Strange Slowdown

Strange Slowdown

[Matthias Schniedermeyer]

Hi



I have a Dual PII with 2GB-RAM on a GX-Chipset Mainboard (Tyan Thunder
100)

Until yesterday that machine ran with Kernel 2.2.19. After upgrading to
2.4.(4/9) it got about 20times SLOWER than with 2.2.19.

After i switched "High Memory-Support" to "OFF"(4GB Before) the speed went
to normal, but now less than half RAM is used.

Any suggestions?

(Kernel 2.2.19 ran with "Maximum Physical Memory = 2GB")




Bis denn

-- 
Real Programmers consider "what you see is what you get" to be just as 
bad a concept in Text Editors as it is in women. No, the Real Programmer
wants a "you asked for it, you got it" text editor -- complicated, 
cryptic, powerful, unforgiving, dangerous.

Re: Strange Slowdown

[Alan Cox]

> After i switched "High Memory-Support" to "OFF"(4GB Before) the speed went
> to normal, but now less than half RAM is used.
> Any suggestions?

This sounds like the top of memory is running uncached due to wrong mtrr
settings from the BIOS. Can you post your /proc/mtrr 

Re: Strange Slowdown

[Matthias Schniedermeyer]

> > After i switched "High Memory-Support" to "OFF"(4GB Before) the speed went
> > to normal, but now less than half RAM is used.
> > Any suggestions?
> 
> This sounds like the top of memory is running uncached due to wrong mtrr
> settings from the BIOS. Can you post your /proc/mtrr 

Because of other reasons i'm back to 2.2.19

/proc/mtrr from 2.2.19 show this
-- /proc/mtrr --
reg00: base=0x00000000 (   0MB), size=1024MB: write-back, count=1
reg01: base=0x40000000 (1024MB), size= 512MB: write-back, count=1
reg02: base=0x60000000 (1536MB), size= 256MB: write-back, count=1
reg03: base=0x70000000 (1792MB), size= 128MB: write-back, count=1
reg04: base=0x78000000 (1920MB), size=  64MB: write-back, count=1
reg05: base=0x7c000000 (1984MB), size=  32MB: write-back, count=1
-- end --
(32MB "missing". Seems like Linux uses these "missing" MBs.)

For the 2.4.X-Kernel i had switched off the MTRR-Kernel-Option!

Maybe i should try it another time with MTRR-Support switched on. 
Or i should use "mem=2016M".



Bis denn

-- 
Real Programmers consider "what you see is what you get" to be just as 
bad a concept in Text Editors as it is in women. No, the Real Programmer
wants a "you asked for it, you got it" text editor -- complicated, 
cryptic, powerful, unforgiving, dangerous.

RE: Strange Slowdown

[Marvin Justice]

> > > After i switched "High Memory-Support" to "OFF"(4GB
> Before) the speed went
> > > to normal, but now less than half RAM is used.
> > > Any suggestions?
> >
> > This sounds like the top of memory is running uncached due
> to wrong mtrr
> > settings from the BIOS. Can you post your /proc/mtrr
>
> Because of other reasons i'm back to 2.2.19
>
> /proc/mtrr from 2.2.19 show this
> -- /proc/mtrr --
> reg00: base=0x00000000 (   0MB), size=1024MB: write-back, count=1
> reg01: base=0x40000000 (1024MB), size= 512MB: write-back, count=1
> reg02: base=0x60000000 (1536MB), size= 256MB: write-back, count=1
> reg03: base=0x70000000 (1792MB), size= 128MB: write-back, count=1
> reg04: base=0x78000000 (1920MB), size=  64MB: write-back, count=1
> reg05: base=0x7c000000 (1984MB), size=  32MB: write-back, count=1
> -- end --
> (32MB "missing". Seems like Linux uses these "missing" MBs.)
>
> For the 2.4.X-Kernel i had switched off the MTRR-Kernel-Option!
>
> Maybe i should try it another time with MTRR-Support switched on.
> Or i should use "mem=2016M".
>

I had a similar problem which could be solved by either passing the "mem="
option at boot time as you suggest(in which case you permanently lose the
upper 32MB) or by using the /proc/mtrr interface ( as described in
Documentation/mtrr.txt) to assign the "missing" region. For your system you
might try something like:

echo "base=0x7e000000 size=0x2000000 type=write-back" >| /proc/mtrr

Marvin Justice