May 31, 2022
Editor’s Note: This content is contributed Snehal Ullagaddi, Product Marketing Manager for Vivado ML at AMD
With the new Vivado® ML Editions 2022.1 release, we have introduced some major improvements and enhancements to the tool set.
The Vivado Design Suite 2022.1 release provides:
Let's look at some of the major highlights.
With the current release, we have pushed the envelope and brought in improvements for Versal® devices. Work has been done extensively at every step of the implementation stage.
The image below shows the changes made to its corresponding stage on the left. The average improvement that we see is about 5% for default strategies and about 8% for stacked silicon interconnect (SSI) technology based devices. Usually, an average of 2–3% geomean improvements are good. But this time, we are seeing between 5–8 % improvement.
With 2022.1, we introduced a new feature for estimating resources prior to the synthesis stage, called ML-based resource estimation. This feature provides instant resource estimation data for all IPs, as well as IPs within a block design.
For Versal devices, we now have intelligent design runs and ML strategy runs. All timing closure features are available from a single right-click on a failed design run. We no longer require the user to open the run and take further steps; ML strategy runs are automatically created. The user no longer needs to create these runs manually from the report’s QoR suggestions.