Re: Slow lmbench on IBM Netfinity

Paul Ho (pH7@w3com.com)
Thu, 9 Dec 1999 15:16:27 -0500

At 07:52 PM +0100 12/09/99, Ragnar Kjørstad wrote:
>On Thu, Dec 09, 1999 at 01:43:19PM -0500, Paul Ho wrote:
>> Thanks for the reply. /proc/cpuinfo show 2 cpu with reasonable BogoMIPS.
>> However, 2nd cpu show up as processor 2. i.e. I have processor 0 and 2.
>> Is this normal?
>
>I'm not sure. our SMB machine show up with processor 0 and 1...
>
>Even if the mips indicator was normal, it could still be mtrr-releated
>(I guess).
>
>>From Configure.help:
>
> Saying Y here also fixes a problem with buggy SMP BIOSes which only
> set the MTRRs for the boot CPU and not the secondary CPUs. This can
> lead to all sorts of problems.
>
>
>I'd suggest you try with MTTR enabled - if it doesn't help you should
>reply to your own message saying your processors are numbered 0 and 2.
>Probably someone else can tell you if it's ok or not (I'm not a hardcore
>kernel-hacker).

MTRR is on.
/proc/mtrr show:
reg00: base=0x00000000 ( 0MB), size= 256MB: write-through, count=1

is this looks normal?

I suspect memory problem because of the high Mmap Latency, Prot Fault and
Page Fault in lmbench:

>File & VM system latencies in microseconds - smaller is better
>--------------------------------------------------------------
>Host OS 0K File 10K File Mmap Prot Page
>
> Create Delete Create Delete Latency Fault Fault
>--------- ------------- ------ ------ ------ ------ ------- ----- -----
>i686-linu Linux 2.2.13- 148 29 320 42 191531 20 7.4K

Paul

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