This thread has been locked.
If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.
Hi TI
The customer ran the cyclictest program to check the performance of the RT Linux SDK,
A 1 ms control cycle is required to control the robot, and the stress test was conducted with Core3 assigned as a separate process.
If there is no load, there is no problem even if it is operated for more than 8 hours,
If you apply stress, the delay time is longer in about 20 minutes, causing problems.
I would like to ask if there is a separate test method or a solution to the delay time so that the tasks assigned to CORE3 do not cause delay time.
I am attaching their test procedures and results.
Please give me some advice.
Thanks.
Regards,
Jack
Hi Praneeth
Can you please check if the Core3' latency could be reduced less than 20us even under heavy stress conditions?
Customer need the improvement of realtime scheduling of Core 3 even under stress load.
Thanks.
Regards,
Jack
Hi Jack,
In the release notes is the best latency measured so far. 20uS is not realistic on this processor.
Best Regards,
Schuyler
20us is not realistic for A53.
Your results are 8.6. With 9.0 SDK this is what we get are in https://software-dl.ti.com/processor-sdk-linux-rt/esd/AM62X/09_00_00_03/exports/docs/devices/AM62X/linux/RT_Linux_Performance_Guide.html#stress-ng-and-cyclic-test . Here is some further runs:
root@am62xx-evm:~# stress-ng -c4 --cpu-method=all &
root@am62xx-evm:~# cyclictest -m -Sp98 -i1000 -h800 -l100000000 -q > output
# Max Latencies: 00104 00108 00128 00103
Further improvement with kernel option CONFIG_ARM_ARCH_TIMER_EVTSTREAM=n
root@am62xx-evm:~# stress-ng -c4 --cpu-method=all &
root@am62xx-evm:~# cyclictest -m -Sp98 -i200 -h800 -l100000000 -q > output