Merge pull request #18 from aktondak/README_update

README addition

Author: aktondak

runner/README.md

@@ -0,0 +1,50 @@
+# Runner Directory Archive Creation Guide
+
+This directory contains test runners and benchmarks for different NPU platforms: `npu3`, `phx`, and `strx`. Each platform has its own subdirectory with specific test configurations and binaries.
+
+## Directory Structure
+
+```
+runner/
+├── npu3/           # NPU3 platform tests
+├── phx/            # Phoenix platform tests
+└── strx/           # Strix platform tests
+```
+
+## Creating Archives
+
+Use the `ar` utility to create archives. **Note**: The `ar` utility preserves directory paths by default and does NOT automatically flatten file structure.
+
+### Platform Archives
+
+```bash
+# Archive all NPU3 subdirectories
+(cd npu3 && find . -type f -print0 | xargs -0 ar -cr xrt_smi_npu3.a)
+
+# Archive all PHX subdirectories  
+(cd phx && find . -type f -print0 | xargs -0 ar -cr xrt_smi_phx.a)
+
+# Archive all STRX subdirectories
+(cd strx && find . -type f -print0 | xargs -0 ar -cr xrt_smi_strx.a)
+```
+
+## File Types in Archives
+
+Each test category typically contains:
+- **ELF files** (`.elf`): Executable binaries for the NPU
+- **JSON files** (`.json`): Configuration profiles and recipes
+- **XCLBIN files** (`.xclbin`): FPGA bitstream files
+- **Archive files** (`.a`): Static library files (XRT SMI)
+
+## Archive Command Options
+
+- `c`: Create archive
+- `r`: Insert files (replace if already exists)
+- The `ar` utility creates archives with all files flattened at the root level
+
+## Notes
+
+- Archives are created in the current working directory
+- Use `ar -t archive_name.a` to list contents of an archive
+- Use `ar -x archive_name.a` to extract all files from an archive
+- All files will be extracted to the current directory without preserving subdirectory structure