how to do GPU performance monitor in Xavier (Solved)

Autonomous Machines Jetson & Embedded Systems Jetson AGX Xavier
1

cannot find nvidia-msi command, cannot find tegrastats command

2

Hi,

Do you mean nvidia-smi?
nvidia-smi can only be used on the desktop environment.

For Jetson, please use tegrastats instead:

sudo ./tegrastats

Thanks.

3

Hi AastaLLL,
I enter “sudo ./tegrastats” command as you mentioned on Xavier. But I get the result as below. Do I need to install any other library to execute the command?
sudo: ./tegrastats: command not found

Thank you

4

Has this problem been solved?

5

I try the solution

and could use the command(sudo python3 tegrastats2.py --bin=/home/nvidia/tegrastats --output=./a.log --params “–interval 500”) to get the information as below.

But can not get visualization charts by the below command
python3 visualize.py --start=“2018-06-09 02:42:30” --end=“2018-06-09 02:43:15” --input=“/home/find/ddown/a.log” --output=./freq.xls

2019-06-12 08:16:24:
RAM 2210/24746MB (lfb 5352x4MB) CPU [0%,1%,3%,1%,2%,0%,0%,2%] EMC_FREQ @2133 GR3D_FREQ 3%@1185 APE 245 MTS fg 0% bg 0% AO@35.5C GPU@36C tj@49C AUX@36C CPU@36.5C Tdiode@36.25C
2019-06-12 08:16:24:
RAM 2210/24746MB (lfb 5352x4MB) CPU [0%,1%,2%,2%,0%,3%,1%,1%] EMC_FREQ @2133 GR3D_FREQ 0%@1185 APE 245 MTS fg 0% bg 0% AO@35C GPU@36.5C tj@49C AUX@36C CPU@36C Tdiode@36.25C
2019-06-12 08:16:25:
RAM 2210/24746MB (lfb 5352x4MB) CPU [2%,3%,2%,2%,1%,1%,2%,0%] EMC_FREQ @2133 GR3D_FREQ 0%@1185 APE 245 MTS fg 0% bg 1% AO@35.5C GPU@36C tj@49C AUX@36C CPU@36C Tdiode@36C
2019-06-12 08:16:25:
RAM 2210/24746MB (lfb 5352x4MB) CPU [0%,0%,1%,0%,0%,0%,2%,1%] EMC_FREQ @2133 GR3D_FREQ 0%@1185 APE 245 MTS fg 0% bg 0% AO@35.5C GPU@36C tj@49C AUX@36C CPU@36C Tdiode@36C

6

Run tegrastats on Jetson Xavier like you would a normal linux command:

$ tegrastats

Command line options include:

–interval [milliseconds]
–verbose
–stop [other instances of tegrastats]
–logfile

7

The “./” should not be part of that command (unless your working directory is the /usr/bin directory)

jwatte@jetson-nano:~$ which tegrastats
/usr/bin/tegrastats
jwatte@jetson-nano:~$ tegrastats
RAM 1486/3957MB (lfb 1x2MB) CPU [17%@102,11%@102,16%@102,10%@102] EMC_FREQ 0% GR3D_FREQ 5% PLL@31.5C CPU@37.5C PMIC@100C GPU@32C AO@38.5C thermal@34.75C POM_5V_IN 1594/1594 POM_5V_GPU 119/119 POM_5V_CPU 159/159
RAM 1486/3957MB (lfb 1x2MB) CPU [19%@102,11%@102,13%@102,9%@102] EMC_FREQ 0% GR3D_FREQ 39% PLL@31.5C CPU@37.5C PMIC@100C GPU@32C AO@38.5C thermal@34.75C POM_5V_IN 1634/1614 POM_5V_GPU 119/119 POM_5V_CPU 159/159
^C

(This happened to be on a nano, but it’s the same on my Xavier)

8

Isn’t there any tool that delivers the information from tegrastats in the form of graphs?

I can read any one line of output from tegrastats, yet, from a screen full of constantly changing characters I can not get any idea for my GPU and CPU usage. Clearly, a tool like “System Monitor” is infinitely more useful. Unfortunately, it reports only the CPU usage and nothing about the GPU.

9

It would be really helpful is there is one, else can someone please explain some of the information displayed here:
My tegrastats output

RAM 12681/15698MB (lfb 318x4MB) SWAP 22/7849MB (cached 0MB) CPU [84%@2265,23%@2265,58%@2265,1%@2265,18%@2265,59%@2265,33%@2265,28%@2265] EMC_FREQ 0% GR3D_FREQ 0% AO@44C GPU@47.5C Tboard@43C Tdiode@45.75C AUX@43.5C CPU@47.5C thermal@45.9C PMIC@100C GPU 0/0 CPU 6329/6329 SOC 2624/2624 CV 0/0 VDDRQ 308/308 SYS5V 2323/2323

GPU@47.5C Tboard@43C Tdiode@45.75C AUX@43.5C CPU@47.5C thermal@45.9C PMIC@100C GPU 0/0 CPU 6329/6329

What do these mean?

10

Hi tangenishikomba,

Please refer to https://docs.nvidia.com/jetson/l4t/index.html#page/Tegra%2520Linux%2520Driver%2520Package%2520Development%2520Guide%2FAppendixTegraStats.html%23

11

I found this site:

The visualization in the terminal is really great and works perfectly :)

Have fun.

3 Likes