Summary

• Add CUDA version and driver version display in GPU list view
• Display power consumption in current/max format when max power limit is available
• Add customizable device_type field to support future NPU and other accelerator types
• Expose device_type through API metrics for remote monitoring

Changes

1. Enhanced GPU Information Display

   • Show CUDA version and driver version in GPU list (for NVIDIA/Jetson GPUs)
   • Display power consumption as "current/max W" format when max limit is available
   • Remove decimal places from power values for cleaner output

2. Device Type Support

   • Add device_type field to GpuInfo struct (defaults to "GPU")
   • Display device type instead of hardcoded "GPU" prefix in device list
   • Prepare for future support of NPUs and other accelerator types

3. API and Remote Monitoring

   • Add type label to all_smi_gpu_info metric in API mode
   • Update metrics parser to read device_type from remote hosts
   • Ensure mock server outputs device_type information

Test plan

• Test local view mode with NVIDIA GPUs
• Test remote view mode with mock server
• Verify CUDA/driver info display
• Verify power limit display format
• Test API metrics output includes device_type