Difference between Uncertanity & OCV

Can anyone tell the basic definition of Uncertanity & OCV?
What is the difference between them?
Why they play an important role?

If you've any documents on these topics, please upload it here.

Thanks in advance.

Hi, you can search this paper for detail:
"My Head Hurts My Timing Stinks and I dont Love On-Chip Variation"

Hi Kumar,

Uncertainity = skew + jitter
Skew: Diffrence between the arrival times of the clock at the flops....Spatial variatio(as Rabaey puts it)
Jitter: Variation in clock period for each cycle...inherent feature of an Oscillator as there is bound to be small variation in the clock period generated by an Oscillator in consecutive cycles....temporal variation(quoting Rabaey)...
Skew will effect the setup and hold checks of the design...as a positive skew will help in meeting setup and may cause hold to fail
as a negative skew will help in meeting hold and may cause setup failure

Jitter is the variation in clock period so it effectively reduces the time available for datapath i.e the eq for setup check will be datapath delay < clockpath delay - 2*jitter
Jitter doesnt have an effect on hold as hold check is done on the same clock edge.

OCV is Onchip variation...OCV will result in better timing(WNS and TNS) with minimal impact on the clock skew and insertion at the cost of increased runtime and power. What happens in OCV is it allows the tool to use both minimum and maximum delays to apply to different paths at the same time.

For a setup check it takes up max delays on launching flop and datapath and min delays on capture flop
For a hold check it takes up min delays on launching flop and datapath and min delays on capture flop.

So as you can see it is contrary to normal analysis wherein we can specify whetther to use max delays or min delays in a specified path.

For setup check it takes up max delays on launch,data and capture
For hold check it takes up min delays on launch,data and capture.

When you perform OCV analysis you will get more accurate timing reports...

Hope its Useful...

The common factor among both is that they are unpredictable..what ever models we do are just approximations...i agree with phoenixpavan on uncertainity but OCV can hamper the performance of the chip. OCV can happen because of process variation also. Some part of the chip may be at higher temp. than the other or may be at higher voltage than the other....OCV occurs because of the manufacturing process and not because of design.