[Warning: the hack presented in this message is *really* sick]
>Any advice on where to look for info on this type of problem is
>appreciated,
If I understand you correctly, you want to make sure that your machine is
fully loaded all of the time, without going into an unbounded load
explosion, right ? Last time I had a similar problem (large simulation, >2
CPU-years on 4 quad PPROs) I devised the following ugly hack:
-- begin script
#!/bin/sh
# Get some initial parameters
NUM_CPUS=`grep -i processor /proc/cpuinfo | wc -l | sed 's/ *//'`
# Do the loops
while DoMoreSims; do
# Only schedule a new sim when the load is less than the number of CPUs
until [ `sed 's/\..*//' < /proc/loadavg` -lt $NUM_CPUS ]; do
sleep 15
done
StartNewSim &
# Sleep for a bit to let the load averages settle
sleep 25
done
-- end script
As you see, new simulators are only started when the load is lower than the
number of CPUs. A cleaner option is probably to just start one simulator
per CPU and then wait() for them to finish before starting new ones. The
reason I didn't use this method is that a) part of my simulation was not
CPU-bound, so the load method got higher throughput and b) it was possible
to run multiple starter scripts on the same machine (for different sims)
which is verfy hard to cross-balance with the wait()-method.
HTH,
Jan-Derk Bakker.
--
Jan-Derk Bakker, [EMAIL PROTECTED]
The lazy man's proverb:
'There's no business like slow business !'
-
Linux SMP list: FIRST see FAQ at http://www.irisa.fr/prive/mentre/smp-faq/
To Unsubscribe: send "unsubscribe linux-smp" to [EMAIL PROTECTED]
