Vitis™ AI is a Xilinx® development kit for AI inference on Xilinx hardware platforms. Inference in machine learning is computation-intensive and requires high memory bandwidth to meet the low-latency and high-throughput requirements of various applications.
The Vitis AI Optimizer provides the ability to optimize neural network models. Currently, the Vitis AI Optimizer includes only one tool called the pruner. The Vitis AI Optimizer removes redundant kernels in neural networks thereby reducing the overall computational cost for inference. The pruned models generated by the Vitis AI Optimizer are then quantized by the Vitis AI quantizer and deployed to Xilinx FPGA, SoC, or ACAP devices. For more information on the Vitis AI quantizer and deployment, see the Vitis AI User Guide (UG1414).
The Vitis AI Optimizer supports TensorFlow and PyTorch. The corresponding tool names (_p_ denotes pruning) are listed in the following table:
| Framework | Tool Name |
|---|---|
| TensorFlow | vai_p_tensorflow (TF1.15), vai_p_tensorflow2 (TF2.x) |
| PyTorch | vai_p_pytorch |
Licensing Information
The use of the Vitis AI Optimizer requires purchasing a license for the tool. You can request a quotation for either the node-locked (part# EF-AI-OPTIMIZER-NL) or floating license (part# EF-AI-OPTIMIZER-FL) by contacting your local Xilinx Distributor or Sales Office. This is a perpetual license with no annual maintenance or renewal costs.
Should you wish to evaluate the AI Optimizer prior to considering purchase, you can request access by emailing xilinx_ai_optimizer@amd.com, or request access to the AI Optimizer Lounge.