On 6 October 2016 at 22:35, Steve Muckle steve.muckle@linaro.org wrote:
Apply various clarifications and fixups to the README for the cpufreq governor efficiency test.
Acked-by: Vincent Guittot vincent.guittot@linaro.org
Signed-off-by: Steve Muckle smuckle@linaro.org
doc/examples/cpufreq_governor_efficiency/README | 48 ++++++++++++------------- 1 file changed, 23 insertions(+), 25 deletions(-)
diff --git a/doc/examples/cpufreq_governor_efficiency/README b/doc/examples/cpufreq_governor_efficiency/README index 10482b85f529..0d9123daa1cd 100755 --- a/doc/examples/cpufreq_governor_efficiency/README +++ b/doc/examples/cpufreq_governor_efficiency/README @@ -1,23 +1,22 @@ Measure the efficiency of cpufreq governors using rt-app
BACKGROUND:
- DVFS adds a latency in the execution of task because of the time to
- decide to move at max freq. We need to measure this latency and check
- that the governor stays in an acceptable range.
- DVFS adds a latency in the execution of a task because some or all of the
- task may not be executed at maximum frequency. We need to measure this
- latency and check that the governor stays in an acceptable range.
- When rt-app runs a json file, a log file is created for each thread.
- This log file records the number of loop that has been executed and
- When rt-app runs a json file a log file is created for each thread.
- This log file records the number of loops that have been executed and the duration for executing these loops (per phase). We can use these
- figures to evaluate to latency that is added by a cpufreq governor
- figures to evaluate the latency that is added by a cpufreq governor and its "performance efficiency".
- We use the run+sleep pattern to do the measurement, for the run time per
- loop, the performance governor should run the expected duration as the
- CPU stays a max freq. At the opposite, the powersave governor will give
- use the longest duration (as it stays at lowest OPP). Other governor will
- be somewhere between the 2 previous duration as they will use several OPP
- and will go back to max frequency after a defined duration which depends
- on its monitoring period.
We use the run+sleep pattern to do the measurement. For the run time per
loop the performance governor should run the expected duration as the CPU
stays at max freq. At the opposite, the powersave governor will give us the
longest duration (as it stays at lowest OPP). Other governors will be
somewhere between the 2 previous durations as they may use any number of
OPPs for various portions of the workload.
The formula:
@@ -34,20 +33,22 @@ Usage: ./calibration.sh <cpu> cpu: cpu number on which you want to run the test
- ./test.sh <governor> <cpu> <runtime> <sleeptime> [<loops>]
- ./test.sh <governor> <cpu> <runtime> <period> [<loops>] governor: target CPUFreq governor you want to test
- cpu: cpu number on which you want to run the test. Be the same
as the one passing to "calibration.sh".- runtime: running time in ms per loop of the workload pattern
- sleeptime: sleeping time in ms per loop of the workload pattern
- cpu: cpu number on which you want to run the test. It should
be the same as the one passed to "calibration.sh".- runtime: running time in ms per loop of the workload pattern,
at fmax- period: period in ms of the loop of the workload pattern loops: repeat times of the workload pattern. default: 10
Example:
- "./calibration.sh 0" means to calculate the computing capacity of CPU0 which
- will be used in the following test.
- "./calibration.sh 0" means to calculate the computing capacity of CPU0
- which will be used in the following test.
- "./test.sh ondemand 0 100 100 20" means to
- test "ondemand" on CPU0 with workload pattern "run 100ms + sleep 100ms"(20 loops).
- "./test.sh ondemand 0 100 200 20" means to test "ondemand" on CPU0 with a
- periodic workload where the busy portion is 100ms of work (if it were to be
- run at fmax), the period is 200ms, and the workload runs for 20 cycles.
NOTE: - Make sure there are "sed", "cut", "grep", "rt-app", etc tools on @@ -55,6 +56,3 @@ NOTE:
- Run the test while the system is idle.
- You can change the target governor's parameters after running "calibration.sh",
but before "test.sh".-- 2.7.3