fbpx
Netint technologies

Quadra T1A

Video Processing Unit

Quadra T1A is an AIC form-factor video processing unit with one Codensity G5 ASIC. Operating in x86 and Arm-based servers. Quadra T1A enables video operators to move from software to hardware-based encoding to power real-time video, cloud gaming, and metaverse applications at a 40x less TCO than CPU-based solutions.

0 x

TCO Reduction

0 x

Less CO2 Emissions

Netint Codensity, ASIC-based Quadra T1A Video Processing Unit

Quadra T1A

Quadra T1
HOW IT WORKS

High-Density Transcoding Solution

Quadra T1A uses the Codensity G5 ASIC to support AV1, HEVC, and H.264 real-time video encoding up to 8K resolution with 10-bit HDR. 

The Quadra T1A Video Processing Unit plugs into any x86 or Arm server for a simple upgrade path to hardware video transcoding. 

The high throughput of the Quadra T1A enables ultra-low latency encoding of 16 broadcast quality 1080p60 streams using AV1, HEVC, or H.264.

ABOUT

Software Integration with FFmpeg Library Support.

NETINT Video Processing Units are simple to integrate with using FFmpeg. The Xcoder lib API allows existing video transcoding applications to achieve significant performance and capacity upgrades by switching to ASIC-powered VPUs.

Quadra T1A / T1U / T2 Video Processing Units

Ultra-High Density Video Encoding for x86 and Arm Servers

Realtime and ultra-low latency video platforms and services utilize NETINT Quadra T1A, T1U, and T2 Video Processing Units in their x86 and ARM-based servers to reduce operational costs by as much as 40X, and carbon emissions 80X as compared to CPU-powered software-based video transcoders. All while maintaining broadcast video quality.

Elegant, simple, effective

By utilizing the ubiquitous PCIe 2.5" U.2 and AIC form factors, Quadra VPUs offers an elegant, simple, yet effective hardware architecture for scaling video encoding and processing in the datacenter.

Quadra VPU Comparison

Quadra T1A, T1U, and T2 VPUs are built on the newest NETINT Codensity™ G5 ASIC that was designed for enterprise-class video transcoding operations capable of reaching 8K resolution at 60 fps using AV1, HEVC, and H.264.

The addition of 2 Deep Neural Network engines capable of 18 Trillion Operations per Second (TOPS), enables functionalities such as object detection, classification and segmentation, and ROI to provide additional data to the encoding engine for image quality improvement and content-adaptive rate control. 

To accommodate the widest range of use cases and applications, three Quadra VPU variants are available.

Applications for Quadra VPUs.

The advantage that silicon brings to modern video workflows is the vast reduction in power needed to operate the device.

Greatly enhanced energy efficiency directly translates to reduced carbon emissions and a reduction in operations costs making many interactive and advanced video applications economically viable. Additionally, the integration of high-scale video encoding with built-in AI processing opens premium video use cases and advanced applications.

Using a 1RU NVME server hosting ten (10) Quadra T1 VPUs, the combined system can achieve simultaneous encoding and transcoding of 160 broadcast quality 1080p60 live streams. For services needing 4Kp60, 8Kp60 and full adaptive bitrate ladders Quadra VPUs provide flexibility across every supported codec, AV1, HEVC, and H.264.

Quadra VPU Video Function Diagram
Quadra VPU Video Function Diagram

Leveraging FFmpeg, and a comprehensive suite of APIs, video engineers can easily integrate Quadra VPUs into their existing encoding infrastructure. Following is a functional block diagram that illustrates Quadra VPU video encoding and processing operations.

Quadra - T1A-T1U-T2 - Form Performance Codecs

Quadra T1 VPU Specifications

PHYSICAL

  • Hardware Interface AIC HH HL
  • Interface PCIe 4.0 x4, NVME
  • Protocol NVMe
  • Power Consumption 20W
  • Operation Temperature 0 – 50°C
  • RoHS Compliance European Union (EU) ROHS Compliance
  • Product Health Monitoring Self-Monitoring, Analysis

FEATURES

  • AI Deep Neural Network Engine 18 TOPS
  • Region of Interest (ROI) Enables region optimization
  • Closed Captioning EIA CEA-708
  • High Dynamic Range (HDR) HDR10, HDR10+, HLG (Encode & Decode)
  • IDR Insert Forced IDR frame inserts at any location
  • Flexable GOP Structure 8 presets & customizable GOP structure
  • Video 2D Processing Engine Crop, Padding/Scaling/Overlay, YUV/RGB Conv.

CODEC

Encode Codecs

AVC/H.264 Baseline, Main, High, High 10 HEVC/H.264 Main, High 10 AV1 Main

Decode Codecs

AVC/H.264 Baseline, Main, High, High 10 HEVC/H.264 Main, High 10 VP9 Profile 0,2

Level

1 to 6.2 Main-Tier

Resolution

32×32 to 8192×5120

Bitrate

64kbit/s to 700Mbit/s

Software Integration

FFmpeg SDK, LibXcoder API Integration

Capacity

1x 8Kp60, 4x 4Kp60, or 16x 1080p60

Applications

USE CASES AND ADVANCED APPLICATIONS

How Quadra VPUs Are Being Used

WebRTC Live Streaming

High Density Encoding Architecture | Real Time Encoding | Scalability | Future Proofing | HDR10+/8K ready

Cloud Mobile Gaming | AV | VR

Ultra-low latency network architecture | Low TCO | Scalable infrastructure | High Performance | High Res/VQ/Graphics .

Social Media Live Video Streaming

System Scale | Many Edge DCs | Network Bandwidth OPEX | Edge Logistics | Power/Footprint/ Space

Security

High Density Encoding Architecture | Faster than real time encoding | HDR/8K ready

Video Conferencing

Latency | Real Time Encoding | Scalability | Encoding Density

Ad Insertion

High Density Encoding Architecture | Real Time Encoding | Scalability | Future Proofing | HDR10+/8K ready