[CI] Add matrix strategy for backend support in build-and-run workflow

- Introduce `fail-fast: false` and matrix strategy for `opencl` and `ptx` backends.
- Modify build and execution steps to dynamically use the selected backend.

Author: mikepapadim

.github/workflows/build-and-run.yml

@@ -33,6 +33,13 @@ jobs:
     needs: code-quality
     timeout-minutes: 30
 
+    strategy:
+      fail-fast: false
+      matrix:
+        backend:
+          - name: opencl
+          - name: ptx
+
     steps:
       - name: Checkout GPULlama3
         uses: actions/checkout@v4
@@ -54,7 +61,7 @@ jobs:
           source venv/bin/activate
           echo "=== Building TornadoVM ==="
           
-          make
+          make BACKEND=${{ matrix.backend.name }}
           
           echo "=== Searching for TornadoVM SDK directory ==="
           SDK_DIR=$(find dist -type d -maxdepth 3 -path "*/tornadovm-*-opencl" | head -n 1)
@@ -88,69 +95,69 @@ jobs:
         run: |
           cd ${{ github.workspace }}
           export PATH="$TORNADO_SDK/bin:$JAVA_HOME/bin:$PATH"
-          ./llama-tornado --gpu --opencl \
+          ./llama-tornado --gpu --${{ matrix.backend.name }} \
             --model $MODELS_DIR/Llama-3.2-1B-Instruct-F16.gguf \
             --prompt "Say hello"
       - name: FP16 - Run Qwen3-4B-f16.gguf
         run: |
           cd ${{ github.workspace }}
           export PATH="$TORNADO_SDK/bin:$JAVA_HOME/bin:$PATH"
-          ./llama-tornado --gpu --opencl \
+          ./llama-tornado --gpu --${{ matrix.backend.name }} \
             --model $MODELS_DIR/Qwen3-4B-f16.gguf \
             --prompt "Say hello"
       - name: FP16 - Run Mistral-7B-Instruct-v0.3.fp16.gguf
         run: |
           cd ${{ github.workspace }}
           export PATH="$TORNADO_SDK/bin:$JAVA_HOME/bin:$PATH"
-          ./llama-tornado --gpu --opencl \
+          ./llama-tornado --gpu --${{ matrix.backend.name }} \
             --model $MODELS_DIR/Mistral-7B-Instruct-v0.3.fp16.gguf \
             --prompt "Say hello"
       - name: FP16 - Run Qwen2.5-1.5b-instruct-fp16.gguf
         run: |
           cd ${{ github.workspace }}
           export PATH="$TORNADO_SDK/bin:$JAVA_HOME/bin:$PATH"
-          ./llama-tornado --gpu --opencl \
+          ./llama-tornado --gpu --${{ matrix.backend.name }} \
             --model $MODELS_DIR/qwen2.5-1.5b-instruct-fp16.gguf \
             --prompt "Say hello"
       - name: FP16 - Run Phi-3-mini-4k-instruct-fp16.gguf
         run: |
           cd ${{ github.workspace }}
           export PATH="$TORNADO_SDK/bin:$JAVA_HOME/bin:$PATH"
-          ./llama-tornado --gpu --opencl \
+          ./llama-tornado --gpu --${{ matrix.backend.name }} \
             --model /$MODELS_DIR/Phi-3-mini-4k-instruct-fp16.gguf \
             --prompt "Say hello"
       - name: Q8 - Run Llama-3.2-1B-Instruct-Q8_0.gguf
         run: |
           cd ${{ github.workspace }}
           export PATH="$TORNADO_SDK/bin:$JAVA_HOME/bin:$PATH"
-          ./llama-tornado --gpu --opencl \
+          ./llama-tornado --gpu --${{ matrix.backend.name }} \
             --model $MODELS_DIR/Llama-3.2-1B-Instruct-Q8_0.gguf \
             --prompt "Say hello"
       - name: Q8 - Run Qwen3-0.6B-Q8_0.gguf
         run: |
           cd ${{ github.workspace }}
           export PATH="$TORNADO_SDK/bin:$JAVA_HOME/bin:$PATH"
-          ./llama-tornado --gpu --opencl \
+          ./llama-tornado --gpu --${{ matrix.backend.name }} \
             --model $MODELS_DIR/Qwen3-0.6B-Q8_0.gguf \
             --prompt "Say hello"
       - name: Q8 - Run Phi-3-mini-4k-instruct-Q8_0.gguf
         run: |
           cd ${{ github.workspace }}
           export PATH="$TORNADO_SDK/bin:$JAVA_HOME/bin:$PATH"
-          ./llama-tornado --gpu --opencl \
+          ./llama-tornado --gpu --${{ matrix.backend.name }} \
             --model $MODELS_DIR/Phi-3-mini-4k-instruct-Q8_0.gguf \
             --prompt "Say hello"
       - name: Q8 - Run Qwen2.5-1.5b-instruct-q8_0.gguf
         run: |
           cd ${{ github.workspace }}
           export PATH="$TORNADO_SDK/bin:$JAVA_HOME/bin:$PATH"
-          ./llama-tornado --gpu --opencl \
+          ./llama-tornado --gpu --${{ matrix.backend.name }} \
             --model $MODELS_DIR/qwen2.5-1.5b-instruct-q8_0.gguf \
             --prompt "Say hello"
       - name: Q8 - Mistral-7B-Instruct-v0.3.Q8_0.gguf
         run: |
           cd ${{ github.workspace }}
           export PATH="$TORNADO_SDK/bin:$JAVA_HOME/bin:$PATH"
-          ./llama-tornado --gpu --opencl \
+          ./llama-tornado --gpu --${{ matrix.backend.name }} \
             --model $MODELS_DIR/Mistral-7B-Instruct-v0.3.Q8_0.gguf \
             --prompt "Say hello"