Re: BogoMIPS?

linux kernel account (linker@nightshade.z.ml.org)
Wed, 4 Feb 1998 18:46:07 -0500 (EST)


On Wed, 4 Feb 1998, Chuck Carson wrote:

> 1) What is a BogoMIP?

A timing loop. It times the operation of a trivial loop. If all programs
were trivial loops, it would be a good performance mesaurment. However, I
know of no useful programs that are a trivial loop (not even sleep in
Linux :) )..

> 2) I have noticed that at boot time the system will calculate (?) the
> bogomips for each
> cpu add their combined total.
>
> eg:
>
> Total of 2 processors activated (594.74 BogoMIPS)

Yep..

664.x here.. (dual 166 mmxes, tyan IIID)

> After numerous kernel recompilings I have noticed that this number changes
> constantly and in
> many times will be much more or much less than the previous kernel. (I
> believe I have seen
> this as high as 720.xx)

This is caused by alignment changes, make sure you are compiling as
'pentium'.. Also, using something like PGCC affects the alignment.

> 3) /proc/cpuinfo shows twp cpu's listed but the system does not behave as if
> it was scheduling
> both cpu's (when compiling software mainly). Before, under 4.2 I compiled
> the same code on
> the same machine _much_ faster. How can I verify both cpu's are being
> scheduled without
> relying on some text file created at boot up?

Make sure make is started with the -j2 option.
Do a top while make is running.. You should see two tasks using 99% of the
cpu each (rather then 2 using 50% each).

I've never seen the kernel report two cpus and have it not work.

> btw: This is a Tyan board with 2 matched Intel 200mmx chips. Any known
> problems with these boards?

Dont know about your board, but mine (TIIID) is specifically forbidden
from running dual 200mmxes.

The board doesn't report it's vendor ID correctly as per mp spec.

There are alot of people very pissed at tyan right now (like myself)
because they refuse to reply to tech email (regarding the devation from
the MP spec)..

If I had just bought a tyan board, I would send it RIGHT back with a note
saying that it is flawed and does not comply with the MP spec because it
does not provide a correct product or vendor id (which makes it impossible
for the kernel to detect it which is needed for IO-APIC pinout mapping
which is needed for load balenced IRQs which is need by high end SMP
systems)..

> Thanks,
> Chuck