mirror of
https://github.com/Dispatcharr/Dispatcharr.git
synced 2026-07-26 11:34:32 +00:00
Huge improvement to hardware acceleration script. Renamed for accuracy.
This commit is contained in:
SergeantPanda
parent
634d16d402
commit
0953e044b7
3 changed files with 553 additions and 389 deletions
|
|
@ -99,7 +99,7 @@ echo "Starting init process..."
|
|||
. /app/docker/init/01-user-setup.sh
|
||||
. /app/docker/init/02-postgres.sh
|
||||
. /app/docker/init/03-init-dispatcharr.sh
|
||||
. /app/docker/init/04-check-gpu.sh
|
||||
. /app/docker/init/04-check-hwaccel.sh
|
||||
|
||||
# Start PostgreSQL
|
||||
echo "Starting Postgres..."
|
||||
|
|
|
|||
|
|
@ -1,388 +0,0 @@
|
|||
#!/bin/bash
|
||||
|
||||
echo "🔍 Checking for GPU acceleration devices..."
|
||||
|
||||
# Helper function for device access checks
|
||||
check_dev() {
|
||||
local dev=$1
|
||||
if [ -e "$dev" ]; then
|
||||
if [ -r "$dev" ] && [ -w "$dev" ]; then
|
||||
echo "✅ Device $dev is accessible."
|
||||
else
|
||||
echo "⚠️ Device $dev exists but is not accessible. Check permissions or container runtime options."
|
||||
fi
|
||||
else
|
||||
echo "ℹ️ Device $dev does not exist."
|
||||
fi
|
||||
}
|
||||
|
||||
# Check Intel/AMD VAAPI devices
|
||||
echo "🔍 Checking for Intel/AMD (VAAPI) devices..."
|
||||
for dev in /dev/dri/renderD* /dev/dri/card*; do
|
||||
[ -e "$dev" ] && check_dev "$dev"
|
||||
done
|
||||
|
||||
# Check NVIDIA device nodes
|
||||
echo "🔍 Checking for NVIDIA devices..."
|
||||
NVIDIA_FOUND=false
|
||||
for dev in /dev/nvidia*; do
|
||||
[ -e "$dev" ] && NVIDIA_FOUND=true && check_dev "$dev"
|
||||
done
|
||||
if [ "$NVIDIA_FOUND" = false ]; then
|
||||
echo "ℹ️ No NVIDIA device nodes found under /dev."
|
||||
fi
|
||||
|
||||
# Check group membership for GPU access - context-aware based on hardware
|
||||
echo "🔍 Checking user group memberships..."
|
||||
VIDEO_GID=$(getent group video | cut -d: -f3)
|
||||
RENDER_GID=$(getent group render | cut -d: -f3)
|
||||
NVIDIA_CONTAINER_TOOLKIT_FOUND=false
|
||||
NVIDIA_ENV_MISMATCH=false
|
||||
|
||||
# Detect GPU type for smarter group membership requirements
|
||||
INTEL_GPU_DETECTED=false
|
||||
AMD_GPU_DETECTED=false
|
||||
if command -v lspci >/dev/null 2>&1; then
|
||||
if lspci | grep -q "Intel Corporation.*VGA"; then
|
||||
INTEL_GPU_DETECTED=true
|
||||
echo "ℹ️ Intel GPU detected - video group membership is recommended."
|
||||
elif lspci | grep -q "Advanced Micro Devices.*VGA"; then
|
||||
AMD_GPU_DETECTED=true
|
||||
echo "ℹ️ AMD GPU detected - render group membership is recommended."
|
||||
fi
|
||||
fi
|
||||
|
||||
# Explicitly check if $POSTGRES_USER is in the video/render groups based on detected hardware
|
||||
echo "🔍 Verifying if $POSTGRES_USER user is in required groups..."
|
||||
USER_IN_VIDEO_GROUP=false
|
||||
USER_IN_RENDER_GROUP=false
|
||||
|
||||
# For Intel GPUs or when GPU type can't be determined, check video group
|
||||
if [ -n "$VIDEO_GID" ]; then
|
||||
if id -nG "$POSTGRES_USER" 2>/dev/null | grep -qw "video"; then
|
||||
USER_IN_VIDEO_GROUP=true
|
||||
echo "✅ User $POSTGRES_USER is in the 'video' group."
|
||||
elif [ "$INTEL_GPU_DETECTED" = true ] || ([ "$NVIDIA_FOUND" = false ] && [ "$DRI_DEVICES_FOUND" = true ]); then
|
||||
echo "⚠️ User $POSTGRES_USER is NOT in the 'video' group - hardware acceleration for Intel GPUs may not work!"
|
||||
else
|
||||
echo "ℹ️ User $POSTGRES_USER is not in the 'video' group - this is fine for NVIDIA GPUs with container toolkit."
|
||||
fi
|
||||
fi
|
||||
|
||||
# For AMD GPUs, check render group
|
||||
if [ -n "$RENDER_GID" ]; then
|
||||
if id -nG "$POSTGRES_USER" 2>/dev/null | grep -qw "render"; then
|
||||
USER_IN_RENDER_GROUP=true
|
||||
echo "✅ User $POSTGRES_USER is in the 'render' group."
|
||||
elif [ "$AMD_GPU_DETECTED" = true ]; then
|
||||
echo "⚠️ User $POSTGRES_USER is NOT in the 'render' group - hardware acceleration for AMD GPUs may not work!"
|
||||
else
|
||||
echo "ℹ️ User $POSTGRES_USER is not in the 'render' group - not needed for Intel/NVIDIA GPUs."
|
||||
fi
|
||||
fi
|
||||
|
||||
# Check if NVIDIA Container Toolkit is present through environment or CLI tool
|
||||
# IMPORTANT: Only mark as found if both env vars AND actual NVIDIA devices exist
|
||||
if [ "$NVIDIA_FOUND" = true ] && command -v nvidia-container-cli >/dev/null 2>&1; then
|
||||
NVIDIA_CONTAINER_TOOLKIT_FOUND=true
|
||||
# Check for environment variables set by NVIDIA Container Runtime, but only if NVIDIA hardware exists
|
||||
elif [ "$NVIDIA_FOUND" = true ] && [ -n "$NVIDIA_VISIBLE_DEVICES" ] && [ -n "$NVIDIA_DRIVER_CAPABILITIES" ]; then
|
||||
NVIDIA_CONTAINER_TOOLKIT_FOUND=true
|
||||
echo "✅ NVIDIA Container Toolkit detected (via environment variables)."
|
||||
echo " The container is properly configured with Docker Compose's 'driver: nvidia' syntax."
|
||||
elif [ -n "$NVIDIA_VISIBLE_DEVICES" ] && [ -n "$NVIDIA_DRIVER_CAPABILITIES" ] && [ "$NVIDIA_FOUND" = false ]; then
|
||||
NVIDIA_ENV_MISMATCH=true
|
||||
fi
|
||||
|
||||
# For NVIDIA GPUs with Container Toolkit, video group is optional
|
||||
if [ "$NVIDIA_FOUND" = true ] && [ "$NVIDIA_CONTAINER_TOOLKIT_FOUND" = true ]; then
|
||||
if [ -n "$VIDEO_GID" ] && id -nG "$POSTGRES_USER" 2>/dev/null | grep -qw "video"; then
|
||||
echo "✅ User $POSTGRES_USER is in the 'video' group (GID $VIDEO_GID)."
|
||||
echo " Note: With NVIDIA Container Toolkit properly configured, this is usually not required."
|
||||
elif [ -n "$VIDEO_GID" ]; then
|
||||
echo "ℹ️ User $POSTGRES_USER is not in the 'video' group, but NVIDIA Container Toolkit is present."
|
||||
echo " This is typically fine as the Container Toolkit handles device permissions."
|
||||
fi
|
||||
# For other GPU types (or NVIDIA without Toolkit), video/render group is important
|
||||
else
|
||||
if [ -n "$VIDEO_GID" ]; then
|
||||
if id -nG "$POSTGRES_USER" 2>/dev/null | grep -qw "video"; then
|
||||
echo "✅ User $POSTGRES_USER is in the 'video' group (GID $VIDEO_GID)."
|
||||
else
|
||||
echo "⚠️ User $POSTGRES_USER is NOT in the 'video' group (GID $VIDEO_GID). Hardware acceleration may not work."
|
||||
fi
|
||||
elif [ -n "$RENDER_GID" ]; then
|
||||
if id -nG "$POSTGRES_USER" 2>/dev/null | grep -qw "render"; then
|
||||
echo "✅ User $POSTGRES_USER is in the 'render' group (GID $RENDER_GID)."
|
||||
else
|
||||
echo "⚠️ User $POSTGRES_USER is NOT in the 'render' group (GID $RENDER_GID). Hardware acceleration may not work."
|
||||
fi
|
||||
else
|
||||
echo "⚠️ Neither 'video' nor 'render' groups found on this system."
|
||||
fi
|
||||
fi
|
||||
|
||||
# Check NVIDIA Container Toolkit support
|
||||
echo "🔍 Checking NVIDIA container runtime support..."
|
||||
|
||||
# More reliable detection of NVIDIA Container Runtime
|
||||
NVIDIA_RUNTIME_ACTIVE=false
|
||||
|
||||
# Method 1: Check for nvidia-container-cli tool
|
||||
if command -v nvidia-container-cli >/dev/null 2>&1; then
|
||||
NVIDIA_RUNTIME_ACTIVE=true
|
||||
echo "✅ NVIDIA Container Runtime detected (nvidia-container-cli found)."
|
||||
|
||||
if nvidia-container-cli info >/dev/null 2>&1; then
|
||||
echo "✅ NVIDIA container runtime is functional."
|
||||
else
|
||||
echo "⚠️ nvidia-container-cli found, but 'info' command failed. Runtime may be misconfigured."
|
||||
fi
|
||||
fi
|
||||
|
||||
# Method 2: Check for NVIDIA Container Runtime specific files
|
||||
if [ -e "/dev/.nv" ] || [ -e "/.nv" ] || [ -e "/.nvidia-container-runtime" ]; then
|
||||
NVIDIA_RUNTIME_ACTIVE=true
|
||||
echo "✅ NVIDIA Container Runtime files detected."
|
||||
fi
|
||||
|
||||
# Method 3: Check cgroup information for NVIDIA
|
||||
if grep -q "nvidia" /proc/self/cgroup 2>/dev/null; then
|
||||
NVIDIA_RUNTIME_ACTIVE=true
|
||||
echo "✅ NVIDIA Container Runtime cgroups detected."
|
||||
fi
|
||||
|
||||
# Final verdict based on hardware AND runtime
|
||||
if [ "$NVIDIA_FOUND" = true ] && [ "$NVIDIA_RUNTIME_ACTIVE" = true ]; then
|
||||
echo "✅ NVIDIA Container Runtime is properly configured with hardware access."
|
||||
elif [ "$NVIDIA_FOUND" = true ] && [ "$NVIDIA_RUNTIME_ACTIVE" = false ]; then
|
||||
echo "ℹ️ NVIDIA GPU detected, but using direct device passthrough instead of Container Runtime."
|
||||
echo " This works but consider using the 'deploy: resources: reservations: devices:' method in docker-compose."
|
||||
elif [ "$NVIDIA_FOUND" = false ] && [ "$NVIDIA_RUNTIME_ACTIVE" = true ]; then
|
||||
echo "⚠️ NVIDIA Container Runtime appears to be configured, but no NVIDIA devices found."
|
||||
echo " Check that your host has NVIDIA drivers installed and GPUs are properly passed to the container."
|
||||
else
|
||||
# No need to show NVIDIA environment variable warnings if they're default in the container
|
||||
echo "ℹ️ Using Intel/AMD GPU hardware for acceleration."
|
||||
fi
|
||||
|
||||
# Run nvidia-smi if available
|
||||
if command -v nvidia-smi >/dev/null 2>&1; then
|
||||
echo "🔍 Running nvidia-smi to verify GPU visibility..."
|
||||
if nvidia-smi >/dev/null 2>&1; then
|
||||
echo "✅ nvidia-smi successful - GPU is accessible to container!"
|
||||
echo " This confirms hardware acceleration should be available to FFmpeg."
|
||||
else
|
||||
echo "⚠️ nvidia-smi command failed. GPU may not be properly mapped into container."
|
||||
fi
|
||||
else
|
||||
echo "ℹ️ nvidia-smi not installed or not in PATH."
|
||||
fi
|
||||
|
||||
# Show relevant environment variables with contextual suggestions
|
||||
echo "🔍 Checking GPU-related environment variables..."
|
||||
|
||||
# Set flags based on device detection
|
||||
DRI_DEVICES_FOUND=false
|
||||
for dev in /dev/dri/renderD* /dev/dri/card*; do
|
||||
if [ -e "$dev" ]; then
|
||||
DRI_DEVICES_FOUND=true
|
||||
break
|
||||
fi
|
||||
done
|
||||
|
||||
# Give contextual suggestions based on detected hardware
|
||||
if [ "$DRI_DEVICES_FOUND" = true ]; then
|
||||
# Detect Intel/AMD GPU model
|
||||
if command -v lspci >/dev/null 2>&1; then
|
||||
GPU_INFO=$(lspci -nn | grep -i "VGA\|Display" | head -1)
|
||||
if [ -n "$GPU_INFO" ]; then
|
||||
echo "🔍 Detected GPU: $GPU_INFO"
|
||||
# Extract model for cleaner display in summary
|
||||
GPU_MODEL=$(echo "$GPU_INFO" | sed -E 's/.*: (.*) \[.*/\1/' | sed 's/Corporation //' | sed 's/Technologies //')
|
||||
fi
|
||||
fi
|
||||
|
||||
if [ -n "$LIBVA_DRIVER_NAME" ]; then
|
||||
echo "ℹ️ LIBVA_DRIVER_NAME is set to '$LIBVA_DRIVER_NAME'"
|
||||
else
|
||||
# Check if we can detect the GPU type
|
||||
if command -v lspci >/dev/null 2>&1; then
|
||||
if lspci | grep -q "Intel Corporation.*VGA" | grep -i "HD Graphics"; then
|
||||
# Newer Intel GPUs (Gen12+/Tiger Lake and newer)
|
||||
if lspci | grep -q "Intel Corporation.*VGA" | grep -E "Xe|Alchemist|Tiger Lake|Gen1[2-9]"; then
|
||||
echo "💡 Consider setting LIBVA_DRIVER_NAME=iHD for this modern Intel GPU"
|
||||
else
|
||||
# Older Intel GPUs
|
||||
echo "💡 Consider setting LIBVA_DRIVER_NAME=i965 for this Intel GPU"
|
||||
fi
|
||||
elif lspci | grep -q "Advanced Micro Devices.*VGA"; then
|
||||
echo "💡 Consider setting LIBVA_DRIVER_NAME=radeonsi for this AMD GPU"
|
||||
else
|
||||
echo "ℹ️ Auto-detection of VAAPI driver is usually reliable, but if you have issues:"
|
||||
echo " - For newer Intel GPUs: LIBVA_DRIVER_NAME=iHD"
|
||||
echo " - For older Intel GPUs: LIBVA_DRIVER_NAME=i965"
|
||||
echo " - For AMD GPUs: LIBVA_DRIVER_NAME=radeonsi"
|
||||
fi
|
||||
else
|
||||
echo "ℹ️ Intel/AMD GPU detected. If VAAPI doesn't work automatically, you may need to set LIBVA_DRIVER_NAME"
|
||||
echo " based on your specific hardware (iHD for newer Intel, i965 for older Intel, radeonsi for AMD)"
|
||||
fi
|
||||
fi
|
||||
fi
|
||||
|
||||
if [ "$NVIDIA_FOUND" = true ]; then
|
||||
# Try to get NVIDIA GPU model info
|
||||
if command -v nvidia-smi >/dev/null 2>&1; then
|
||||
NVIDIA_MODEL=$(nvidia-smi --query-gpu=name --format=csv,noheader 2>/dev/null | head -1)
|
||||
if [ -n "$NVIDIA_MODEL" ]; then
|
||||
echo "🔍 Detected NVIDIA GPU: $NVIDIA_MODEL"
|
||||
GPU_MODEL=$NVIDIA_MODEL
|
||||
fi
|
||||
fi
|
||||
|
||||
if [ -n "$NVIDIA_VISIBLE_DEVICES" ]; then
|
||||
echo "ℹ️ NVIDIA_VISIBLE_DEVICES is set to '$NVIDIA_VISIBLE_DEVICES'"
|
||||
else
|
||||
echo "💡 Consider setting NVIDIA_VISIBLE_DEVICES to 'all' or specific indices (e.g., '0,1')"
|
||||
fi
|
||||
|
||||
if [ -n "$NVIDIA_DRIVER_CAPABILITIES" ]; then
|
||||
echo "ℹ️ NVIDIA_DRIVER_CAPABILITIES is set to '$NVIDIA_DRIVER_CAPABILITIES'"
|
||||
else
|
||||
echo "💡 Consider setting NVIDIA_DRIVER_CAPABILITIES to 'all' or 'compute,video,utility' for full functionality"
|
||||
fi
|
||||
|
||||
if [ -n "$CUDA_VISIBLE_DEVICES" ]; then
|
||||
echo "ℹ️ CUDA_VISIBLE_DEVICES is set to '$CUDA_VISIBLE_DEVICES'"
|
||||
fi
|
||||
fi
|
||||
|
||||
# Check FFmpeg hardware acceleration support
|
||||
echo "🔍 Checking FFmpeg hardware acceleration capabilities..."
|
||||
if command -v ffmpeg >/dev/null 2>&1; then
|
||||
HWACCEL=$(ffmpeg -hide_banner -hwaccels 2>/dev/null | grep -v "Hardware acceleration methods:" || echo "None found")
|
||||
echo "Available FFmpeg hardware acceleration methods:"
|
||||
echo "$HWACCEL"
|
||||
else
|
||||
echo "⚠️ FFmpeg not found in PATH."
|
||||
fi
|
||||
|
||||
# Add hardware detection through lspci if available
|
||||
NVIDIA_GPU_IN_LSPCI=false
|
||||
if command -v lspci >/dev/null 2>&1; then
|
||||
if lspci | grep -i "NVIDIA" | grep -i "VGA\|3D\|Display" >/dev/null; then
|
||||
NVIDIA_GPU_IN_LSPCI=true
|
||||
echo "🔍 NVIDIA GPU detected in hardware listing (lspci)."
|
||||
if [ "$NVIDIA_FOUND" = false ]; then
|
||||
echo "⚠️ NVIDIA GPU exists in system but no /dev/nvidia* devices are mapped to the container."
|
||||
echo " You have DRI devices mapped, but this provides limited functionality for NVIDIA GPUs."
|
||||
echo " Consider using the NVIDIA Container Runtime for optimal performance."
|
||||
fi
|
||||
fi
|
||||
fi
|
||||
|
||||
# Provide a final summary of the hardware acceleration setup
|
||||
echo "📋 ===================== SUMMARY ====================="
|
||||
|
||||
# Identify which GPU type is active and working
|
||||
if [ "$NVIDIA_FOUND" = true ] && (nvidia-smi >/dev/null 2>&1 || [ -n "$NVIDIA_VISIBLE_DEVICES" ]); then
|
||||
if [ -n "$NVIDIA_MODEL" ]; then
|
||||
echo "🔰 NVIDIA GPU: $NVIDIA_MODEL"
|
||||
else
|
||||
echo "🔰 NVIDIA GPU: ACTIVE (model detection unavailable)"
|
||||
echo "ℹ️ Note: GPU model information couldn't be retrieved, but devices are present."
|
||||
echo " This may be due to missing nvidia-smi tool or container limitations."
|
||||
fi
|
||||
|
||||
if [ "$NVIDIA_CONTAINER_TOOLKIT_FOUND" = true ]; then
|
||||
echo "✅ NVIDIA Container Toolkit: CONFIGURED CORRECTLY"
|
||||
elif [ -n "$NVIDIA_VISIBLE_DEVICES" ] && [ -n "$NVIDIA_DRIVER_CAPABILITIES" ]; then
|
||||
echo "✅ NVIDIA Docker configuration: USING MODERN DEPLOYMENT"
|
||||
else
|
||||
echo "⚠️ NVIDIA setup method: DIRECT DEVICE MAPPING (functional but not optimal)"
|
||||
fi
|
||||
# Display FFmpeg NVIDIA acceleration methods
|
||||
if echo "$HWACCEL" | grep -q "cuda\|nvenc\|cuvid"; then
|
||||
echo "✅ FFmpeg NVIDIA acceleration: AVAILABLE"
|
||||
else
|
||||
echo "⚠️ FFmpeg NVIDIA acceleration: NOT DETECTED"
|
||||
fi
|
||||
elif [ "$NVIDIA_GPU_IN_LSPCI" = true ] && [ "$DRI_DEVICES_FOUND" = true ]; then
|
||||
# NVIDIA through DRI only (suboptimal but possible)
|
||||
echo "🔰 NVIDIA GPU: DETECTED BUT SUBOPTIMALLY CONFIGURED"
|
||||
echo "⚠️ Your NVIDIA GPU is only accessible through DRI devices, not NVIDIA container runtime"
|
||||
echo "💡 RECOMMENDATION: Use the proper NVIDIA container configuration:"
|
||||
echo " deploy:"
|
||||
echo " resources:"
|
||||
echo " reservations:"
|
||||
echo " devices:"
|
||||
echo " - driver: nvidia"
|
||||
echo " count: all"
|
||||
echo " capabilities: [gpu]"
|
||||
|
||||
if [ "$USER_IN_VIDEO_GROUP" = true ] || [ "$USER_IN_RENDER_GROUP" = true ]; then
|
||||
echo "✅ GPU group membership for $POSTGRES_USER: CORRECT"
|
||||
else
|
||||
echo "⚠️ GPU group membership for $POSTGRES_USER: MISSING (may affect DRI access)"
|
||||
fi
|
||||
|
||||
if echo "$HWACCEL" | grep -q "vaapi"; then
|
||||
echo "✅ FFmpeg VAAPI acceleration: AVAILABLE"
|
||||
else
|
||||
echo "⚠️ FFmpeg VAAPI acceleration: NOT DETECTED"
|
||||
fi
|
||||
elif [ "$DRI_DEVICES_FOUND" = true ]; then
|
||||
# Intel/AMD detection with model if available
|
||||
if [ -n "$GPU_MODEL" ]; then
|
||||
echo "🔰 GPU: $GPU_MODEL"
|
||||
elif [ -n "$LIBVA_DRIVER_NAME" ]; then
|
||||
echo "🔰 ${LIBVA_DRIVER_NAME^^} GPU: ACTIVE"
|
||||
else
|
||||
echo "🔰 INTEL/AMD GPU: ACTIVE (model detection unavailable)"
|
||||
echo "ℹ️ Note: Basic GPU drivers appear to be loaded (device nodes exist), but"
|
||||
echo " couldn't identify specific model. This doesn't necessarily indicate a problem."
|
||||
fi
|
||||
|
||||
# Check group membership based on detected GPU type
|
||||
if [ "$INTEL_GPU_DETECTED" = true ]; then
|
||||
if [ "$USER_IN_VIDEO_GROUP" = true ]; then
|
||||
echo "✅ Video group membership for $POSTGRES_USER: CORRECT (required for Intel GPU)"
|
||||
else
|
||||
echo "⚠️ Video group membership for $POSTGRES_USER: MISSING (required for Intel GPU)"
|
||||
fi
|
||||
elif [ "$AMD_GPU_DETECTED" = true ]; then
|
||||
if [ "$USER_IN_RENDER_GROUP" = true ]; then
|
||||
echo "✅ Render group membership for $POSTGRES_USER: CORRECT (required for AMD GPU)"
|
||||
else
|
||||
echo "⚠️ Render group membership for $POSTGRES_USER: MISSING (required for AMD GPU)"
|
||||
fi
|
||||
elif [ "$NVIDIA_FOUND" = true ] && [ "$NVIDIA_CONTAINER_TOOLKIT_FOUND" = false ]; then
|
||||
# For NVIDIA without container toolkit, check video group
|
||||
if [ "$USER_IN_VIDEO_GROUP" = true ]; then
|
||||
echo "✅ Video group membership for $POSTGRES_USER: CORRECT (helps with direct NVIDIA device access)"
|
||||
else
|
||||
echo "⚠️ Video group membership for $POSTGRES_USER: MISSING (may affect direct NVIDIA device access)"
|
||||
fi
|
||||
else
|
||||
# Generic case or NVIDIA with container toolkit (where group membership is less important)
|
||||
if [ "$USER_IN_VIDEO_GROUP" = true ] || [ "$USER_IN_RENDER_GROUP" = true ]; then
|
||||
echo "✅ GPU group membership for $POSTGRES_USER: CORRECT"
|
||||
else
|
||||
echo "ℹ️ GPU group membership for $POSTGRES_USER: NOT REQUIRED (using NVIDIA Container Toolkit)"
|
||||
fi
|
||||
fi
|
||||
|
||||
# Display FFmpeg VAAPI acceleration method
|
||||
if echo "$HWACCEL" | grep -q "vaapi"; then
|
||||
echo "✅ FFmpeg VAAPI acceleration: AVAILABLE"
|
||||
else
|
||||
echo "⚠️ FFmpeg VAAPI acceleration: NOT DETECTED"
|
||||
fi
|
||||
else
|
||||
echo "❌ NO GPU ACCELERATION DETECTED"
|
||||
echo "⚠️ Hardware acceleration is unavailable or misconfigured"
|
||||
fi
|
||||
|
||||
echo "📋 =================================================="
|
||||
echo "✅ GPU detection script complete."
|
||||
552
docker/init/04-check-hwaccel.sh
Normal file
552
docker/init/04-check-hwaccel.sh
Normal file
|
|
@ -0,0 +1,552 @@
|
|||
#!/bin/bash
|
||||
|
||||
echo "🔍 Checking for GPU acceleration devices..."
|
||||
|
||||
# Helper function for device access checks
|
||||
check_dev() {
|
||||
local dev=$1
|
||||
if [ -e "$dev" ]; then
|
||||
if [ -r "$dev" ] && [ -w "$dev" ]; then
|
||||
echo "✅ Device $dev is accessible."
|
||||
else
|
||||
echo "⚠️ Device $dev exists but is not accessible. Check permissions or container runtime options."
|
||||
fi
|
||||
else
|
||||
echo "ℹ️ Device $dev does not exist."
|
||||
fi
|
||||
}
|
||||
|
||||
# Initialize device detection flags
|
||||
ANY_GPU_DEVICES_FOUND=false
|
||||
DRI_DEVICES_FOUND=false
|
||||
NVIDIA_FOUND=false
|
||||
NVIDIA_GPU_IN_LSPCI=false
|
||||
INTEL_GPU_IN_LSPCI=false
|
||||
AMD_GPU_IN_LSPCI=false
|
||||
|
||||
# Check for all GPU types in hardware via lspci
|
||||
if command -v lspci >/dev/null 2>&1; then
|
||||
# Check for NVIDIA GPUs
|
||||
if lspci | grep -i "NVIDIA" | grep -i "VGA\|3D\|Display" >/dev/null; then
|
||||
NVIDIA_GPU_IN_LSPCI=true
|
||||
NVIDIA_MODEL=$(lspci | grep -i "NVIDIA" | grep -i "VGA\|3D\|Display" | head -1 | sed -E 's/.*: (.*) \[.*/\1/' | sed 's/Corporation //')
|
||||
fi
|
||||
|
||||
# Check for Intel GPUs - making sure it's not already detected as NVIDIA
|
||||
if lspci | grep -i "Intel" | grep -v "NVIDIA" | grep -i "VGA\|3D\|Display" >/dev/null; then
|
||||
INTEL_GPU_IN_LSPCI=true
|
||||
INTEL_MODEL=$(lspci | grep -i "Intel" | grep -v "NVIDIA" | grep -i "VGA\|3D\|Display" | head -1 | sed -E 's/.*: (.*) \[.*/\1/' | sed 's/Corporation //')
|
||||
fi
|
||||
|
||||
# Check for AMD GPUs - making sure it's not already detected as NVIDIA or Intel
|
||||
if lspci | grep -i "AMD\|ATI\|Advanced Micro Devices" | grep -v "NVIDIA\|Intel" | grep -i "VGA\|3D\|Display" >/dev/null; then
|
||||
AMD_GPU_IN_LSPCI=true
|
||||
AMD_MODEL=$(lspci | grep -i "AMD\|ATI\|Advanced Micro Devices" | grep -v "NVIDIA\|Intel" | grep -i "VGA\|3D\|Display" | head -1 | sed -E 's/.*: (.*) \[.*/\1/' | sed 's/Corporation //' | sed 's/Technologies //')
|
||||
fi
|
||||
|
||||
# Display detected GPU hardware
|
||||
if [ "$NVIDIA_GPU_IN_LSPCI" = true ]; then
|
||||
echo "🔍 Hardware detection: NVIDIA GPU ($NVIDIA_MODEL)"
|
||||
fi
|
||||
if [ "$INTEL_GPU_IN_LSPCI" = true ]; then
|
||||
echo "🔍 Hardware detection: Intel GPU ($INTEL_MODEL)"
|
||||
fi
|
||||
if [ "$AMD_GPU_IN_LSPCI" = true ]; then
|
||||
echo "🔍 Hardware detection: AMD GPU ($AMD_MODEL)"
|
||||
fi
|
||||
fi
|
||||
|
||||
# Silently check for any GPU devices first
|
||||
for dev in /dev/dri/renderD* /dev/dri/card* /dev/nvidia*; do
|
||||
if [ -e "$dev" ]; then
|
||||
ANY_GPU_DEVICES_FOUND=true
|
||||
break
|
||||
fi
|
||||
done
|
||||
|
||||
# Only if devices might exist, show detailed checks
|
||||
if [ "$ANY_GPU_DEVICES_FOUND" = true ]; then
|
||||
# Check Intel/AMD VAAPI devices
|
||||
echo "🔍 Checking for VAAPI device nodes (Intel/AMD)..."
|
||||
for dev in /dev/dri/renderD* /dev/dri/card*; do
|
||||
if [ -e "$dev" ]; then
|
||||
DRI_DEVICES_FOUND=true
|
||||
check_dev "$dev"
|
||||
fi
|
||||
done
|
||||
|
||||
# Check NVIDIA device nodes
|
||||
echo "🔍 Checking for NVIDIA device nodes..."
|
||||
for dev in /dev/nvidia*; do
|
||||
if [ -e "$dev" ]; then
|
||||
NVIDIA_FOUND=true
|
||||
check_dev "$dev"
|
||||
fi
|
||||
done
|
||||
|
||||
# Show GPU device availability messages
|
||||
if [ "$NVIDIA_FOUND" = false ] && [ "$NVIDIA_GPU_IN_LSPCI" = true ]; then
|
||||
echo "⚠️ No NVIDIA device nodes available despite hardware detection."
|
||||
echo " You may be able to use VAAPI for hardware acceleration, but NVENC/CUDA won't be available."
|
||||
echo " For optimal performance, configure proper NVIDIA container runtime."
|
||||
elif [ "$NVIDIA_FOUND" = false ]; then
|
||||
echo "ℹ️ No NVIDIA device nodes found under /dev."
|
||||
fi
|
||||
|
||||
# Check for Intel/AMD GPUs that might not be fully accessible
|
||||
if [ "$DRI_DEVICES_FOUND" = false ] && [ "$INTEL_GPU_IN_LSPCI" = true ]; then
|
||||
echo "⚠️ Intel GPU detected in hardware but no DRI devices found."
|
||||
echo " Hardware acceleration will not be available."
|
||||
echo " Make sure /dev/dri/ devices are properly mapped to the container."
|
||||
elif [ "$DRI_DEVICES_FOUND" = false ] && [ "$AMD_GPU_IN_LSPCI" = true ]; then
|
||||
echo "⚠️ AMD GPU detected in hardware but no DRI devices found."
|
||||
echo " Hardware acceleration will not be available."
|
||||
echo " Make sure /dev/dri/ devices are properly mapped to the container."
|
||||
fi
|
||||
else
|
||||
# No GPU devices found, skip the detailed checks
|
||||
echo "❌ No GPU acceleration devices detected in this container."
|
||||
echo "ℹ️ Checking for potential configuration issues..."
|
||||
|
||||
# Check if the host might have GPUs that aren't passed to the container
|
||||
if command -v lspci >/dev/null 2>&1; then
|
||||
if lspci | grep -i "VGA\|3D\|Display" | grep -i "NVIDIA\|Intel\|AMD" >/dev/null; then
|
||||
echo "⚠️ Host system appears to have GPU hardware, but no devices are accessible to the container."
|
||||
echo " - For NVIDIA GPUs: Ensure NVIDIA Container Runtime is configured properly"
|
||||
echo " - For Intel/AMD GPUs: Verify that /dev/dri/ devices are passed to the container"
|
||||
echo " - Check your Docker run command or docker-compose.yml for proper device mapping"
|
||||
else
|
||||
echo "ℹ️ No GPU hardware detected on the host system. CPU-only transcoding will be used."
|
||||
fi
|
||||
else
|
||||
echo "ℹ️ Unable to check host GPU hardware (lspci not available). CPU-only transcoding will be used."
|
||||
fi
|
||||
|
||||
echo "📋 =================================================="
|
||||
echo "✅ GPU detection script complete. No GPUs available for hardware acceleration."
|
||||
exit 0
|
||||
fi
|
||||
|
||||
# Check group membership for GPU access - context-aware based on hardware
|
||||
echo "🔍 Checking user group memberships and device access..."
|
||||
VIDEO_GID=$(getent group video | cut -d: -f3)
|
||||
RENDER_GID=$(getent group render | cut -d: -f3)
|
||||
NVIDIA_CONTAINER_TOOLKIT_FOUND=false
|
||||
NVIDIA_ENV_MISMATCH=false
|
||||
|
||||
# Improved device access check function
|
||||
check_user_device_access() {
|
||||
local device=$1
|
||||
local user=$2
|
||||
if [ -e "$device" ]; then
|
||||
if su -c "test -r $device && test -w $device" - $user 2>/dev/null; then
|
||||
echo "✅ User $user has full access to $device"
|
||||
return 0
|
||||
else
|
||||
echo "⚠️ User $user cannot access $device (permission denied)"
|
||||
return 1
|
||||
fi
|
||||
else
|
||||
# Device doesn't exist, no need to report here
|
||||
return 2
|
||||
fi
|
||||
}
|
||||
|
||||
# Direct device access verification for DRI (Intel/AMD)
|
||||
echo "🔍 Verifying if $POSTGRES_USER has direct access to GPU devices..."
|
||||
HAS_DRI_ACCESS=false
|
||||
DRI_ACCESS_COUNT=0
|
||||
DRI_DEVICE_COUNT=0
|
||||
|
||||
for dev in /dev/dri/renderD* /dev/dri/card*; do
|
||||
if [ -e "$dev" ]; then
|
||||
DRI_DEVICE_COUNT=$((DRI_DEVICE_COUNT + 1))
|
||||
if check_user_device_access "$dev" "$POSTGRES_USER"; then
|
||||
DRI_ACCESS_COUNT=$((DRI_ACCESS_COUNT + 1))
|
||||
HAS_DRI_ACCESS=true
|
||||
fi
|
||||
fi
|
||||
done
|
||||
|
||||
# Direct device access verification for NVIDIA
|
||||
HAS_NVIDIA_ACCESS=false
|
||||
NVIDIA_ACCESS_COUNT=0
|
||||
NVIDIA_DEVICE_COUNT=0
|
||||
|
||||
for dev in /dev/nvidia*; do
|
||||
if [ -e "$dev" ]; then
|
||||
NVIDIA_DEVICE_COUNT=$((NVIDIA_DEVICE_COUNT + 1))
|
||||
if check_user_device_access "$dev" "$POSTGRES_USER"; then
|
||||
NVIDIA_ACCESS_COUNT=$((NVIDIA_ACCESS_COUNT + 1))
|
||||
HAS_NVIDIA_ACCESS=true
|
||||
fi
|
||||
fi
|
||||
done
|
||||
|
||||
# Summary of device access
|
||||
if [ $DRI_DEVICE_COUNT -gt 0 ]; then
|
||||
if [ $DRI_ACCESS_COUNT -eq $DRI_DEVICE_COUNT ]; then
|
||||
echo "✅ User $POSTGRES_USER has access to all DRI devices ($DRI_ACCESS_COUNT/$DRI_DEVICE_COUNT)"
|
||||
echo " VAAPI hardware acceleration should work properly."
|
||||
else
|
||||
echo "⚠️ User $POSTGRES_USER has limited access to DRI devices ($DRI_ACCESS_COUNT/$DRI_DEVICE_COUNT)"
|
||||
echo " VAAPI hardware acceleration may not work properly."
|
||||
echo " Consider adding $POSTGRES_USER to the 'video' and/or 'render' groups."
|
||||
fi
|
||||
fi
|
||||
|
||||
if [ $NVIDIA_DEVICE_COUNT -gt 0 ]; then
|
||||
if [ $NVIDIA_ACCESS_COUNT -eq $NVIDIA_DEVICE_COUNT ]; then
|
||||
echo "✅ User $POSTGRES_USER has access to all NVIDIA devices ($NVIDIA_ACCESS_COUNT/$NVIDIA_DEVICE_COUNT)"
|
||||
echo " NVIDIA hardware acceleration should work properly."
|
||||
else
|
||||
echo "⚠️ User $POSTGRES_USER has limited access to NVIDIA devices ($NVIDIA_ACCESS_COUNT/$NVIDIA_DEVICE_COUNT)"
|
||||
echo " NVIDIA hardware acceleration may not work properly."
|
||||
if [ "$NVIDIA_CONTAINER_TOOLKIT_FOUND" = false ]; then
|
||||
echo " Consider adding $POSTGRES_USER to the 'video' group or use NVIDIA Container Toolkit."
|
||||
fi
|
||||
fi
|
||||
fi
|
||||
|
||||
# Check for traditional group memberships (as additional information)
|
||||
USER_IN_VIDEO_GROUP=false
|
||||
USER_IN_RENDER_GROUP=false
|
||||
|
||||
if [ -n "$VIDEO_GID" ]; then
|
||||
if id -nG "$POSTGRES_USER" 2>/dev/null | grep -qw "video"; then
|
||||
USER_IN_VIDEO_GROUP=true
|
||||
echo "ℹ️ User $POSTGRES_USER is in the 'video' group (GID $VIDEO_GID)."
|
||||
fi
|
||||
fi
|
||||
|
||||
if [ -n "$RENDER_GID" ]; then
|
||||
if id -nG "$POSTGRES_USER" 2>/dev/null | grep -qw "render"; then
|
||||
USER_IN_RENDER_GROUP=true
|
||||
echo "ℹ️ User $POSTGRES_USER is in the 'render' group (GID $RENDER_GID)."
|
||||
fi
|
||||
fi
|
||||
|
||||
# Check if NVIDIA Container Toolkit is present through environment or CLI tool
|
||||
# IMPORTANT: Only mark as found if both env vars AND actual NVIDIA devices exist
|
||||
if [ "$NVIDIA_FOUND" = true ] && command -v nvidia-container-cli >/dev/null 2>&1; then
|
||||
NVIDIA_CONTAINER_TOOLKIT_FOUND=true
|
||||
# Check for environment variables set by NVIDIA Container Runtime, but only if NVIDIA hardware exists
|
||||
elif [ "$NVIDIA_FOUND" = true ] && [ -n "$NVIDIA_VISIBLE_DEVICES" ] && [ -n "$NVIDIA_DRIVER_CAPABILITIES" ]; then
|
||||
NVIDIA_CONTAINER_TOOLKIT_FOUND=true
|
||||
echo "✅ NVIDIA Container Toolkit detected (via environment variables)."
|
||||
echo " The container is properly configured with Docker Compose's 'driver: nvidia' syntax."
|
||||
elif [ -n "$NVIDIA_VISIBLE_DEVICES" ] && [ -n "$NVIDIA_DRIVER_CAPABILITIES" ] && [ "$NVIDIA_FOUND" = false ]; then
|
||||
NVIDIA_ENV_MISMATCH=true
|
||||
fi
|
||||
|
||||
# Removed duplicate video group checks here - consolidated into the earlier checks that include GID
|
||||
|
||||
# Check NVIDIA Container Toolkit support
|
||||
echo "🔍 Checking NVIDIA container runtime support..."
|
||||
|
||||
# More reliable detection of NVIDIA Container Runtime
|
||||
NVIDIA_RUNTIME_ACTIVE=false
|
||||
|
||||
# Method 1: Check for nvidia-container-cli tool
|
||||
if command -v nvidia-container-cli >/dev/null 2>&1; then
|
||||
NVIDIA_RUNTIME_ACTIVE=true
|
||||
echo "✅ NVIDIA Container Runtime detected (nvidia-container-cli found)."
|
||||
|
||||
if nvidia-container-cli info >/dev/null 2>&1; then
|
||||
echo "✅ NVIDIA container runtime is functional."
|
||||
else
|
||||
echo "⚠️ nvidia-container-cli found, but 'info' command failed. Runtime may be misconfigured."
|
||||
fi
|
||||
fi
|
||||
|
||||
# Method 2: Check for NVIDIA Container Runtime specific files
|
||||
if [ -e "/dev/.nv" ] || [ -e "/.nv" ] || [ -e "/.nvidia-container-runtime" ]; then
|
||||
NVIDIA_RUNTIME_ACTIVE=true
|
||||
echo "✅ NVIDIA Container Runtime files detected."
|
||||
fi
|
||||
|
||||
# Method 3: Check cgroup information for NVIDIA
|
||||
if grep -q "nvidia" /proc/self/cgroup 2>/dev/null; then
|
||||
NVIDIA_RUNTIME_ACTIVE=true
|
||||
echo "✅ NVIDIA Container Runtime cgroups detected."
|
||||
fi
|
||||
|
||||
# Final verdict based on hardware AND runtime with improved messaging
|
||||
if [ "$NVIDIA_FOUND" = true ] && ([ "$NVIDIA_RUNTIME_ACTIVE" = true ] || [ "$NVIDIA_CONTAINER_TOOLKIT_FOUND" = true ]); then
|
||||
echo "✅ NVIDIA Container Runtime is properly configured with hardware access."
|
||||
elif [ "$NVIDIA_FOUND" = true ] && [ "$NVIDIA_RUNTIME_ACTIVE" = false ] && [ "$NVIDIA_CONTAINER_TOOLKIT_FOUND" = false ]; then
|
||||
echo "ℹ️ NVIDIA devices accessible via direct passthrough instead of Container Runtime."
|
||||
echo " This works but consider using the 'deploy: resources: reservations: devices:' method in docker-compose."
|
||||
elif [ "$NVIDIA_FOUND" = false ] && [ "$NVIDIA_RUNTIME_ACTIVE" = true ]; then
|
||||
echo "⚠️ NVIDIA Container Runtime appears to be configured, but no NVIDIA devices found."
|
||||
echo " Check that your host has NVIDIA drivers installed and GPUs are properly passed to the container."
|
||||
elif [ "$DRI_DEVICES_FOUND" = true ] && [ "$NVIDIA_GPU_IN_LSPCI" = true ]; then
|
||||
echo "ℹ️ Limited GPU access: Only DRI devices available for NVIDIA hardware."
|
||||
echo " VAAPI acceleration may work but NVENC/CUDA won't be available."
|
||||
echo " For full NVIDIA capabilities, configure the NVIDIA Container Runtime."
|
||||
elif [ "$DRI_DEVICES_FOUND" = true ]; then
|
||||
echo "ℹ️ Using Intel/AMD GPU hardware for acceleration via VAAPI."
|
||||
else
|
||||
echo "⚠️ No GPU acceleration devices detected. CPU-only transcoding will be used."
|
||||
fi
|
||||
|
||||
# Run nvidia-smi if available
|
||||
if command -v nvidia-smi >/dev/null 2>&1; then
|
||||
echo "🔍 Running nvidia-smi to verify GPU visibility..."
|
||||
if nvidia-smi >/dev/null 2>&1; then
|
||||
echo "✅ nvidia-smi successful - GPU is accessible to container!"
|
||||
echo " This confirms hardware acceleration should be available to FFmpeg."
|
||||
else
|
||||
echo "⚠️ nvidia-smi command failed. GPU may not be properly mapped into container."
|
||||
fi
|
||||
else
|
||||
echo "ℹ️ nvidia-smi not installed or not in PATH."
|
||||
fi
|
||||
|
||||
# Show relevant environment variables with contextual suggestions
|
||||
echo "🔍 Checking GPU-related environment variables..."
|
||||
|
||||
# Set flags based on device detection
|
||||
DRI_DEVICES_FOUND=false
|
||||
for dev in /dev/dri/renderD* /dev/dri/card*; do
|
||||
if [ -e "$dev" ]; then
|
||||
DRI_DEVICES_FOUND=true
|
||||
break
|
||||
fi
|
||||
done
|
||||
|
||||
# Give contextual suggestions based on detected hardware
|
||||
if [ "$DRI_DEVICES_FOUND" = true ]; then
|
||||
# Detect Intel/AMD GPU model - skip this if we already detected GPUs earlier
|
||||
if [ "$NVIDIA_GPU_IN_LSPCI" = false ] && [ "$INTEL_GPU_IN_LSPCI" = false ] && [ "$AMD_GPU_IN_LSPCI" = false ] && command -v lspci >/dev/null 2>&1; then
|
||||
GPU_INFO=$(lspci -nn | grep -i "VGA\|Display" | head -1)
|
||||
if [ -n "$GPU_INFO" ]; then
|
||||
echo "🔍 Detected GPU: $GPU_INFO"
|
||||
# Extract model for cleaner display in summary
|
||||
GPU_MODEL=$(echo "$GPU_INFO" | sed -E 's/.*: (.*) \[.*/\1/' | sed 's/Corporation //' | sed 's/Technologies //')
|
||||
fi
|
||||
else
|
||||
# Use already detected GPU model info
|
||||
if [ "$NVIDIA_GPU_IN_LSPCI" = true ]; then
|
||||
GPU_MODEL=$NVIDIA_MODEL
|
||||
elif [ "$INTEL_GPU_IN_LSPCI" = true ]; then
|
||||
GPU_MODEL=$INTEL_MODEL
|
||||
elif [ "$AMD_GPU_IN_LSPCI" = true ]; then
|
||||
GPU_MODEL=$AMD_MODEL
|
||||
fi
|
||||
fi
|
||||
|
||||
if [ -n "$LIBVA_DRIVER_NAME" ]; then
|
||||
echo "ℹ️ LIBVA_DRIVER_NAME is set to '$LIBVA_DRIVER_NAME'"
|
||||
else
|
||||
# Check if we can detect the GPU type
|
||||
if command -v lspci >/dev/null 2>&1; then
|
||||
echo "ℹ️ VAAPI driver auto-detection is usually reliable. Settings below only needed if you experience issues."
|
||||
|
||||
# Intel GPU detection with more future-proof approach
|
||||
if lspci | grep -q "Intel Corporation.*VGA"; then
|
||||
# Check for newer Intel generations that use iHD
|
||||
if lspci | grep -q "Intel Corporation.*VGA" | grep -E "Arc|Xe|Alchemist|Tiger Lake|Alder Lake|Raptor Lake|Meteor Lake|Gen1[2-9]"; then
|
||||
echo "💡 If needed: LIBVA_DRIVER_NAME=iHD for modern Intel GPUs (Gen12+/Arc/Xe)"
|
||||
# Check for very old Intel that definitely needs i965
|
||||
elif lspci | grep -q "Intel Corporation.*VGA" | grep -E "Haswell|Broadwell|Skylake|Kaby Lake|Coffee Lake|Whiskey Lake|Comet Lake"; then
|
||||
echo "💡 If needed: LIBVA_DRIVER_NAME=i965 for older Intel GPUs (Gen11 and below)"
|
||||
else
|
||||
# Generic Intel case - could be either, but iHD is more likely for newer hardware
|
||||
echo "💡 If needed: Try LIBVA_DRIVER_NAME=iHD first, or LIBVA_DRIVER_NAME=i965 if that fails"
|
||||
fi
|
||||
elif lspci | grep -q "Advanced Micro Devices.*VGA"; then
|
||||
echo "💡 If needed: LIBVA_DRIVER_NAME=radeonsi for AMD GPUs"
|
||||
else
|
||||
echo "ℹ️ Common VAAPI driver options if auto-detection fails:"
|
||||
echo " - For modern Intel GPUs (Gen12+/Arc/Xe): LIBVA_DRIVER_NAME=iHD"
|
||||
echo " - For older Intel GPUs: LIBVA_DRIVER_NAME=i965"
|
||||
echo " - For AMD GPUs: LIBVA_DRIVER_NAME=radeonsi"
|
||||
fi
|
||||
else
|
||||
echo "ℹ️ Intel/AMD GPU detected. Auto-detection should work in most cases."
|
||||
echo " If VAAPI doesn't work, you may need to set LIBVA_DRIVER_NAME manually."
|
||||
fi
|
||||
fi
|
||||
fi
|
||||
|
||||
# Check FFmpeg hardware acceleration support
|
||||
echo "🔍 Checking FFmpeg hardware acceleration capabilities..."
|
||||
if command -v ffmpeg >/dev/null 2>&1; then
|
||||
HWACCEL=$(ffmpeg -hide_banner -hwaccels 2>/dev/null | grep -v "Hardware acceleration methods:" || echo "None found")
|
||||
|
||||
# Define expected acceleration methods based on detected hardware
|
||||
EXPECTED_METHODS=""
|
||||
MISSING_METHODS=""
|
||||
|
||||
if [ "$NVIDIA_FOUND" = true ]; then
|
||||
EXPECTED_NVIDIA="cuda" #cuvid nvenc nvdec" keeping these in case future support is added
|
||||
for method in $EXPECTED_NVIDIA; do
|
||||
if echo "$HWACCEL" | grep -q "$method"; then
|
||||
EXPECTED_METHODS="$EXPECTED_METHODS $method"
|
||||
else
|
||||
MISSING_METHODS="$MISSING_METHODS $method"
|
||||
fi
|
||||
done
|
||||
fi
|
||||
|
||||
if [ "$INTEL_GPU_IN_LSPCI" = true ] && [ "$DRI_DEVICES_FOUND" = true ]; then
|
||||
EXPECTED_INTEL="vaapi qsv"
|
||||
for method in $EXPECTED_INTEL; do
|
||||
if echo "$HWACCEL" | grep -q "$method"; then
|
||||
EXPECTED_METHODS="$EXPECTED_METHODS $method"
|
||||
else
|
||||
MISSING_METHODS="$MISSING_METHODS $method"
|
||||
fi
|
||||
done
|
||||
fi
|
||||
|
||||
if [ "$AMD_GPU_IN_LSPCI" = true ] && [ "$DRI_DEVICES_FOUND" = true ]; then
|
||||
EXPECTED_AMD="vaapi"
|
||||
for method in $EXPECTED_AMD; do
|
||||
if echo "$HWACCEL" | grep -q "$method"; then
|
||||
EXPECTED_METHODS="$EXPECTED_METHODS $method"
|
||||
else
|
||||
MISSING_METHODS="$MISSING_METHODS $method"
|
||||
fi
|
||||
done
|
||||
fi
|
||||
|
||||
# Show all available methods
|
||||
echo "🔍 All available FFmpeg hardware acceleration methods:"
|
||||
echo "$HWACCEL"
|
||||
|
||||
# Show expected methods based on hardware
|
||||
if [ -n "$EXPECTED_METHODS" ]; then
|
||||
echo "✅ Hardware-appropriate acceleration methods available:$EXPECTED_METHODS"
|
||||
fi
|
||||
|
||||
# Show missing expected methods
|
||||
if [ -n "$MISSING_METHODS" ]; then
|
||||
echo "⚠️ Expected acceleration methods not found:$MISSING_METHODS"
|
||||
echo " This might indicate missing libraries or improper driver configuration."
|
||||
fi
|
||||
|
||||
# Check specific cases of interest
|
||||
if [ "$NVIDIA_FOUND" = true ] && ! echo "$HWACCEL" | grep -q "cuda\|nvenc\|cuvid"; then
|
||||
echo "⚠️ NVIDIA GPU detected but no NVIDIA acceleration methods available."
|
||||
echo " Ensure ffmpeg is built with NVIDIA support and required libraries are installed."
|
||||
fi
|
||||
|
||||
if ([ "$INTEL_GPU_IN_LSPCI" = true ] || [ "$AMD_GPU_IN_LSPCI" = true ]) && [ "$DRI_DEVICES_FOUND" = true ] && ! echo "$HWACCEL" | grep -q "vaapi"; then
|
||||
echo "⚠️ Intel/AMD GPU detected but VAAPI acceleration not available."
|
||||
echo " Ensure ffmpeg is built with VAAPI support and proper drivers are installed."
|
||||
fi
|
||||
else
|
||||
echo "⚠️ FFmpeg not found in PATH."
|
||||
fi
|
||||
|
||||
# Provide a final summary of the hardware acceleration setup
|
||||
echo "📋 ===================== SUMMARY ====================="
|
||||
|
||||
# Identify which GPU type is active and working
|
||||
if [ "$NVIDIA_FOUND" = true ] && (nvidia-smi >/dev/null 2>&1 || [ -n "$NVIDIA_VISIBLE_DEVICES" ]); then
|
||||
if [ -n "$NVIDIA_MODEL" ]; then
|
||||
echo "🔰 NVIDIA GPU: $NVIDIA_MODEL"
|
||||
else
|
||||
echo "🔰 NVIDIA GPU: ACTIVE (model detection unavailable)"
|
||||
echo "ℹ️ Note: GPU model information couldn't be retrieved, but devices are present."
|
||||
echo " This may be due to missing nvidia-smi tool or container limitations."
|
||||
fi
|
||||
|
||||
if [ "$NVIDIA_CONTAINER_TOOLKIT_FOUND" = true ]; then
|
||||
echo "✅ NVIDIA Container Toolkit: CONFIGURED CORRECTLY"
|
||||
elif [ -n "$NVIDIA_VISIBLE_DEVICES" ] && [ -n "$NVIDIA_DRIVER_CAPABILITIES" ]; then
|
||||
echo "✅ NVIDIA Docker configuration: USING MODERN DEPLOYMENT"
|
||||
else
|
||||
echo "⚠️ NVIDIA setup method: DIRECT DEVICE MAPPING (functional but not optimal)"
|
||||
fi
|
||||
|
||||
# Add device accessibility status
|
||||
if [ $NVIDIA_DEVICE_COUNT -gt 0 ]; then
|
||||
if [ $NVIDIA_ACCESS_COUNT -eq $NVIDIA_DEVICE_COUNT ]; then
|
||||
echo "✅ Device access: ALL NVIDIA DEVICES ACCESSIBLE ($NVIDIA_ACCESS_COUNT/$NVIDIA_DEVICE_COUNT)"
|
||||
else
|
||||
echo "⚠️ Device access: LIMITED NVIDIA DEVICE ACCESS ($NVIDIA_ACCESS_COUNT/$NVIDIA_DEVICE_COUNT)"
|
||||
echo " Some hardware acceleration features may not work properly."
|
||||
fi
|
||||
fi
|
||||
|
||||
# Display FFmpeg NVIDIA acceleration methods
|
||||
if echo "$HWACCEL" | grep -q "cuda\|nvenc\|cuvid"; then
|
||||
echo "✅ FFmpeg NVIDIA acceleration: AVAILABLE"
|
||||
else
|
||||
echo "⚠️ FFmpeg NVIDIA acceleration: NOT DETECTED"
|
||||
fi
|
||||
elif [ "$NVIDIA_GPU_IN_LSPCI" = true ] && [ "$DRI_DEVICES_FOUND" = true ]; then
|
||||
# NVIDIA through DRI only (suboptimal but possible)
|
||||
if [ -n "$NVIDIA_MODEL" ]; then
|
||||
echo "🔰 NVIDIA GPU: $NVIDIA_MODEL (SUBOPTIMALLY CONFIGURED)"
|
||||
else
|
||||
echo "🔰 NVIDIA GPU: DETECTED BUT SUBOPTIMALLY CONFIGURED"
|
||||
fi
|
||||
echo "⚠️ Your NVIDIA GPU is only accessible through DRI devices"
|
||||
echo " - VAAPI acceleration may work for some tasks"
|
||||
echo " - NVENC/CUDA acceleration is NOT available"
|
||||
|
||||
# Add device accessibility status
|
||||
if [ $DRI_DEVICE_COUNT -gt 0 ]; then
|
||||
if [ $DRI_ACCESS_COUNT -eq $DRI_DEVICE_COUNT ]; then
|
||||
echo "✅ Device access: ALL DRI DEVICES ACCESSIBLE ($DRI_ACCESS_COUNT/$DRI_DEVICE_COUNT)"
|
||||
echo " VAAPI acceleration should work properly."
|
||||
else
|
||||
echo "⚠️ Device access: LIMITED DRI DEVICE ACCESS ($DRI_ACCESS_COUNT/$DRI_DEVICE_COUNT)"
|
||||
echo " VAAPI acceleration may not work properly."
|
||||
fi
|
||||
fi
|
||||
|
||||
echo "💡 RECOMMENDATION: Use the proper NVIDIA container configuration:"
|
||||
echo " deploy:"
|
||||
echo " resources:"
|
||||
echo " reservations:"
|
||||
echo " devices:"
|
||||
echo " - driver: nvidia"
|
||||
echo " count: all"
|
||||
echo " capabilities: [gpu]"
|
||||
|
||||
if echo "$HWACCEL" | grep -q "vaapi"; then
|
||||
echo "✅ FFmpeg VAAPI acceleration: AVAILABLE"
|
||||
else
|
||||
echo "⚠️ FFmpeg VAAPI acceleration: NOT DETECTED"
|
||||
fi
|
||||
elif [ "$DRI_DEVICES_FOUND" = true ]; then
|
||||
# Intel/AMD detection with model if available
|
||||
if [ -n "$GPU_MODEL" ]; then
|
||||
echo "🔰 GPU: $GPU_MODEL"
|
||||
elif [ -n "$LIBVA_DRIVER_NAME" ]; then
|
||||
echo "🔰 ${LIBVA_DRIVER_NAME^^} GPU: ACTIVE"
|
||||
else
|
||||
echo "🔰 INTEL/AMD GPU: ACTIVE (model detection unavailable)"
|
||||
echo "ℹ️ Note: Basic GPU drivers appear to be loaded (device nodes exist), but"
|
||||
echo " couldn't identify specific model. This doesn't necessarily indicate a problem."
|
||||
fi
|
||||
|
||||
# Add device accessibility status
|
||||
if [ $DRI_DEVICE_COUNT -gt 0 ]; then
|
||||
if [ $DRI_ACCESS_COUNT -eq $DRI_DEVICE_COUNT ]; then
|
||||
echo "✅ Device access: ALL DRI DEVICES ACCESSIBLE ($DRI_ACCESS_COUNT/$DRI_DEVICE_COUNT)"
|
||||
echo " VAAPI hardware acceleration should work properly."
|
||||
else
|
||||
echo "⚠️ Device access: LIMITED DRI DEVICE ACCESS ($DRI_ACCESS_COUNT/$DRI_DEVICE_COUNT)"
|
||||
echo " VAAPI hardware acceleration may not work properly."
|
||||
fi
|
||||
fi
|
||||
|
||||
# Display FFmpeg VAAPI acceleration method
|
||||
if echo "$HWACCEL" | grep -q "vaapi"; then
|
||||
echo "✅ FFmpeg VAAPI acceleration: AVAILABLE"
|
||||
else
|
||||
echo "⚠️ FFmpeg VAAPI acceleration: NOT DETECTED"
|
||||
fi
|
||||
else
|
||||
echo "❌ NO GPU ACCELERATION DETECTED"
|
||||
echo "⚠️ Hardware acceleration is unavailable or misconfigured"
|
||||
fi
|
||||
|
||||
echo "📋 =================================================="
|
||||
echo "✅ GPU detection script complete."
|
||||
Loading…
Add table
Add a link
Reference in a new issue