Re: Weird bogomips problem

lilo (lilo@lions.reseq.ungov)
Tue, 12 Mar 1996 22:40:22 -0600 (CST)


> On Sun, 10 Mar 1996, David C. Niemi wrote:
>
> > My wife's machine registers only 1/10th as many bogomips as it should
> > while booting. It is a 486 DX2-66 VLB system with 256 KB secondary cache
> > and 8 MB RAM. What is really strange is that the turbo button really *is*
> > on, and other benchmarks look normal on this machine, including dhrystone,
> > floating point, and one that measures bandwidth to main memory. Does
> > anyone have an idea of what could be wrong that would cause it to have
> > 3.08 Bogomips instead of 33?

I can report to whoever made the suggestion, I changed the write caching on
my motherboard from ``writeback'' to ``write-through'', and as a result,
find in /proc/cpuinfo:

BogoMips : 39.93

And the classic and portable BogoMips runs produce, respectively:

Classic Calibrating delay loop.. ok - 31.65 BogoMips

Portable Calibrating delay loop.. ok - 39.70 BogoMips

As you can see, all is well. It'll be interesting to see if this has any
effect whatsover on processing, and it may not---but you'd think that having
any timing loops faster by over a factor of ten (2.66 --> 39.70) would have
somewhat a negative effect on processing.... Hi, ho, if I see anything
interesting I'll let you know.

Thanks again all! :)

:) lilo