After running for a long time, tegrastats shows negative averages

Robotics & Edge Computing Jetson & Embedded Systems Jetson AGX Xavier
1

Hi,

We kept the tegrastats utility running over the weekend, and we found that some of the averages seem to turn negative at some point. Not that I think it’s a huge problem, but still something you might want to address.

I had it running at intervals of 100ms.

RAM 7663/31927MB (lfb 5326x4MB) CPU [15%@2263,0%@2263,100%@2263,0%@2263,100%@2263,0%@2263,0%@2263,0%@2263] EMC_FREQ 0%@2133 GR3D_FREQ 0%@1377 APE 150 MTS fg 0% bg 0% AO@58.5C GPU@59.5C Tdiode@62C PMIC@100C AUX@59.5C CPU@61.5C thermal@60.3C Tboard@54C GPU 1076/-948 CPU 4150/374 SOC 4150/-236 CV 0/0 VDDRQ 768/416 SYS5V 2804/420
RAM 7663/31927MB (lfb 5326x4MB) CPU [0%@2263,0%@2263,100%@2263,0%@2263,100%@2263,0%@2263,0%@2263,8%@2263] EMC_FREQ 0%@2133 GR3D_FREQ 0%@1377 APE 150 MTS fg 0% bg 0% AO@58.5C GPU@59.5C Tdiode@62C PMIC@100C AUX@59.5C CPU@61C thermal@60.3C Tboard@54C GPU 1076/-948 CPU 4150/374 SOC 4150/-236 CV 0/0 VDDRQ 768/416 SYS5V 2804/420
RAM 7663/31927MB (lfb 5326x4MB) CPU [8%@2263,0%@2263,100%@2263,0%@2263,100%@2263,0%@2263,0%@2263,0%@2263] EMC_FREQ 0%@2133 GR3D_FREQ 0%@1377 APE 150 MTS fg 0% bg 0% AO@58.5C GPU@59.5C Tdiode@62C PMIC@100C AUX@59.5C CPU@61.5C thermal@60.3C Tboard@54C GPU 1076/-948 CPU 4150/374 SOC 4150/-236 CV 0/0 VDDRQ 768/416 SYS5V 2804/420
3

hello fransklaver,

may I have more details to reproduce this issue,
for example,

  1. which JetPack release it is? please check release tag for details, i.e. $ cat /etc/nv_tegra_release
  2. are you having any process running in the background, or, you just leaving Xavier idle and keep tegrastats reports the statistics over the weekend.
  3. can these erroneous recover after you restart the tegrastats utility?
4

4.3, but due to development history and circumstances we’re using a kernel from JP 4.4.

No, I had nppi processing on live camera images running during that time. I just now notice that the GR3D_FREQ is at 0% in my example lines, which is odd, as the program was still running at the time I noticed the negative values.

I think restarting tegrastats removes the erroneous values, yes, but I would need to go and check again to be sure.