Release Notes

Current Release

Version 6.3

Release Notes

Feature Updates Utilizes Vitis, Vivado, and XRT tools with v2026.1 The VEK280 build platform is migrated to the Yocto EDF flow. The VART-X source code is bundled to a wheel file and it will be downloaded automatically during the platform build. Yolox-m postprocessing is integrated into x_plus_ml_app in C/C++ (libpostprocess_yolo-1.0.so). Runtime SW supports any batch size. However, the optimal runtime batch size is the batch size determined at compilation. Known Issues/Limitation The setup_yocto_host.sh script is used to set up the required packages on host machine for Yocto EDF build and it is supported on Ubuntu 20.04 and 22.04 only. The first-time Yocto EDF build will be slow because the repo sync + full bitbake build can take several hours on a fresh machine without shared sstate/downloads mirrors. The subsequent incremental builds will be fast. keras framework is not supported. If keras application is used for inference, the recommended flow is to export the model into .onnx and to use ONNXRuntime framework to perform the quantization and compilation. AI Analyzer graphical results report inaccurate NPU clock frequency and inference time.

Previous releases

Version 6.2

Release Notes
Feature Updates Supports VE2802_NPU_IP_O00_A304_M3_generic NPU IP in addition to the previously supported NPU IPs in Version 6.1. Known Issues/Limitation A performance gap exists between single instance execution to multi-instance execution. SSDResnet34 snapshots have accuracy issues. AI Analyzer graphical results report inaccurate NPU clock frequency and inference time.

Version 6.1

Release Notes

Feature Updates Utilizes Vitis, Vivado, XRT, and Petalinux tools with v2025.2 Migrates to the SDT BSP flow (previous releases used the XSCT BSP flow) Fully accelerates YOLOv8, YOLOv9, and SSDResNet34 models on AIE when Mixed Precision is enabled Delivers performance improvements across supported workloads: Includes an example performance result for ResNet50: more than 9100 FPS on AIE and up to 8700 FPS at the software level. Adds support for INT8, BF16, and Mixed Precision snapshots in X+ML application. Updates the pre-processing IP in the reference design to support BF16 outputs. Adds support for PL-Stream based YOLO tail acceleration Adds support for Monochrome and RGB inputs Adds support for network performance (AI) analyzer Adds support for AMD GPU in Docker Known Issues/Limitation Models without the batch Size in the input or output shape are not supported. The batchSize (being 1 or more) is to be expected in the highest dimension of the tensor. A performance gap exists between single instance execution to multi-instance execution. SSDResnet34 snapshots have accuracy issues. AI Analyzer graphical results report inaccurate NPU clock frequency and inference time.

Note

• In X + ML, X refers to the hardware-accelerated pre-processing task and ML refers to the inference task running on the NPU.

• Removed YOLOX PL Tail support from reference design and application (as the Vitis AI tool supports mixed precision and achieves good performance).

Version 5.1

Release Notes

Feature Updates Utilizes Vitis, and Vivado with v2025.1.1 , XRT, and Petalinux with v2025.1. Reference design supports integration of two NPU IPs (Multiple NPU instances) feature. Supports another 38-column NPU IP that can be used for a design with NPU connected on either one or two DDRs (used in interleaving mode). Supports mixed precision implementation on YOLO models for better performance and accuracy trade-off. New Layers Support DepthtoSpace, PixelShuffle, Deconvolution, Up-conv, Transpose-conv (2D version) Only square boxes are supported (n, n) with n multiple of 2 Input channels must be a multiple of 16 × n × n Depthwise support upgradation on VE2802 to provide better performance Applications renamed ‘npu_runner_demo’ to ‘vart_ml_demo’ ‘npu_runner.py’ to ‘vart_ml_runner.py’ Updated the End-to-End (X+ML) application to support parallel ML pipelines using multi-threading. Known Issues/Limitation Batch size is limited when tail is accelerated on AIE; Validated only batchSize=1. Some models may not work on a smaller IPs (for instance the one used by the Multiple Instances of NPU design) or in MIXED or BF16 precision due to the layers not fitting in the memory tiles. The vart_ml_demo application prints incorrect prediction scores. End-to-End (X+ML) application supports INT8 snapshots (Yolo models using Mixed Precision do not support pre-processing in Programmable Logic). The X+ML application does not support multiple processes.

For documentation on the previous releases, visit the Vitis AI Lounge.

Version 2025.1

Release Notes

New Features Utilizes Vitis, Vivado, Petalinux, and XRT Tools v2025.1 Supports VE2202 and VE2602 Performant IPs, along with VE2802 and VE2302 Removes support for functional IP and related content Fixes bugs in corner cases | Supports all batch sizes / various input sizes | No new layer support BF16 support for YOLOv5, YOLOv7, YOLOv8, YOLOX. See Supported Models for other CNNs compatible with BF16 Improved performance/efficiency for SSD-ResNet34, YOLOv5, and YOLOX Supports NHWC native format for input and output Introduces Apptainer as an alternative to Docker Non-service container; no root required (better security/usability) Addresses Docker-related issues; future releases may phase out Docker Accelerates input pre-processing within AIE accelerated graphs Supports acceleration of YOLO tail graph on AIE Redesigned performance summary table for comprehensive details

Version 2024.2

Release Notes

New Features Supports PyTorch v1.12, TensorFlow v2.9.0, ONNX Runtime v1.20.1 Enables zero-copy with C++ and Python APIs; supported in End-to-End (X+ML) app as well NPU IP (VE2802 – performance): Three variants (38/24/16 columns, INT8) and one variant (38 columns, BF16) Officially supports and verifies Aurora’s VoVNet and ResNet18 models NPU IP (VE2802 – functional): One 7-column NPU IP NPU IP (VE2302 – performance): Full-columns and half-columns variants Updated NPU IP names Utilizes Vitis, Vivado, Petalinux, and XRT Tools v2024.2 Tool for ONNX-to-PL acceleration for YOLO tail graphs; quick-start updated for YOLOX tail Broader model coverage. See Supported Models