“I wanted to share my work: Linux Q83 on Orion O6” “Sharing my project: Linux Q83 on Orion O6”

Orion Series O6
1

I would like to share my recent project: running Linux Q83 on the Orion O6 platform .

The main goal of this work was to test the compatibility, performance, and stability of Linux Q83 on Orion O6 hardware. I focused on system installation, configuration, and optimization for this environment.

Key points of the project:

  • Successful installation of Linux Q83 on Orion O6
  • Hardware compatibility testing (CPU, GPU, peripherals)
  • System performance measurements
  • Bug fixes and improvements

This project may be useful for anyone interested in experimenting with Linux on less common platforms or exploring Orion O6 hardware capabilities.

1 Like
2

does the npu/gpu work, can we run ollama on gpu on it ?

3

The underlying dependency of ollama is llama.cpp.
Therefore, if you need to use GPU inference for LLM, it is recommended to use the combination of llama.cpp and GPU acceleration: enable -DGGML_VULKAN=ON, compile llama.cpp, and obtain the “c++ API for opening vulkan configuration: llama-cli”.
Use this llama-cli for GPU inference on Orion O6.

4

Note that there’s already one thread on this topic (Does Vulkan work?) where several of us found that while it works it does not improve anything.

5

guess its time to test a gpt-oss gguf with ollama

ollama pull hf.co/unsloth/gpt-oss-20b-GGUF:Q8_0