Accelerated Compute Instances are virtual machines optimized for transcoding. Transcoding is the process of converting digital media files from one format to another, which typically involves changing the file’s encoding, bit rate, resolution, or format to ensure compatibility with different devices, platforms, or bandwidth conditions.

Accelerated Compute Instances are backed by NETINT Quadra T1U video processing units (VPUs), which are ASIC-based media accelerator cards purpose-built for transcoding tasks. When compared to transcoding setups that rely on CPUs or GPUs, Accelerated Compute Instances may offer significant performance and cost advantages. To learn more, see Use cases.

Accelerated Compute Instance plans are ideal for media service providers engaged in video processing, including those that stream content directly to viewers over the internet.

Availability

During the Beta program, we are supporting approved customers who have identified workloads that will benefit from NETINT T1U VPU Accelerated plans.

Accelerated Compute Instances are currently available in the following regions:

• Chennai, IN
• Frankfurt 2, DE
• Los Angeles, US
• Miami, US

View these locations on the Akamai Connected Cloud map.

Plans and pricing

The following plans are offered at no cost during the Beta period:

Additionally, all network transfer associated with these plans will be free for the duration of the Beta period, or until further notice. Deployment of Accelerated plans won't impact existing network transfer pools.

Use cases

Over-the-top (OTT) media service providers (streaming platforms) offer content directly to viewers on the internet, bypassing cable and broadcast platforms. To ensure smooth, high-quality delivery of streamed content, video files are transcoded into various formats and resolutions to accommodate different devices and network conditions. Transcoding-optimized Accelerated Compute Instances may offer significant advantages over other solutions, especially in the following use cases.

Live Streaming

Accelerated Compute Instances can generally achieve greater density and handle more streams simultaneously than other solutions. This allows providers to use fewer hardware units. Accelerated Compute Instances also consume less power and generate less heat than general-purpose CPUs or GPUs, reducing energy and cooling costs.

NETINT Quadra TIU cards leverage hardware acceleration to handle video encoding and decoding tasks, which increases the efficiency of those operations. This translates into lower latency and higher throughput, which is essential for streaming and on-demand video.

Accelerated Compute Instances are also ideal for continuous stream handling. This is because they can optimize bitrate efficiency by dynamically adjusting encoding parameters to match network bandwidth, which reduces buffering and enhances the viewer experience.

Video on demand (VOD) Transcoding

On-demand transcoding typically involves processing large libraries of video content, where the focus is on throughput and quality rather than immediate real-time delivery. NETINT Quadra T1U cards support a variety of codecs, including AV1. They can handle multi-pass encoding and advanced compression techniques to ensure the final output meets quality standards.

These cards excel at batch processing, as they support simultaneous transcoding of multiple files without compromising quality. The ability to handle large volumes of content with high throughput and quality ensures that VOD providers can meet their audience's expectations.

Get started with Accelerated Compute Instances

Supported Operating Systems

When creating your NETINT Quadra T1U VPU-backed Accelerated Compute Instance, we recommend using one of the following distributions:

• Ubuntu 20.04
• Ubuntu 22.04
• Ubuntu 22.10

Configure your card with the Quadra SDK

To use a NETINT VPU, some initial configuration is necessary to ensure optimal performance. This includes installing the appropriate drivers, setting up the required software, and integrating tools like FFmpeg with NETINT’s libxcoder library.

To configure your card:

1. Install pre-requisites

   Shell

   sudo apt-get install nasm yasm -y
   

2. Download the latest release of the Quadra SDK and extract it to /root/ of your Accelerated Compute Instance.

3. Install the required software and firmware:

   1. Navigate to the folder containing the extracted the files and run the command

      bash quadra_quick_installer.sh

   2. Type 'Y' and press Enter to confirm that you want to use the release packages listed.

      A script menu appears.

   3. Complete each of the following by typing the corresponding number and pressing Enter:

      • Setup environment variables (1)
      • Install the prerequisite software libraries for your operating system (3)
      • Install the NVMe CLI tool (4)
        (You may receive an error message if you're running a newer version of Ubuntu. If so, ignore and continue to the next step.)
      • Install Libxcoder (5)
      • Install your preferred version of FFmpeg (7-15)

4. Type '22' and press Enter to exit the script menu.

NETINT Card Initialization

NETINT cards require initialization after a system reboot or virtual machine (VM) restart. Without proper initialization, the cards will not function as expected. We recommend applying a systemd process that will automatically initialize the cards during the boot process, eliminating the need for manual initialization and ensuring the cards are always ready for use.

1. Use the following command to create the systemd file:

sudo tee /etc/systemd/system/ni-vpu.service << EOF
[Unit]
Description=Run statup Process for the Netint VPU Devices
Before=docker.service

[Service]
Type=oneshot
RemainAfterExit=true
#Quardra Product Line
ExecStart=-/usr/local/bin/init_rsrc -t 5
StandardOutput=syslog
StandardError=syslog
SyslogIdentifier=ni_vpu_devices
User=nvme
Group=nvme


[Install]
WantedBy=multi-user.target
EOF

2. Enable the systemd process:

sudsystemctl able ni-vpu.service

3. Manually start the systemd process:

sudo systemctl tart ni-vpu.service

Support

For Support during the beta period, please contact your Account Representative or Open a Support ticket.