Benchmarking a 4way

[ober]

I have a 4way xeon with 4 gigs of ram, and 2meg l2 cache on each cpu.
So far I have run a few benchmarks comparing the three primary BSDs.
http://nikto.linbsd.org/obsd nbsd and fbsd
The results are all run together.
I am cleaning them up for proper display in charts.
However what I am looking for is any suggestions for
really testing the MP aspects of the OS, and something that would
really thrash the locking in the kernel.

Thanks in advance.

-Ober

[Gary Thorpe]

Post by ober
I have a 4way xeon with 4 gigs of ram, and 2meg l2 cache on each cpu.
So far I have run a few benchmarks comparing the three primary BSDs.
http://nikto.linbsd.org/obsd nbsd and fbsd
The results are all run together.
I am cleaning them up for proper display in charts.
However what I am looking for is any suggestions for
really testing the MP aspects of the OS, and something that would
really thrash the locking in the kernel.
Thanks in advance.
-Ober

Just a note: only FreeBSD really has locking inside of it's kernel. As
far as I can tell, NetBSD and OpenBSD still rely on masking interrupts
and I think only allow one task/process inside the kernel at a time.
Anyone know if this has evolved further?

__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com