Talk:Performance - Revision history

Kaihsu: /* Scaling */

2008-12-08T15:44:06Z

Scaling

← Older revision		Revision as of 15:44, 8 December 2008
Line 17:		Line 17:
	Manually:		Manually:
	Use updated [[calc_testjobs_linux.sh]]		Use updated [[calc_testjobs_linux.sh]]
			<source lang="bash">
	DATA=scaling.xvg		DATA=scaling.xvg
	MAXSLOTS=8		MAXSLOTS=8
Line 25:		Line 26:
	\| awk '/BENCHMARK/ {print $3, $4}' >> $DATA;		\| awk '/BENCHMARK/ {print $3, $4}' >> $DATA;
	done		done
			</source>
	All the above remarks apply here, too.		All the above remarks apply here, too.

Oliver: /* Scaling */

2008-12-07T15:41:53Z

Scaling

← Older revision		Revision as of 15:41, 7 December 2008
Line 3:		Line 3:

	== Scaling ==		== Scaling ==
			How to run scaling tests/testing methodology:

			Use [[test_scaling.py]] which in turn uses [[calc_testjobs_linux.sh]].
			cd testjobs
			test_scaling.py ''NSLOTS''
			where ''NSLOTS'' is the maximum number of available cpus/cores. Results are the files
			scaling.xvg # numbers
			scaling.png # graphs
			* Only measures ''wall time'', hence must be run on an empty machine.
			* Uses <code>date +%s</code> so accuracy is limited to seconds.
			* No repeat runs; results may vary slightly.

			Manually:
	Use updated [[calc_testjobs_linux.sh]]		Use updated [[calc_testjobs_linux.sh]]
	DATA=scaling.xvg		DATA=scaling.xvg
			MAXSLOTS=8
	echo -e "# scaling for Hippo\n# numthreads walltime/s" > $DATA		echo -e "# scaling for Hippo\n# numthreads walltime/s" > $DATA
	for NSLOTS in `seq 8`; do		for NSLOTS in `seq $MAXSLOTS`; do
	echo "-- NSLOTS = $NSLOTS";		echo "-- NSLOTS = $NSLOTS";
	../../calc_testjobs_linux.sh -n $NSLOTS walp_octane_NPT_sp_MD \		../../calc_testjobs_linux.sh -n $NSLOTS walp_octane_NPT_sp_MD \
	\| awk '/BENCHMARK/ {print $3, $4}' >> $DATA;		\| awk '/BENCHMARK/ {print $3, $4}' >> $DATA;
	done		done
	~~Note:~~		All the above remarks apply here, too.
	* only run this on an empty machine: we are using ''wall time'' for timing

	~~Slightly more sophisticated: use [[test_scaling.py]].~~
	~~cd testjobs~~
	~~test_scaling.py ''NSLOTS''~~
	~~where ''NSLOTS'' is the maximum number of available cpus/cores. Results are~~ the ~~files~~
	~~scaling.xvg # numbers~~
	~~scaling~~.~~png # graphs~~

Oliver: /* Scaling */

2008-12-07T15:23:25Z

Scaling

← Older revision		Revision as of 15:23, 7 December 2008
Line 13:		Line 13:
	Note:		Note:
	* only run this on an empty machine: we are using ''wall time'' for timing		* only run this on an empty machine: we are using ''wall time'' for timing

			Slightly more sophisticated: use [[test_scaling.py]].
			cd testjobs
			test_scaling.py ''NSLOTS''
			where ''NSLOTS'' is the maximum number of available cpus/cores. Results are the files
			scaling.xvg # numbers
			scaling.png # graphs

Oliver: /* Scaling */

2008-12-07T13:41:02Z

Scaling

← Older revision		Revision as of 13:41, 7 December 2008
Line 4:		Line 4:
	== Scaling ==		== Scaling ==
	Use updated [[calc_testjobs_linux.sh]]		Use updated [[calc_testjobs_linux.sh]]
			DATA=scaling.xvg
			echo -e "# scaling for Hippo\n# numthreads walltime/s" > $DATA
	for NSLOTS in `seq 8`; do		for NSLOTS in `seq 8`; do
	echo "-- NSLOTS = $NSLOTS";		echo "-- NSLOTS = $NSLOTS";
	../../calc_testjobs_linux.sh -n $NSLOTS walp_octane_NPT_sp_MD \		../../calc_testjobs_linux.sh -n $NSLOTS walp_octane_NPT_sp_MD \
	\| awk '/BENCHMARK/ {print $3, $4}' >> ~~scaling.dat~~;		\| awk '/BENCHMARK/ {print $3, $4}' >> $DATA;
	done		done
	Note:		Note:
	* only run this on an empty machine: we are using ''wall time'' for timing		* only run this on an empty machine: we are using ''wall time'' for timing

Oliver: benchmarking scaling

2008-12-07T13:38:51Z

benchmarking scaling

← Older revision		Revision as of 13:38, 7 December 2008
Line 1:		Line 1:
	== Integration with tests ==		== Integration with tests ==
	We could write a script that does the benchmark while running the test. It would even be possible to automatically post it (with the user's consent, of course). — [[User:Oliver\|Oli]] 16:03, 14 November 2008 (UTC)		We could write a script that does the benchmark while running the test. It would even be possible to automatically post it (with the user's consent, of course). — [[User:Oliver\|Oli]] 16:03, 14 November 2008 (UTC)

			== Scaling ==
			Use updated [[calc_testjobs_linux.sh]]
			for NSLOTS in `seq 8`; do
			echo "-- NSLOTS = $NSLOTS";
			../../calc_testjobs_linux.sh -n $NSLOTS walp_octane_NPT_sp_MD \
			\| awk '/BENCHMARK/ {print $3, $4}' >> scaling.dat;
			done
			Note:
			* only run this on an empty machine: we are using ''wall time'' for timing

Oliver: automatic benchmark in distribution?

2008-11-14T16:03:23Z

automatic benchmark in distribution?

New page

== Integration with tests ==
We could write a script that does the benchmark while running the test. It would even be possible to automatically post it (with the user's consent, of course). — [[User:Oliver|Oli]] 16:03, 14 November 2008 (UTC)