• Supports mainstream frameworks and the latest models capable of diverse deep learning tasks, CNN, RNN, and NLP
• Provides a comprehensive set of pre-optimized AI models that are ready to deploy on Xilinx devices. You can find the closest model and start re-training for your applications
• Provides a powerful open-source AI quantizer that supports pruned and unpruned model quantization, calibration, and fine-tuning.
• Provides a user-friendly compilation and deployment flow to meet customer-defined models and operators
• Provides layer by layer analysis to help with bottlenecks by AI profiler
• Offers the AI library with open-source high-level C++ and Python APIs for maximum portability from edge to cloud
• Provides a smooth solution by whole graphic optimizer (WeGO) to deploy Pytorch and Tensorflow models on cloud DPU by integrating Vitis AI within the frameworks
• Offers efficient, scalable, and customizable DPU IP cores, achieving different needs such as throughput, latency, power, and lower precision

AI model zoo is open to all users with rich and off-the-shelf deep learning models from the most popular frameworks such as Pytorch, Tensorflow, Tensorflow 2, and Caffe. AI model zoo provides optimized and retrainable AI models that enable faster deployment, performance acceleration, and productization on all Xilinx platforms. 

With world-leading model compression technology, the AI optimizer reduces model complexity by 5X to 50X with minimal accuracy impact. Deep compression takes the performance of your AI inference to the next level.

The AI quantizer can reduce the computing complexity without losing prediction accuracy by converting the 32-bit floating-point weights and activations to fixed-point like INT8. The fixed-point network model requires less memory bandwidth, thus providing faster speed and higher power efficiency than the floating-point model.

The AI compiler maps the AI model to a high-efficient instruction set and data flow. It also performs sophisticated optimizations such as layer fusion, instruction scheduling, and reuses on-chip memory as much as possible.

Click to Enlarge

The performance profiler allows programmers to perform in-depth analysis of the efficiency and utilization of the AI inference implementation.

The Vitis AI Library is a set of high-level libraries and APIs built for efficient AI inference with DPU cores. It is built based on the Vitis AI runtime (VART) with unified APIs and provides easy-to-use interfaces for the AI model deployment on Xilinx platforms.

The framework inference flow WeGO offers a straightforward path from training to inference by leveraging the native Pytorch and Tensorflow frameworks to deploy DPU un-supported operators to CPU, thus greatly speed up the models’ deployment and evaluation over cloud DPU.

Adaptable Domain Specific Architecture (DSA) to match fast evolving AI algorithms of CNN, RNN, and NLP with industrial leading performance on Xilinx Zynq® SoC, Zynq UltraScale+™ MPSoC, Alveo data center accelerator cards, and Versal® ACAP.