REGISTER | LOGIN
Breaking News
Blog

RTL Power Estimation: The Pursuit of Perfection

NO RATINGS
View Comments: Newest First | Oldest First | Threaded View
leoeltipo
User Rank
Author
Re: Full agree...
leoeltipo   12/19/2014 7:40:34 AM
NO RATINGS
Hi,


Thanks!! No problem for the delay. Interesting topic this of power estimation, even more for the upcoming IoT market where power is going to be THE main issue for some particular applications.

 

Regards,

Leo

ccorbj
User Rank
Author
Re: Full agree...
ccorbj   12/18/2014 3:46:06 PM
NO RATINGS
@leoeltipo - thanks for the comment. Sorry it took me so long to reply. Somehow I don't seem to have got an email notification on your comment

leoeltipo
User Rank
Author
Full agree...
leoeltipo   11/21/2014 2:20:21 AM
NO RATINGS
I have some experience doing power estimation at different levels of abstraction. For sure, signoff netlist with parasitics and a realistic activity file is way the best golden-model.

Some tools, however, mimic physical back end and improve power results. Some of the problems that I observed that are not easily taken into account in RTL estimations are all the different fixes that are only present during back end: DRV, hold, buffering... things difficult to be taken into account in RTL.

I think that RTL can be close to reality for non very aggresive floorplans, where these problems are less important.

Anyway, there is no magic, out of RTL you have an infinite number of possible implementations based on your specific constraints, and the power estimation tool cannot guess the one you want without significantly asking you for extra information.

 

Very interesting post.

 

Regards,

Leandro

Most Recent Comments
Like Us on Facebook
EE Times on Twitter
EE Times Twitter Feed