Here are timings of Berkeley Spice 3, linked both
with and without Electric Fence, run under 2.1.117
(w/o SMP) vs 2.0.34.
Elapsed time in seconds 2.0.34 2.1.117
----------------------- ------ -------
Without Electric Fence 3.82 3.72
With Electric Fence 14.61 325.85 (!)
No swapping was involved; at least 90-100MB seemed
to remain free on my system at all times. As
indicated by "top", Spice's memory footprint (with
efence) was only 15MB. My system has 128MB RAM
(Intel P-II-266 w/ RedHat 5.1). Each measurement
was made with GNU "time" (subjective impressions
were similar) immediately after a fresh reboot.
Spice3 was run on a VT without starting X. It
was given a moderately complex input file to work
on. Spice3 gets many memory allocations in small
pieces, and frees a few of them as it goes along.
Anyone should be able to duplicate this result with
spice-3f4-4.src.rpm from redhat contrib/hurricane,
linking it with libefence, and giving it ltra_1.cir
(from the spice-examples binary rpm that the src
rpm also builds) as input with the following command:
echo "set noquitsave; run; print all; quit" |
time ./spice3 -i ltra_1.cir
Spice emits some complaint messages as it runs,
but the overall slowdown is quite clear, and occurs
with other circuit files also.
Before making this posting (my first), I searched
this archive for mentions of Electric Fence.
Seemingly they are rare and have not addressed
this subject. Unfortunately I am not acquainted
with the kernel internals.
Thanks for any help, and thanks to members of this
list for your wonderful work in creating Linux.
Richard D. McRoberts
Loveland, Colorado USA
rdm@verinet.com
-
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.altern.org/andrebalsa/doc/lkml-faq.html