Xilinx Video SDK Release Notes¶

Note

Version: 1.5 (Production)

Note

A patch is required to use version 1.5 of the Xilinx Video SDK on AWS VT1 instances running the latest version of the U30 firmware. See instructions here.

Table of Contents

What’s New
Supported Platforms and Operating Systems
Limitations
Known Issues

What’s New ¶

This version provides various performance and stability improvements, including:

Support for Amazon EC2 VT1 instances as well as Amazon EKS and ECS
Support for Amazon Linux 2
Improved density, especially when processing many low-resolution streams
Improved runtime for both single and multi-process execution models
Improved resiliency of XMA and XVBM libraries

Supported Platforms and Operating Systems ¶

The Xilinx Video SDK currently supports the following platforms:

Xilinx Alveo U30
Amazon EC2 VT1

The Xilinx Video SDK currently supports the following software tools:

FFmpeg 4.1

The Xilinx Video SDK has been tested and validated by Xilinx on the following operating environments:

Ubuntu 20.04 (Kernel 5.4)
Ubuntu 18.04 (Kernel 5.4)
RHEL 7.8 (Kernel 4.9.184)
Amazon Linux 2

The above list of validated operating environments is not intended to be exclusionary. If you choose to run the Xilinx Video SDK using a different operating environment and encounter issues, please file a GitHub issue.

Limitations ¶

The scaler is tuned for downscaling and expects non-increasing resolutions in an ABR ladder. Increasing resolutions between outputs will reduce video quality. For more details see Using FFmpeg for Video Scaling.
The Xilinx Video SDK is optimized for live video real-time transcoding. For example, even though a single Alveo U30 card has the capability of running two Ultra HD (4K) transcodes in real-time at 60 frames per second (2x4kp60), running a single 4k stream at 120 fps is not supported. Similarly, a single Alveo U30 card supports up to 8x1080p60 transcodes in real-time, but it does not support fewer streams running at 120fps or beyond.
When processing 4x 1080p60 streams, the scaler can only sustain a rate of 60fps for up to 5 outputs. Therefore, given 4 1080p60 input streams, real-time performance is only possible for up to 20 scaled outputs.
The maximum bit-rate tested is 35 Mb/s; if the bit-rate is higher, the encoder might not work properly.
ABR video transcode use cases have been tested with up to 5 output renditions per input for a maximum supported number of channels of 8 running real-time per card.
The lookahead feature is not supported when encoding streams to a 4k resolution.

Known Issues ¶

Potential “No U30 devices found” error when sourcing the setup.sh script on AWS VT1 instances¶

Description: Sourcing the setup.sh script on AWS VT1 gives a “No U30 devices found” error.
Solution: A patch is required to use version 1.5 of the Xilinx Video SDK on AWS VT1 instances running the latest version of the U30 firmware. See instructions here.

Potential “failed” error code when sourcing the setup.sh script¶

Description: If you source the setup script on a machine where the Xilinx Video SDK environment has already been initialized, you will see the “failed” status being reported.

Solution: This is not an indication of a problem. This status simply states that your environment is already set up and that the system is ready for use.:

-----Load xrm plugins-----
{
    "response": {
        "name": "loadXrmPlugins",
        "requestId": "1",
        "status": "failed",
        "data": {
            "failed": "plugin xrmU30DecPlugin (version 2) is already loaded"
        }
    }
}

---------------------------------------

Potential “Too many packets buffered for output stream 0:1” error when running FFmpeg¶

Description: Some FFmpeg jobs may fail with a error message indicating “Too many packets buffered for output stream 0:1.”. This is a known FFmpeg issue (https://trac.ffmpeg.org/ticket/6375) which is not specific to the Xilinx Video SDK.
Solution: Run FFmpeg with the -max_muxing_queue_size 9999 option

Multiple “Starting new cluster” messages when encoding to Matroska¶

Description: When encoding to .mkv, FFmpeg may have issues synchronizing video and audio data leading to messages of the following form: [matroska @ 0x563b3f675100] Starting new cluster
Solution: Pass -max_interleave_delta 0 on the FFmpeg command line

Streams with rotation metadata may cause a segfault while transcoding¶

Description: The hardware transcode pipeline can be unstable on a stream with rotation metadata
Solution: Use the FFmpeg -noautorotate option in cases where rotation metadata is present.

Generating a device status report with the xrmadm tool does not work on Amazon EC2 VT1 instances¶

Description: The xrmadm /opt/xilinx/xrm/test/list_cmd.json command fails with a Traceback error message.
Solution: None.

Reverting to the golden image may not work by default on some on-prem systems¶

Description: The xbmgmt program --revert-to-golden and xbmgmt flash --factory_reset command expects the golden image of the card to be flashed at a particular memory location. If this is not the case, the command will not work.
Solution: Relocate the golden image by following the instructions below.
1. To reflash the golden image, your system must be already configured with release 0.96.0 or newer. If this is not the case, contact Xilinx.
2. Set up your environment:
```
source /opt/xilinx/xrt/setup.sh
```
3. List the Management BDF of your devices:
```
sudo /opt/xilinx/xrt/bin/xbmgmt examine
```
4. For each device, run the following command to flash the golden image to the expected memory location:
```
sudo /opt/xilinx/xrt/bin/xbflash.qspi --qspips-flash --input /opt/xilinx/firmware/u30/gen3x4/base/data/BOOT_golden.BIN --offset 0x6000000 --bar-offset 0x10000 --card <BDF>
```
5. Once the golden image has been properly relocated, you can revert the card to the golden image by following the card recovery instructions.