Announcement of SDK 1.2
We are happy to inform the introduction of the AMD Alveo Media Acceleration portfolio (AMA) SDK 1.2, which adds even more media accelerator capabilities to the AMD Alveo MA35D.
SDK version 1.2
With this update, the SDK can now accommodate:
The video pipeline from encoding a video to still jpeg pictures to decoding of still jpeg images executing more intricate ABR ladders that use many devices facilitating the implementation of ladders that can handle more encoding than a single device.
Video Caliber
The following enhances video quality:
- A fresh -caber encoding setting. The encoder will save bits on basic material in this mode by capping content at VMAF ~= 90.
- A sneak peek at a multimedia encoding option using machine learning that adapts bitrate depending on frame complexity.
OS Assistance
With increased OS compatibility, the SDK provides more host machine options, including:
- Fedora 40 support with the 6.9 kernel, offering another option for a supported operating system.
- A new firmware that supports PCI passthrough and boosts virtual machine flexibility.
Extra Information
Additional updates are included in the release, such as:
- A rework of the 2D pixel processing filters and an update to ffmpeg 6.1.1 release result in simpler, more user-friendly command lines.
- A redesign of card management tools that enhances their functionality and output.
AMD Alveo MA35D Media Accelerator
Key Features
Accelerator for New Media in High-Volume Streaming: AI-Powered Video Processing Scales Interactive Media at a Low Cost.
ASIC Design for Elevated Channel Density: Optimized for low latency, performance, power, and space per channel, supporting H.264, HEVC, and AV1.
Ideal Video Resolution at Lower Bandwidth: For the best possible experience, compression efficiency combined with AI processing and specialized VQ engines
Scaling Interactive Media Applications at a Low Cost: Scale emergent use cases, live events, cloud gaming, and video collaboration while lowering cost-per-channel. Hardware-Powered Video Processing for Interactive, High-Volume Streaming.
AMD Alveo MA35D Media Accelerator Unveilment: Their first ASIC-based media accelerator, the AMD Alveo MA35D Media Accelerator, provides AI-enabled video processing to help expand a new class of media services at a reasonable cost.
ASIC Structure
ASIC Design for Elevated Channel Density
An ASIC-based video processing unit designed from the bottom up for high-density, ultra-low latency streaming powers the AMD Alveo MA35D Media Accelerator. With four distinct encoder engines on each device (two per card), streaming providers can accommodate both new and older endpoints with flexibility and support several standards at the same time. Based on application needs, the design may be adjusted to balance latency, channel density, and compression efficiency. It is optimized for space, power, and cost-per-channel.
SDK for Advanced Media Acceleration
Kit for Software Development
For video professionals looking to smoothly incorporate the technology into their own development environment, the AMD Alveo MA35D comes with a full software stack. The popular video frameworks FFmpeg and Gstreamer are supported by the open-source Advanced Media Acceleration (AMA) SDK, which also offers a C-API for more video pipeline modification.
Start Now: With the AMD Alveo MA35D media accelerator card, get going right now. Available right now from approved wholesalers and AMD.
Media Accelerator Card Alveo MA35D: With its ultra-low latency, excellent video quality, compatibility for next-generation codecs, and ability to grow live streaming cost-effectively, the AMD Alveo MA35D transforms high-volume streaming.
Summary of Alveo MA35D Products: A new age of interactive media applications that scale for cost, space, and power per video channel is being ushered in by the AMD Alveo MA35D media accelerator.
Alveo MA35D Media Accelerator
Since live streaming has taken over the worldwide media business, a new category of Emerging low latency applications are changing the cost structure and related infrastructure and video processing technologies’ rollout approach.
A new age of interactive media is being ushered in with the AMD Alveo MA35D media accelerator. Programs that adjust for power, area, and cost per video channel. bolstering elevated Up to 32x 1080p60 streams of channel density, the accelerator is designed to minimize rack space, battery use, and network bandwidth all while preserving the quality of the video very low latency.
By assisting in the reduction of infrastructure, content, and CAPEX and OPEX Providers may grow profitably while maintaining the highest level of customer satisfaction. The AMD Alveo MA35D card, with its Half-Height, Half-Length (HHHL) PCIe form factor, offers full compatibility with video codec technologies, such as H.264, H.265, and the AV1 is a new standard.
Important Uses
Video processor for high-density, low-latency streaming
The AMD Alveo MA35D media accelerator was created from the ground up for interactive media application streaming. Its unique ASIC design provides the optimum economics in terms of cost, space, and power-per-stream. The accelerator is powered by two of these specially designed video processing units (VPU), which also include multiple video encoders, a hardened scaler to implement ABR ladders, a hardware compositor engine, an AI processor, video quality engines, and other features.
- Video Processing Unit for Ultra-Low Latency, High Density Streaming.
- Cost, area, and power per stream are all decreased via specialized ASIC design.
- Support for latencies as low as 8 ms (4 Kp60) for AV1, H.264, and H.265.
- Maximum resolution of 8K.
Through the acceleration of the whole video pipeline on a single device, the AMD Alveo MA35D platform maximizes channel density and minimizes chip-to-chip latency by decreasing data flow between CPU and accelerator.
The AMD Alveo MA35D accelerator is designed to meet the demanding latency needs of media applications and new use cases that provide gamification, customization, and monetization services like online wagering, live auctions, and live shopping. The card is designed such that, for a variety of endpoints, latency decreases linearly with resolution and format reductions.
Intelligent Video Pipeline Powered by AI
Each device on the platform has an AI processor that works as an encoder’s co-processor. The specialized engine analyses the material frame by frame. It dynamically modifies the settings of the encoding (codec) to improve the apparent visual quality while lowering bitrate. Techniques for optimization include:
- At ideal video quality, compression efficiency aids in bandwidth reduction.
- “Smart streaming” powered by AI improves experience quality at lower bitrates.
- Region-of-Interest (ROI) encoding: this technique optimizes certain video segments, such text and faces, to improve perceived visual quality at a low bitrate.
- Artifact Detection & Correction eliminates ringing, blurring, and “blockiness” from videos with a lot of motion or complexity.
- Content Aware Encoding (CAE) gives the encoder prediction insights from scene analysis to optimize bitrate.
Through the integration of AI into a pipeline that is entirely hardware accelerated, the AMD Alveo MA35D is able to execute intelligent optimizations at low latency and large scale. Because of its intrinsic customization capabilities, the processor can adapt to new AI methods and application cases.
Scaling Interactive Media Applications at a Reduced Cost
- Affordable Scale For Apps Using Interactive Media.
- Providers are able to grow economically for real-time, high-volume streaming.
- Provides a platform for a new type of interactive services and monetization.
It is necessary to maximize the number of channels per server while lowering the power and bandwidth consumption per stream in order to scale high-volume streaming applications profitably. With its thin, Half-Height, Half-Length (HHHL) PCIe form factor, the AMD Alveo MA35D media accelerator may optimize streams per server, rack, and data center by providing up to 32x 1080p streams per card for a 1U rack server.
The card helps lower power and bandwidth OPEX so infrastructure and content providers may grow financially. It does this by reducing bitrate by 52% when compared to a software solution and using 1 watt per stream.
