[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: SMP Kernel on UP Machine



Michael Madden <madden@maddentechnologies.com> [2002-08-19 21:32:00 -0500]:
> Just curious... is there any danger or benefit in running a SMP
> kernel on a single processor machine?  I am considering installing
> a SMP kernel on my UP Debian 3.0 machine.

It should work fine.  But it will be slightly slower.  Depending upon
the compilation #ifdef's in the code will remove the semaphore
operations which are needed in a SMP kernel but not in a UP kernel.
Most people want the kernel that is as fast as possible and so
targeted kernels are made available.

If you just want a uniform system installation across several machines
and that is more important than squeezing the ultimate in performance
out of each machine then you are good to go to run a SMP kernel on a
UP machine.  I don't think the performance difference is more than 5%
between the two kernels.  It may be less.  I have not measured it
recently enough to remember.  It is a small but measurable amount.

Yes Virginia MP machines are slower than UP machines, clock cycle for
clock cycle, both in the kernel code and in the CPU hardware itself.
But with a MP machine you can do more things at once and overall you
are ahead.

Bob

Attachment: pgpVvoMdkk6a9.pgp
Description: PGP signature


Reply to: