Server upgrade! 4x RAM + GPU! Including CUDA+ollama!
___
/ __| ___ _ _ __ __ ___ _ _
\__ \ / -_) | '_| \ V / / -_) | '_|
|___/ \___| |_| \_/ \___| |_|
_ _
_ _ _ __ __ _ _ _ __ _ __| | ___ | |
| || | | '_ \ / _` | | '_| / _` | / _` | / -_) |_|
\_,_| | .__/ \__, | |_| \__,_| \__,_| \___| (_)
|_| |___/
_ _
| | | __ __
|_ _| \ \ /
|_| /_\_\
___ _ __ __
| _ \ /_\ | \/ |
| / / _ \ | |\/| |
|_|_\ /_/ \_\ |_| |_|
_
_| |_
|_ _|
|_|
___ ___ _ _ _
/ __| | _ \ | | | | | |
| (_ | | _/ | |_| | |_|
\___| |_| \___/ (_)
___ _ _ _
|_ _| _ _ __ | | _ _ __| | (_) _ _ __ _
| | | ' \ / _| | | | || | / _` | | | | ' \ / _` |
|___| |_||_| \__| |_| \_,_| \__,_| |_| |_||_| \__, |
|___/
___ _ _ ___ _ _ _ _ _
/ __| | | | | | \ /_\ _| |_ ___ | | | | __ _ _ __ __ _ | |
| (__ | |_| | | |) | / _ \ |_ _| / _ \ | | | | / _` | | ' \ / _` | |_|
\___| \___/ |___/ /_/ \_\ |_| \___/ |_| |_| \__,_| |_|_|_| \__,_| (_)
╔─*──*──*──*──*──*──*──*──*──*──*──*──*──*──*──*─╗
║1 ........................................ 1║
║2* ........................................ *2║
║3 ........................................ 3║
║1 ...........Posted: 2024-12-04........... 1║
║2* Tags: sysadmin linux hardware ai debian *2║
║3 ........................................ 3║
║1 ........................................ 1║
╚────────────────────────────────────────────────╝
This is both a server announcement and a guide!
I quadrupled my server memory and installed a GPU.
Quadro P620
## Lenovo ST50 "Unqualified DIMM"
## Making the most out of my Nvidia p620 (?)
### Getting the GPU working
Time to install the appropriate drivers.
```
sudo apt update
sudo apt install nvidia-driver
```
I got some warnings about some driver bugs with certain Linux kernel, but after
confirming I wasn't affected by checking my kernel version `uname -r` I just
accepted/moved on. My display stopped displaying during the install process.
After reboot, check if it's working:
```
nvidia-smi
```
I got:
```
baudrillard@simulacra ~ % nvidia-smi
Wed Dec 4 16:18:35 2024
+---------------------------------------------------------------------------------------+
| NVIDIA-SMI 535.183.01 Driver Version: 535.183.01 CUDA Version: 12.2 |
|-----------------------------------------+----------------------+----------------------+
| 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 Quadro P620 On | 00000000:02:00.0 On | N/A |
| 34% 44C P8 N/A / N/A | 188MiB / 2048MiB | 0% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+
+---------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=======================================================================================|
| 0 N/A N/A 10606 G /usr/lib/xorg/Xorg 185MiB |
+---------------------------------------------------------------------------------------+
```
Can check OpenGL, if you're using an actual display (maybe works in xrdp, too):
```
sudo apt update
sudo apt install mesa-utils
glxinfo | grep OpenGL
```
#### Caveats
##### xrdp
After I installed the GPU I noticed xrdp stopped working!
IDK why but I rebooted or something and now it's working, so just try connecting
a few times, restarting, etc.
### Testing `ollama`
I checked these pages to make sure my GPU is supported:
* https://developer.nvidia.com/cuda-gpus
* https://github.com/ollama/ollama/blob/main/docs/gpu.md
I saw the P620 is supported on both pages, so pretty sure!
I installed the `nvidia-cuda-toolkit`:
```
sudo apt install nvidia-cuda-toolkit
```
Verify CUDA installation:
```
nvcc --version
```
Although this prompted me about how `gcc-11` has a serious bug that should not
be shipped in `trixie`. I just accepted the risk.
I'm going to install `nvtop` to see if my GPU is being used by `ollama` (I ran
`ollama run llama3`. While `nvtop` was running I asked `ollama` to "write a
story that's mathematically complex" and I saw my GPU usage shoot up. I was also
able to see by running `nvidia-smi`, you could also do something like `watch -n
0.5 nvidia-smi`.
All looks pretty good to me!
## Final notes
Darn, that fan is sure annoying on that little low-profile P620.