Dedicated video processing architecture featuring AV1 high-speed pipeline delivers 32x 1080p streams per card with AI-enabled video quality optimization
SANTA CLARA, Calif., April 6, 2023 (GLOBE NEWSWIRE) — AMD (NASDAQ: AMD) today is purpose-built to drive the AV1 compression standard and a new era of live interactive streaming services at scale.Over 70% of the global video market is dominated by live content1A new class of low-latency, high-volume interactive streaming applications is emerging, including watch parties, live shopping, online auctions, and social streaming.
The Alveo MA35D Media Accelerator delivers high channel density, power efficiency and ultra-low latency performance of up to 32x 1080p60 streams per card, the surge needed today to scale such compute-intensive content delivery. reduce infrastructure costs. Compared to the previous generation Alveo U30 media accelerator, the Alveo MA35D offers up to four times the channel density.2up to 4x lower latency at 4K3 and 1.8x compression efficiencyFour Achieve the same VMAF score (a common video quality index).
Dan Gibbons, General Manager of AMD’s AECG Data Center Group, said: “Users and content he ASIC tailored to meet the bespoke needs of these providers in order to reduce both the capital and operational costs of delivering immersive experiences to creators at scale. With architecture he developed the Alveo MA35D.”
dedicated video processing unit
Alveo MA35D uses a dedicated VPU to accelerate the entire video pipeline. Performing all video processing functions on the VPU minimizes data movement between CPUs and accelerators, reduces overall latency, and channels at up to 32x 1080p60, 8x 4Kp60, or 4x 8Kp30 streams per card Density is maximized. The platform offers ultra-low latency support for mainstream H.264 and H.265 codecs and a next-generation AV1 transcoder that reduces bitrates by up to 52% to save bandwidth compared to comparable software implementations It has an engine.Five.
“AMD’s announcement of the new Alveo MA35D add-in card is a breakthrough in video acceleration for the data center, offering royalty-free high-definition video devices, products, and Alliance for Open Media Chairman Matt Frost to: “Live streaming providers are looking for high-density, low-power, low-latency AV1 solutions, and addressing these will help alliance members such as AMD to improve AV1 deployments and overall We help drive adoption.”
AI-enabled intelligent video pipeline
Accelerators feature integrated AI processors and dedicated video quality engines designed to reduce bandwidth and improve the quality of experience. An AI processor evaluates content frame by frame and dynamically adjusts encoder settings to improve visual quality while minimizing bitrate. Optimization techniques include region-of-interest (ROI) encoding for text and face resolution, artifact detection for correcting scenes with high levels of motion and complexity, and predictive insight for bitrate optimization. Contains content-aware encoding for
Scale your interactive media cost-effectively
Scaling a high-volume streaming service requires maximizing the number of channels per server while minimizing power and bandwidth per stream.By delivering up to 32x 1080p60 streams per card at 1 Watt per stream6a 1U rack server equipped with 8 cards provides up to 256 channels, maximizing the number of streams per server, rack or data center.
Availability of software development kits and products
The platform is accessible with the AMD Media Acceleration Software Development Kit (SDK) that supports the widely used FFmpeg and Gstreamer video frameworks for easy development.
The Alveo MA35D Media Accelerator is currently sampling and production is expected in the third quarter. To accelerate development, we offer an early access program to qualified customers, providing comprehensive documentation and software tools for architectural exploration.
For more than 50 years, AMD has driven innovation in high-performance computing, graphics and visualization technology. Billions of people around the world, leading Fortune 500 companies, and cutting-edge scientific research institutions rely on AMD technology every day to improve the way they live, work, and play. AMD employees are focused on building high-performing, adaptable products that demonstrate leadership by pushing the boundaries of what’s possible. For more information about how AMD enables today and inspires tomorrow, visit the AMD (NASDAQ: AMD) website, blog, LinkedIn and twitter page.
1 Source: Bluewave Consulting and Research, March 2022
2 According to the published specs, the Alveo MA35D supports up to 32 1080p60 streams and the Alveo U30 supports up to 8. The channel density ratio remains the same regardless of resolution. ALV-002
3 According to the published specs, the Alveo MA35D delivers a quarter of the latency in 8ms, while the Alveo U30 delivers 4K H.264 in 32ms (based on each platform’s minimum latency capabilities). ALV-005
Four Based on testing conducted by AMD Labs in April 2023, VMAF scores for Alveo MA35D AV1 encode compared to Alveo U30 H.264 encode across (13) published video files at various resolutions and bitrates. bottom. Actual results may vary. ALV-009
Five Based on tests conducted by AMD Labs in March 2023, Alveo MA35D H.264-encode, H.265-encode, and AV1-encode VMAF scores for (13) open source x264 ultra-fast SW across published videos It was compared with the model’s VMAF score. Files of various resolutions and bitrates. Actual results may vary. ALV-006
6 Typical power for 8 4K streams or 32 1080p60 streams is based on preliminary testing and is subject to change. 50W total thermal design power (TDP)
Source: Advanced Micro Devices, Inc.
Released April 6, 2023