Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

What are things that make timing closure hard

Status
Not open for further replies.

matrixofdynamism

Advanced Member level 2
Joined
Apr 17, 2011
Messages
565
Helped
24
Reputation
48
Reaction score
23
Trophy points
1,298
Activity points
7,369
What are things that make timing closure hard?

Our design will have a specified operating frequency and propagation delays which constitute cell delays and path delays between the cells. The idea of timing closure is to verify that no setup hold violations exist and also figure out what are the false paths in the design which must be ignored in the timing analysis.

To this end, what are common difficulties encountered in timing closure and common mistakes? Any examples would be quite benficial.
 

sharath666

Advanced Member level 2
Joined
Apr 4, 2011
Messages
552
Helped
126
Reputation
252
Reaction score
124
Trophy points
1,323
Location
India
Activity points
3,830
As you have already mentioned failure to satisfy the setup and hold times is one of the consequences of not able to close timing..
The main problem in closing timing is the presence of long combinational paths in which the cell delays and the path delays add up to a value greater than the provided constraint..
 

matrixofdynamism

Advanced Member level 2
Joined
Apr 17, 2011
Messages
565
Helped
24
Reputation
48
Reaction score
23
Trophy points
1,298
Activity points
7,369
So the whole problem is that we want to breakdown the critical path to use less stages of logic or else, add a register within this huge combinational block to pipeline it. This way the critical path shall become shorter and the design can work at higher frequency.

That does not sound hard at all. Why then do people end up struggling to close timing? Why is it not piece of cake?
 

ads-ee

Super Moderator
Staff member
Joined
Sep 10, 2013
Messages
7,820
Helped
1,811
Reputation
3,632
Reaction score
1,772
Trophy points
1,393
Location
USA
Activity points
59,025
add a register within this huge combinational block to pipeline it. That does not sound hard at all. Why then do people end up struggling to close timing? Why is it not piece of cake?
It's not hard to add registers to a combinational block to pipeline it more. The problem then becomes, how do you fix the design with the added latency so that it functions correctly. In some cases that added latency breaks the design and there is no way to fix it without redesigning the architecture of the block.
 

sharath666

Advanced Member level 2
Joined
Apr 4, 2011
Messages
552
Helped
126
Reputation
252
Reaction score
124
Trophy points
1,323
Location
India
Activity points
3,830
That is because when you pipeline you are changing the functionality of the circuit. This may lead to signals getting generated earlier or later. This might impact some other part of the design due to which changes may be required in that part of the design also..
 

matrixofdynamism

Advanced Member level 2
Joined
Apr 17, 2011
Messages
565
Helped
24
Reputation
48
Reaction score
23
Trophy points
1,298
Activity points
7,369
What would go wrong if we instead run the design at a lower frequency so timing is met, rather than try to figure out how to reduce the delay in the critical path?

And by the way, why are hold time violations more rare than setup time violations?
 

ads-ee

Super Moderator
Staff member
Joined
Sep 10, 2013
Messages
7,820
Helped
1,811
Reputation
3,632
Reaction score
1,772
Trophy points
1,393
Location
USA
Activity points
59,025
What would go wrong if we instead run the design at a lower frequency so timing is met, rather than try to figure out how to reduce the delay in the critical path?
Your design might then not meet the upper level system requirements for through put or performance.

And by the way, why are hold time violations more rare than setup time violations?
They aren't, they are just easier to correct, see my other post on this.
 
Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Top