What have been changed in 2015.4 XPE and Report Power for UltraScale and UltraScale+ devices?
The power model and related user entry criteria have been changed for UltraScale and UltraScale+ devices in 2015.4 XPE and Vivado Report Power.
These changes were made in an effort to improve the overall accuracy of the tools.
The following three models were updated:
Logic & Signal Modeling:
Increase in analysis accuracy is based on physical routing characteristics of the design, which impacts the dynamic power.
The "Routing Complexity" field replaces the "Logical Fanout" field. See (Xilinx Answer 66038) for more details.
You will see an increase in the estimated power by ~1-2%, but results depend on the specific situation.
Increase in analysis accuracy is based upon input toggle rate rather than output toggle-rates.
You will need to change the toggle-rates to reflect the inputs, instead of the outputs. The input toggle-rates are usually lower than the output toggle-rates.
You will see a small reduction in estimated power, but results depend on the specific situation.
Increase in analysis accuracy is based on improvements of the modeling to reflect low enable-rate situations and clock power contributions.
The toggle-rates now reflect an average of input and output toggle-rates. You need to change the toggle-rates to reflect an average of input and output toggle-rates.
You will see an increase in estimated power by 1-2%, but results depend on the specific situation.
For customers needing more details about these changes a PDF has been created to provide detailed information about the changes and how they impact the design estimation compared to prior versions.
The PDF is attached to this answer record. If you still have questions, please contact your FAE or open a Service Request.
|Name||File Size||File Type|