On-Prem
Run high-density video encoding/decoding/transcoding inside your own data center using Quadra VPUs without changing ingest paths, codecs, or downstream workflows.
Use this architecture when:
- Regulatory, security, or cost constraints require local processing
- CPU/GPU encoding limits density or power efficiency
- Deterministic performance at sustained throughput is required
This architecture is optimized for control, predictability, and maximum encoding efficiency per server.
What changes
- Video compute shifts from CPU/GPU to VPUs
- Encoding density per server increases significantly
- Power, cooling, and rack footprint per stream decrease
What doesn’t
- Ingest sources and output destinations
- FFmpeg/GStreamer-based workflows and codecs
- Network topology, storage systems, or security posture
VPU placement
- Quadra VPUs reside inside dedicated on-prem encoding servers
- VPUs handle encode/decode/transcode only
- CPU resources are freed for orchestration and control
Scaling model
- Scale by adding VPU-enabled servers
- Predictable throughput per server
- Linear capacity growth without performance degradation
Prerequisites
- Quadra-enabled on-prem servers
- Existing ingest and output endpoints
- Orchestration layer (Bitstreams or existing scheduler)
- Standard FFmpeg/GStreamer or compatible encoding pipelines
Validation path
- Deploy a single Quadra-enabled server
- Benchmark against CPU/GPU systems
- Measure streams per rack unit, power, and cost per stream
- Expand incrementally across facilities
What this is not
- Not a SaaS platform
- Not a general-purpose accelerator
- Not a proprietary pipeline
- Not a rip-and-replace migration
Outcome
Higher encoding density. Lower cost per stream. Predictable performance using your existing infrastructure.
Supported by the VPU Ecosystem, partners operating this architecture in production today.