Article ID: 000083673 Content Type: Troubleshooting Last Reviewed: 09/11/2012

When I turn on the LeonardoSpectrum™ Extended Optimization option, why does the software select the pass with the smallest area when I chose to optimize for delay, or the shortest delay when I chose to optimize for area?

Environment

BUILT IN - ARTICLE INTRO SECOND COMPONENT
Description The LeonardoSpectrum software uses the following rules to determine which of the four extended optimization passes should be selected for a given design block:

    If a target frequency is specified and the design meets the timing requirements (according to the timing estimates in the LeonardoSpectrum software), then:

    • The software will choose the smallest design implementation regardless of wether you specify to optimize for area or delay. Figure 1 shows an example where the frequency equals 100 MHz/10 ns.


    Figure 1. Compilation Results
    PassAreaDelay
    Pass 11009
    Pass 29011
    Pass 31108

    For both area and delay settings, Pass 1 will be chosen since it is the smallest pass that meets the timing requirements.

    If a target frequency has not been specified or the design does not meet the specified target frequency requirements, then:

    • The software will choose the pass with the smallest design for an area setting, and the software will choose the pass with the fastest design for a delay setting.
    • However, there are instances where the software may not adhere to this second rule, if one pass has a design implementation which is considerably smaller or faster. For example, if area is selected and the smallest implementation pass has a very high delay, a different pass may be chosen with slightly more area but a much better delay. The LeonardoSpectrum software uses its own cost function to determine which pass should be chosen in these cases.

    Related Products

    This article applies to 1 products

    Intel® Programmable Devices

    1