Tx2 on jetPack 4.5.1

[npcdna]

When I installed two TX2 systems with the latest jetpack 4.5.1, I went into TX2 and switched to maximum power mode 0. I found that my TX2, CPU2 and cpu3 worked at 0%. Even if I let them work overload and other CPUs worked at about 90%, CPU2 and cpu3 still worked at 0%. Only at the moment of switching mode, they would change. Is this a TX2 hardware problem or not suitable for jetpack 4.5.1.

[WayneWWW]

This is not a bug. These two cores are disabled on purpose.

6 cores are visible but no work is distributed to the 2 Denver cores

Hi. I am wondering why no work is ever assigned to the two Denver cores although I am running in the MAX-N model. I have both tested with nvpmodel -m 0 and by running jetson_clocks. I am able to see all 6 cores in both top, htop and in tegrastats but the CPU load of the two Denver cores always remains at 0%. I am able to launch a new process and move it to the Denver cores with taskset but no work is automatically distributed to these cores. Why is that so? I am running Ubuntu 18.04 from Jet…

Jetson TX2 has 2 cpu online but sitting at 0%

Hello, Our team is currently developing some application on the jetson tx2 and, at the moment, we cannot accurately benchmark it because 2 of the board’s CPU are sitting a 0% (although online), here is a screenshots to show the problem, this frequency is measured on jtop as well as the ubuntu system monitor. [cpu_0pc] Note that, from what we can tell, they are not offline: On jtop, offline CPUs are shown with a red OFF sign (which is clearly not the case here) We have (repeatedly) executed…

2 cores at 0%

Hey, I took a fresh flashed Jetson TX2 device (Jetpack 4.4) and activated all cores sudo nvpmodel -m 0. After that, all 6 cores were displayed in htop, but core 2 and 3 are always at 0%. What can go wrong there? [grafik]

[npcdna]

thanks， i figured that.

[edward1_lin]

Hello Wayne, Could I know why Nvidia default disable the two cores on Jetpack 4.5.1?

[WayneWWW]

Please check the release note in this website. There is the reason for it.

L4T R32.4.4 archive