feat: Add comprehensive benchmark script

- Automates full build process with GFX906 support
- Downloads Llama-2-7B Q4_0 model if not present
- Runs llama-bench with specified parameters for performance testing
- Includes progress indicators and error handling

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <[email protected]>

Author: larkinwc

run_benchmark.sh

@@ -0,0 +1,79 @@
+#!/bin/bash
+
+set -e
+
+echo "=== llama.cpp GFX906 Benchmark Script ==="
+echo
+
+# Colors for output
+RED='\033[0;31m'
+GREEN='\033[0;32m'
+YELLOW='\033[1;33m'
+NC='\033[0m' # No Color
+
+# Step 1: Update submodules
+echo -e "${GREEN}[1/4] Updating submodules...${NC}"
+git submodule update --init --recursive
+
+# Step 2: Build the project with GFX906 support
+echo -e "${GREEN}[2/4] Building llama.cpp with GFX906 support...${NC}"
+cmake -B build -DGGML_HIP=ON -DAMDGPU_TARGETS=gfx906 -DCMAKE_BUILD_TYPE=Release
+cmake --build build --config Release -j$(nproc)
+
+# Step 3: Check for model and download if needed
+MODEL_DIR="models"
+MODEL_FILE="llama-2-7b.Q4_0.gguf"
+MODEL_PATH="${MODEL_DIR}/${MODEL_FILE}"
+MODEL_URL="https://huggingface.co/TheBloke/Llama-2-7B-GGUF/resolve/main/llama-2-7b.Q4_0.gguf"
+
+echo -e "${GREEN}[3/4] Checking for model...${NC}"
+
+# Create models directory if it doesn't exist
+mkdir -p "${MODEL_DIR}"
+
+if [ -f "${MODEL_PATH}" ]; then
+    echo -e "${YELLOW}Model already exists at ${MODEL_PATH}${NC}"
+else
+    echo -e "${YELLOW}Model not found. Downloading from Hugging Face...${NC}"
+    echo "This may take a while (~3.8 GB file)..."
+    
+    # Download with wget (with progress bar)
+    if command -v wget &> /dev/null; then
+        wget -c "${MODEL_URL}" -O "${MODEL_PATH}" --show-progress
+    # Fallback to curl if wget is not available
+    elif command -v curl &> /dev/null; then
+        curl -L "${MODEL_URL}" -o "${MODEL_PATH}" --progress-bar
+    else
+        echo -e "${RED}Error: Neither wget nor curl found. Please install one of them.${NC}"
+        exit 1
+    fi
+    
+    if [ $? -eq 0 ]; then
+        echo -e "${GREEN}Model downloaded successfully!${NC}"
+    else
+        echo -e "${RED}Error downloading model. Please check your internet connection.${NC}"
+        exit 1
+    fi
+fi
+
+# Step 4: Run benchmark
+echo -e "${GREEN}[4/4] Running benchmark...${NC}"
+echo "Configuration:"
+echo "  - Model: ${MODEL_PATH}"
+echo "  - GPU layers: 99 (full offload)"
+echo "  - Flash attention: disabled"
+echo "  - Prompt sizes: 512, 1024"
+echo "  - Generation sizes: 128, 256"
+echo
+
+# Check if the binary exists
+if [ ! -f "build/bin/llama-bench" ]; then
+    echo -e "${RED}Error: llama-bench binary not found. Build may have failed.${NC}"
+    exit 1
+fi
+
+# Run the benchmark
+./build/bin/llama-bench -m "${MODEL_PATH}" -ngl 99 -fa 0 -p 512,1024 -n 128,256
+
+echo
+echo -e "${GREEN}=== Benchmark Complete ===${NC}"