Magic Blue Smoke


New Problem in the Block

Hi, my name is Godwin Maben; I have been working with customers in trying to understand their low power concerns since last 6 years. Interesting thing to note is that low power concerns have become more of a main stream problem as opposed to secondary in the last 6/12 months. An interesting thought occurred to me, why not create a community, where in we all can exchange some ideas to solve low power problems and educate/share knowledge about various techniques used today and their challenges through out the flow.

Hopefully it’s going to be very a useful media to exchange ideas as well as creating a low power community to solve/exchange design/verification problem related to low power.The question that needs to be answered is How come all of a sudden Power has become a major concern and is something which cannot be overlooked nor ignored.

I would put them in the following 2 categories.

  •  Process Nodes:90/65/45nm (Shrinking Device size)
  •  Yield Loss due to increasing Process Variation (includes impact on Vt)

Before we get into too many details about why/what/how/when, let’s try to understand few basic questions about power

Here is a link, which gives insight into some of the basic concepts for Power

Now let’s look at the factors influencing these components of Power.

How can we reduce Dynamic Power? Is everything under designer’s control?

 Well, as implied by the equation 0.5C*V*V*F

(a) Reduce C
(b) Reduce V
(c) Reduce Frequency/Activities

I think all the above is definitely under designer’s control and can be tackled using different techniques.

How can we reduce Leakage Power?

This is a tricky one as it lot depends on the process; Designer’s can try minimizing it, but not completely controlling it.

 Graph Below shows the ITRS data on leakage power trends.


Let’s talk/discuss about these topics in my next session.