interpreting PM performance log data?
Andrew Chapman
Randy Mackay
AC,
The “NLon” column holds the number of slow loops (i.e. loops that ran at least 5% slower than they should which is 10500 microseconds) and the “NLoop” col holds the total number of loops since the last PM message. Generally we print the message every 10seconds so we get 1000 loops when running at 100hz but I expect on the Pixhawk that number will eventually be higher. In any case, the percentage of slow loops should generally stay below 5%. So 50 slow loops or so ok. Above that means something is probably going wrong.
The “MaxT” is the maximum time in microseconds (i.e. 1 millionth of a second) that the mainloop took. I.e. the slowest loop over the past 10seconds took X microseconds. With our 100hz main loop rate it should ideally take 10,000 microseconds. If we saw numbers over 13000 I’d get worried. The only exception is when we arm (EV 10 or 11). In these cases we expect a very slow main loop of over 1 second. So you could probably throw away numbers that are over 200,000 (i.e 0.2 seconds).
-Randy
--
You received this message because you are subscribed to the Google Groups "drones-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to drones-discus...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
Andrew Chapman
Randy Mackay
Ok, sounds good. THR_MID should can be between 300 ~ 700 I think.
-Randy
--
Robert Lefebvre
Andrew Chapman
Below are the results for all the logs you sent me. 6 fail, 4 warnings, 3 pass. Is that what you're expecting?
At the moment it considers it a FAIL if any PM line has >10% slow loops or there are more than 5 PM lines > 6%. And it is a warning if any single PM line shows >6% slow loops.
Most other logs I have pass using these thresholds, and they're easy to tweak of course - do you think I should loosen it up a little so the ones here that trigger the warning pass? Maybe ignore a single instance if <10%?
Dataflash log analysis report for file: logs/robert_pm/AH_Test_10_Quad_Thr1000_No_Mount_w_Logs.log
Performance: PASS
Dataflash log analysis report for file: logs/robert_pm/AH_Test_11_XOcto_Thr1000_No_Mount_w_Logs.log
Performance: FAIL 8 slow loop lines found, max 17.74% on line 4531 [GRAPH]
Dataflash log analysis report for file: logs/robert_pm/AH_Test_12_XOcto_no_Mount_Simple.log
Performance: FAIL 12 slow loop lines found, max 18.74% on line 3770 [GRAPH]
Dataflash log analysis report for file: logs/robert_pm/AH_Test_13_XOcto_w_Mount_Simple.log
Performance: FAIL 12 slow loop lines found, max 15.82% on line 4598 [GRAPH]
Dataflash log analysis report for file: logs/robert_pm/AH_Test_1_Quad_No_Mount.log
Performance: WARN 1 slow loop lines found, max 7.30% on line 4139 [GRAPH]
Dataflash log analysis report for file: logs/robert_pm/AH_Test_2_Quad_w_Mount.log
Performance: WARN 1 slow loop lines found, max 6.90% on line 4947 [GRAPH]
Dataflash log analysis report for file: logs/robert_pm/AH_Test_3_Quad_G_Dt_w_Mount.log
Performance: WARN 1 slow loop lines found, max 8.60% on line 6505 [GRAPH]
Dataflash log analysis report for file: logs/robert_pm/AH_Test_4_Quad_ThrMax_1000.log
Performance: WARN 1 slow loop lines found, max 7.68% on line 2093 [GRAPH]
Dataflash log analysis report for file: logs/robert_pm/AH_Test_5_XOcto_Timer_w_Mount.log
Performance: FAIL 14 slow loop lines found, max 18.56% on line 2983 [GRAPH]
Dataflash log analysis report for file: logs/robert_pm/AH_Test_6_XOcto_02_w_Mount.log
Performance: FAIL 9 slow loop lines found, max 18.50% on line 2424 [GRAPH]
Dataflash log analysis report for file: logs/robert_pm/AH_Test_7_XOcto_02_No_Mount.log
Performance: FAIL 8 slow loop lines found, max 13.79% on line 1538 [GRAPH]
Dataflash log analysis report for file: logs/robert_pm/AH_Test_8_Quad_02_Thr1000_No_Mount.log
Performance: PASS
Dataflash log analysis report for file: logs/robert_pm/AH_Test_9_Quad_02_Thr1000_w_Mount.log
Performance: PASS
Robert Lefebvre
--