Failed to align when lengths of PID are greater than 5.
lixinye-nju opened this issue · 7 comments
The application is very convenient but I ran into alignment issues when the PID is too long.
+-----------------------------------------------------------------------------+
| GPU PID USER GPU MEM %CPU %MEM TIME COMMAND |
| 0 3753 root 4MiB 0.0 0.0 18:13:36 /usr/lib/xorg/Xorg v |
| 0 164425 xxxx 10647MiB 106 0.3 35:59 python main.py -i 6 |
| 1 3753 root 4MiB 0.0 0.0 18:13:36 /usr/lib/xorg/Xorg v |
| 1 161143 xxxx 10923MiB 100 0.6 59:46 python main.py -i 1 |
| 2 3753 root 4MiB 0.0 0.0 18:13:36 /usr/lib/xorg/Xorg v |
| 2 161102 xxxx 10731MiB 100 0.6 01:00:02 python main.py -i 1 |
| 3 3753 root 4MiB 0.0 0.0 18:13:36 /usr/lib/xorg/Xorg v |
| 3 160197 xxxx 10195MiB 141 0.6 01:00:49 python main.py -i 1 |
| 4 3753 root 4MiB 0.0 0.0 18:13:36 /usr/lib/xorg/Xorg v |
| 4 160481 xxxx 11099MiB 101 0.4 01:00:40 python main.py -i 3 |
| 5 3753 root 4MiB 0.0 0.0 18:13:36 /usr/lib/xorg/Xorg v |
| 5 159486 xxxx 10761MiB 101 0.4 01:01:59 python main.py -i 3 |
| 6 3753 root 4MiB 0.0 0.0 18:13:36 /usr/lib/xorg/Xorg v |
| 6 159424 xxxx 10623MiB 98.2 0.5 01:02:30 python main.py -i 5 |
| 7 3753 root 4MiB 0.0 0.0 18:13:36 /usr/lib/xorg/Xorg v |
| 7 159383 xxxx 10311MiB 98.7 0.5 01:02:53 python main.py -i 5 |
+-----------------------------------------------------------------------------+
I patched this with the following starting on line 167:
+ max_pid_length = max([len(x) for x in pid])
+ if max_pid_length >= 6:
+ command_length -= (max_pid_length - 6)
+ format = ("| %3s %" + str(max_pid_length) + "s %8s %8s %5s %5s %9s %-" + str(command_length) + "." + str(command_length) + "s |")
- format = ("| %3s %5s %8s %8s %5s %5s %9s %-" + str(command_length) + "." + str(command_length) + "s |")
This worked for me.
Thanks, the proposed patch looks good. Do you want to send it as a pull request to get the proper credit for it?
Thanks. But I install nvidia-htop through pip and it is inconvenient to modify the code on multiple machines. I really appreciate it if you send it as a pull request.
I patched this with the following starting on line 167:
+ max_pid_length = max([len(x) for x in pid]) + if max_pid_length >= 6: + command_length -= (max_pid_length - 6) + format = ("| %3s %" + str(max_pid_length) + "s %8s %8s %5s %5s %9s %-" + str(command_length) + "." + str(command_length) + "s |") - format = ("| %3s %5s %8s %8s %5s %5s %9s %-" + str(command_length) + "." + str(command_length) + "s |")This worked for me.
No problem, I just wanted to give you the opportunity to claim the credit. I'll make the release shortly
You're good! I was in a meeting. I don't care about the credit, it's just an if statement.
The release is out. I needed to do a bit more changes to satisfy the automated tests.
I also added a test for your case with longer PIDs. Could you please review files https://github.com/peci1/nvidia-htop/blob/master/test/FAKE_STDIN_LONG_PIDS and https://github.com/peci1/nvidia-htop/blob/master/test/FAKE_PS_LONG_PIDS and report to me whether your output looks similar (mostly regarding to the number of spaces/lengths of rows). The first one corresponds to the output of nvidia-smi, the latter one to the output of ps -o "pid,user,%cpu,%mem,etime,command".
The PIDs I have are in the 7-digits while @rookielxy had 6-digits, so I think it's good. Here's my nvidia-smi:
Tue Aug 31 12:10:51 2021
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 460.91.03 Driver Version: 460.91.03 CUDA Version: 11.2 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|===============================+======================+======================|
| 0 GeForce RTX 3090 Off | 00000000:01:00.0 On | N/A |
| 0% 43C P8 25W / 350W | 19515MiB / 24265MiB | 4% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+
| 1 GeForce RTX 3090 Off | 00000000:21:00.0 Off | N/A |
| 0% 44C P8 25W / 350W | 16219MiB / 24268MiB | 0% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
| 0 N/A N/A 658723 C .../jupyterNLPenv/bin/python 18887MiB |
| 0 N/A N/A 1429000 G /usr/bin/gnome-shell 57MiB |
| 1 N/A N/A 658723 C .../jupyterNLPenv/bin/python 16201MiB |
+-----------------------------------------------------------------------------+
Here are my PIDs from ps -o "pid,user,%cpu,%mem,etime,command" as well:
PID USER %CPU %MEM ELAPSED COMMAND
1429344 sjwhitak 0.0 0.0 52:42 bash
1429538 sjwhitak 9.1 0.8 52:00 /usr/lib/firefox/firefox
1429624 sjwhitak 6.5 0.8 52:00 /usr/lib/firefox/firefox -contentproc -c
1429671 sjwhitak 0.4 0.2 52:00 /usr/lib/firefox/firefox -contentproc -c
1429760 sjwhitak 0.0 0.1 51:52 /usr/lib/firefox/firefox -contentproc -c
1429824 sjwhitak 0.0 0.0 51:46 /usr/lib/firefox/firefox -contentproc -p
1440708 sjwhitak 0.0 0.0 00:00 ps -o pid,user,%cpu,%mem,etime,command
and this is nvidia-htop without the header (it's the same as nvidia-smi):
+-----------------------------------------------------------------------------+
| GPU PID USER GPU MEM %CPU %MEM TIME COMMAND |
| 0 658723 xxxxxxx 18887MiB 1.1 10.1 2 days /home/xxxxxxx/jupyt |
| 0 1429000 sjwhitak 36MiB 1.5 0.4 56:20 /usr/bin/gnome-shel |
| 1 658723 xxxxxxx 16201MiB 1.1 10.1 2 days /home/xxxxxxx/jupyt |
+-----------------------------------------------------------------------------+
I'm not really sure what you want me to verify, sorry. I'd say the test files look good to me, though.