Hardware Acceleration

Use your GPU for faster, more efficient transcoding.

Supported Hardware

Type	GPUs	Requires
VAAPI	Intel, AMD	/dev/dri device access
NVENC	NVIDIA	NVIDIA driver + runtime
QSV	Intel	Intel Media SDK
AMF	AMD	AMD drivers

Docker Setup

Intel / AMD (VAAPI)

Pass the DRI device to your container:

services:
  tvarr:
    image: ghcr.io/jmylchreest/tvarr:release
    devices:
      - /dev/dri:/dev/dri

Verify access:

docker exec tvarr ls -la /dev/dri
# Should show renderD128, card0, etc.

NVIDIA (NVENC)

Use the NVIDIA container runtime:

services:
  tvarr:
    image: ghcr.io/jmylchreest/tvarr:release
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu, video, compute]

Or with the legacy runtime flag:

services:
  tvarr:
    image: ghcr.io/jmylchreest/tvarr:release
    runtime: nvidia
    environment:
      - NVIDIA_VISIBLE_DEVICES=all
      - NVIDIA_DRIVER_CAPABILITIES=video,compute

Kubernetes Setup

Intel GPU

spec:
  containers:
    - name: tvarr
      resources:
        limits:
          gpu.intel.com/i915: 1
      volumeMounts:
        - name: dri
          mountPath: /dev/dri
  volumes:
    - name: dri
      hostPath:
        path: /dev/dri

NVIDIA GPU

spec:
  runtimeClassName: nvidia
  containers:
    - name: tvarr
      resources:
        limits:
          nvidia.com/gpu: 1

Verifying Hardware

Check detected hardware in the logs:

docker logs tvarr 2>&1 | grep -i "gpu\|vaapi\|nvenc\|qsv"

Or visit the Transcoders page to see detected capabilities.

Encoder Priority

When multiple encoders are available, tvarr prefers:

1. GPU hardware encoder (fastest, lowest CPU)
2. CPU software encoder (universal fallback)

You can override this per-profile with encoder overrides.

Session Limits

GPUs have encoding session limits:

GPU	Max Sessions
NVIDIA Consumer	3-5
NVIDIA Quadro/Pro	Unlimited
Intel (recent)	~20
AMD	Varies

tvarr auto-detects these limits and won't exceed them.

Troubleshooting

VAAPI Not Working

Check render device permissions:

# Inside container
ls -la /dev/dri/renderD128
# Should be readable/writable

# Check user group
id
# Should include 'video' or 'render' group

NVENC "Too Many Sessions"

You've hit the session limit. Options:

1. Use fewer concurrent streams
2. Use a Quadro/professional GPU
3. Apply the NVENC patch

"No Hardware Encoder Found"

Ensure:

1. Device is passed to container correctly
2. Drivers are installed on host
3. Container user has access to device