000000 (   0MB), size= 256MB: write-back, count=1
> > reg01: base=0x10000000 ( 256MB), size= 128MB: write-back, count=1
> > reg02: base=0x18000000 ( 384MB), size=  64MB: write-back, count=1
> > reg03: base=0x1c000000 ( 448MB), size=  32MB: write-back, count=1
> > reg04: base=0x1e000000 ( 480MB), size=  16MB: write-back, count=1
> > reg05: base=0x1f000000 ( 496MB), size=   8MB: write-back, count=1
> 
> Looks right for a machine with 504Mb of RAM. How does your box run
> if you boot with mem=504M ?

504Mb RAM? That sounds like a 512Mb ram with 8Mb "shared"
video-ram. 845? Sounds like one of those "everything integrated"
chipsets. performance problems? Do you happen to have the video in a
high-bandwidth mode?????

			Roger. 

-- 
** R.E.Wolff@BitWizard.nl ** http://www.BitWizard.nl/ ** +31-15-2600998 **
*-- BitWizard writes Linux device drivers for any device you may have! --*
* The Worlds Ecosystem is a stable system. Stable systems may experience *
* excursions from the stable situation. We are currently in such an      * 
* excursion: The stable situation does not include humans. ***************

Re: Re[2]: 845GE Chipset severe performance problems

[H. Peter Anvin]

Followup to:  <20030401204850.A6890@bitwizard.nl>
By author:    Rogier Wolff <R.E.Wolff@BitWizard.nl>
In newsgroup: linux.dev.kernel
> 
> 504Mb RAM? That sounds like a 512Mb ram with 8Mb "shared"
> video-ram. 845? Sounds like one of those "everything integrated"
> chipsets. performance problems? Do you happen to have the video in a
> high-bandwidth mode?????
> 

Sure enough it is, and I can personally vouch for the fact that my
845GE system (a Shuttle-X box) got a lot nicer after I dropped a
Radeon 9000 into the AGP slot :)

	-hpa
-- 
<hpa@transmeta.com> at work, <hpa@zytor.com> in private!
"Unix gives you enough rope to shoot yourself in the foot."
Architectures needed: ia64 m68k mips64 ppc ppc64 s390 s390x sh v850 x86-64

Re: 845GE Chipset severe performance problems

[Dave Jones]

On Tue, Apr 01, 2003 at 02:44:04PM +0200, Eric Brunet wrote:

 > As there is this thread about mtrr on Intel chipsets, I have some
 > messages in the log about mtrr, and I don't know whether they are
 > harmless warnings or errors that should be reported.
 > 
 > My computer is a 2.4 GhZ Pentium IV with an intel i845G/GL chipset.
 > Motherboard and bios by shuttle.
 > 
 > $ cat /proc/mtrr
 > reg00: base=0x00000000 (   0MB), size= 512MB: write-back, count=1
 > reg01: base=0x1f800000 ( 504MB), size=   8MB: uncachable, count=1
 > reg02: base=0xe0000000 (3584MB), size= 128MB: write-combining, count=2
 > 
 > I have 512 MB of memory, the motherboard doesn't support more than 2GB
 > and I don't see what is this range over 3.5 GB. Also, the two first
 > overlaping ranges look suspicious.

overlapping is allowed. Does this board have onboard graphics ?
It looks like your top 8MB is shared video memory. (Ie, the graphics
chip doesn't have its own RAM). Usually theres things in the BIOS
to adjust the amount of RAM to allocate to the 'card' including
a disable option if you have a real card inserted instead.

 > So... Is this situation normal ?

Could be. If the answer to the above question is yes, then its normal.
What exactly are you finding slow ?

		Dave

Re: 845GE Chipset severe performance problems

[Eric Brunet]

In-Reply-To: <1049039985.14686.11.camel@dhcp22.swansea.linux.org.uk>
References: <188481168784.20030329130300@btinternet.com> <1048949147.6725.3.camel@dhcp22.swansea.linux.org.uk> <153495685337.20030329170457@btinternet.com> <1049039985.14686.11.camel@dhcp22.swansea.linux.org.uk>


As there is this thread about mtrr on Intel chipsets, I have some
messages in the log about mtrr, and I don't know whether they are
harmless warnings or errors that should be reported.

My computer is a 2.4 GhZ Pentium IV with an intel i845G/GL chipset.
Motherboard and bios by shuttle.

$ cat /proc/mtrr
reg00: base=0x00000000 (   0MB), size= 512MB: write-back, count=1
reg01: base=0x1f800000 ( 504MB), size=   8MB: uncachable, count=1
reg02: base=0xe0000000 (3584MB), size= 128MB: write-combining, count=2

I have 512 MB of memory, the motherboard doesn't support more than 2GB
and I don't see what is this range over 3.5 GB. Also, the two first
overlaping ranges look suspicious.

I have the following error/warning messages in the log:

* with kernel 2.4.21pre3 (+ acpi patches)
        mtrr: base(0xe0020000) is not aligned on a size(0x800000)
boundary
  (this happens just after initializing drm)

* with kernel 2.5.63 (+ acpi patches)
        mtrr: MTRR 2 not used
  (this happens once or twice during each sutdown. Note that I don't have
  drm with kernel 2.5.x, so I don't know if the first message is really
gone)

So... Is this situation normal ?

Éric Brunet