Pull request overview

Adds an explicit --vgpu flag to the GPU/CUDA driver installation scripts to force use of the proprietary NVIDIA kernel modules on fractional/vGPU-backed instances (e.g., AWS g6f), where open kernel modules are unsupported.

Changes:

• Add --vgpu CLI flag to installGpuDriver.sh and installCudaDriver.sh.
• When --vgpu is provided, force IS_VGPU=true and prefer proprietary driver candidates.
• Extend vGPU detection with an additional heuristic based on PCI BAR size.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

# If --vgpu flag was passed, skip auto-detection entirely
if [ "$FORCE_VGPU" = true ]; then
    IS_VGPU=true
    echo "  → --vgpu flag set: forcing proprietary driver."
fi
# Method 1: Check PCI descriptions for explicit NVIDIA GRID/vGPU identifiers
if [ "$IS_VGPU" = false ] && sudo lspci -nnk 2>/dev/null | grep -i nvidia | grep -Eqi "vGPU|GRID"; then
    IS_VGPU=true
fi
# Method 2: Check for NVIDIA GRID/vGPU signals in driver state (after driver is present)
if [ -d /proc/driver/nvidia/gpus ] && grep -q -ri "vGPU\|GRID" /proc/driver/nvidia/ 2>/dev/null; then
    IS_VGPU=true

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.