How to Vibe Code in Visual Studio Code on your Dell Pro Max 16 Plus with the Qualcomm Inferencing Card in Linux

1. Download the Qwen3-Coder:30B Instruct prebuilt QPC
   For this playbook we are going to use the prebuilt QPC for the Qwen3-Coder-30B-A3B-Instruct model located here: https://dc00tk1pxen80.cloudfront.net/QPC/1.20.4/Qwen3-Coder-30B-A3B-Instruct/
   Follow the steps below to download and extract the QPC from the archive. The AI100 inferencing card uses up to 2 SoCs, this model is prebuilt to support 4K or 8K contexts we will use the 8K context length version so that we can handle larger code prompts. Note that if you do not have curl already installed you can install it in the terminal using the command (sudo apt install curl).
    

   
       # Use existing Downloads folder for model downloads
   
       cd ~/Downloads
   
   
       
       # Download QPC archive from Zentree as qwen3-coder-a3b-instruct.tar.gz
   
       curl -fSL https://dc00tk1pxen80.cloudfront.net/SDK1.20.4/Qwen/Qwen3-Coder-30B-A3B-Instruct/Qwen3_Coder_30B_A3B_Instruct_qpc_16cores_1pl_8192cl_1bs_2devices_mxfp6_mxint8.tar.gz -o qwen3-coder-a3b-instruct.tar.gz
   
   
       
       # Extract QPC archive
   
       tar -xzvf qwen3-coder-a3b-instruct.tar.gz
   
   
       
       # Rename extracted directory to qwen3-coder-a3b-instruct-qpc
   
       mv $(tar -tzf qwen3-coder-a3b-instruct.tar.gz | head -1 | cut -f1 -d"/") qwen3-coder-a3b-instruct-qpc

2. Serve the Qwen3-Coder:30B Instruct model using vLLM
   Open Terminal and run the following command.
   Note that the following mappings are based on the Zentree model page info for Qwen3-Coder-30B-A3B-Instruct-
   • "-v ~/Downloads/qwen3-coder-a3b-instruct-qpc:/root/qpc" line maps to the QPC download directory from step 1.
   • The model to serve is "meta-llama/Llama-3.2-1B-Instruct" (from the Zentree model page.)
   • max-num-seq is 1 ("Full Batch Size" on Zentree model page.)
   • max-model-len is 8192 (from "Context Length (CL)" on Zentree model page.)
   • max-seq_len-to-capture is 1 (from "Chunking Prompt Length" on Zentree model page.)
   
   If you have not previously used this version of the AI100 Docker container, it fetches ~14GB before opening the Docker container.

   
       docker run --rm -it \
   
       --name qaic-bench \
   
       --device=/dev/accel/accel1 \
   
       --device=/dev/accel/accel2 \
   
       --network host \
   
       --ulimit nofile=1048576 \
   
       -e OMP_NUM_THREADS=8 \
   
       -v ~/Downloads/qwen3-coder-a3b-instruct-qpc:/root/qpc \
   
       ghcr.io/quic/cloud_ai_inference_ubuntu24:1.20.6.0 \
   
       /opt/vllm-env/bin/vllm serve Qwen/Qwen3-Coder-30B-A3B-Instruct \
   
       --max-num-seq 1 \
   
       --max-model-len 8192 \
   
       --max-seq_len-to-capture 1 \
   
       --device qaic \
   
       --device-group 0,1 \
   
       --quantization mxfp6 \
   
       --kv-cache-dtype mxint8 \
   
       --override-qaic-config "qpc_path=/root/qpc/qpc"

   
   The model is ready for inferencing once you see the message: "Application startup complete"
   
   To stop vLLM (therefore unloading the model and closing the container), press Ctrl+C in this Terminal window. Larger models may take up to 1 minute to unload.
   
   References: Docker - Qualcomm Cloud AI SDK User Guid and vLLM - Qualcomm Cloud AI SDK User Guide
3. Install Visual Studio Code
   If you do not have Visual Studio Code installed - first download the .deb / x64 Debian/Ubuntu package from the following Microsoft link - Download Visual Studio Code - Mac, Linux, Windows.
   
   The Visual Studio Code .deb package will download into your Downloads directory.
   
   Follow the sample output below to install it.

   
       # download the .deb/x64 version from link above, it should download into your ~/Downloads directory
   
   
       
       dell@dell:~$ cd ~/Downloads
   
   
       
       (base) dell@dell:~/Downloads$ dir code*
   
   
       
       -rw-rw-r-- 1 dell dell 116511274 Feb 11 00:03 code_1.109.2-1770755809_amd64.deb
   
   
   
   
       
       # need to move the code install .deb to /tmp for installation (otherwise get a sandbox error for install)
   
   
       
       (base) dell@dell:~/Downloads$ cp code_1.109.2-1770755809_amd64.deb /tmp
   
   
       
       (base) dell@dell:~/Downloads$ sudo apt-get install /tmp/code_1.109.2-1770755809_amd64.deb
   
   
       
       Reading package lists... Done
   
   
       
       Building dependency tree... Done
   
   
       
       Reading state information... Done
   
   
       
       Note, selecting 'code' instead of '/tmp/code_1.109.2-1770755809_amd64.deb'
   
   
       
       The following NEW packages will be installed:
   
   
       
       code
   
   
       
       0 upgraded, 1 newly installed, 0 to remove and 4 not upgraded.
   
   
       
       Need to get 0 B/117 MB of archives.
   
   
       
       After this operation, 482 MB of additional disk space will be used.
   
   
       
       Get:1 /tmp/code_1.109.2-1770755809_amd64.deb code amd64 1.109.2-1770755809 [117 MB]
   
   
       
       Preconfiguring packages ...
   
   
       
       Selecting previously unselected package code.
   
   
       
       (Reading database ... 550119 files and directories currently installed.)
   
   
       
       Preparing to unpack .../code_1.109.2-1770755809_amd64.deb ...
   
   
       
       Unpacking code (1.109.2-1770755809) ...
   
   
       
       Setting up code (1.109.2-1770755809) ...
   
   
       
       Processing triggers for gnome-menus (3.36.0-1.1ubuntu3) ...
   
   
       
       Processing triggers for shared-mime-info (2.4-4) ...
   
   
       
       Processing triggers for desktop-file-utils (0.27-2build1) ...
   
   
       
       (base) dell@dell:~/Downloads$

   
   Once installed you can open Visual Studio Code in a terminal using the following command-

   code

4. Install the Continue.dev Extension and Configure
   In Visual Studio Code, navigate to the Extensions view in the sidebar (press Ctrl+Shift+X). Search for Continue - open-source AI code agent and install the extension.
   If you see trust this workspace - select trust this workspace for the installation.
   To configure the Continue extension.
   1. In Code, select the Continue icon in the sidebar on the left (look for the Continue icon, it says Continue if you hover over it.)
   2. Select the "Local Config" drop-down.
   3. Hover over "Local Config" and Click the Settings Gear that appears.
   4. This opens ~/.continue/config.yaml
   5. Modify the file as follows. Note that for YAML the spacing/tabs before the list in the models: section are very important (will fail to load properly without proper format). Also note that we are setting maxTokens to 7168 (7K, under 8K context length for this example) to avoid prompt length errors.
      ~/.continue/config.yaml

      
              name: Local Config
      
              version: 1.0.0
      
              schema: v1
      
              models:
      
              - name: Autodetect
      
              provider: openai
      
              model: AUTODETECT
      
              apiBase: http://localhost:8000/v1
      
              defaultCompletionOptions:
      
              maxTokens: 7168

   6. Save the file using File → Save from the Code menu.
   7. You can press the X to close the config.yaml editor tab.
   
   If the Continue extension is loaded before the model container is running, no models will be detected. To resolve.
   1. Open the "Local Config" drop-down in the Continue extension.
   2. Click "Reload."
5. Use the Continue Extension
   Quick Start Tutorial - Continue

Back to Top