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.

Why Synopsys DC and PrimeTime timing analysis report is same!!!!!

Status
Not open for further replies.

gstekboy

Member level 5
Joined
Oct 18, 2013
Messages
87
Helped
1
Reputation
2
Reaction score
1
Trophy points
8
Activity points
512
I had done the timing analysis of a counter in both Synopsys DC and PrimeTime, but got the same output!! Any problem ??

Design file used is counter.v
The design compiler output is attached as DC.txt
The PrimeTime output is attached as PT.txt

Design compiler output is generated by giving input as counter.v , and clock period of 2.

PrimeTime output is generated by giving input as netlist file of counter , sdf file of counter (both generated from design compiler) and clock period of 2.

View attachment DC.txt View attachment PT.txt
 

ads-ee

Super Moderator
Staff member
Joined
Sep 10, 2013
Messages
7,854
Helped
1,815
Reputation
3,640
Reaction score
1,780
Trophy points
1,393
Location
USA
Activity points
59,349
I would kind of expect them to be the same...

1. DC synthesizes the design into a netlist and runs timing analysis on that netlist.
2. Primetime loads the netlist produced by DC and runs timing analysis on that netlist.

You didn't change the netlist, SDF, or the default wireload models. If they didn't match then I'd be worried.
 

gstekboy

Member level 5
Joined
Oct 18, 2013
Messages
87
Helped
1
Reputation
2
Reaction score
1
Trophy points
8
Activity points
512
I would kind of expect them to be the same...

1. DC synthesizes the design into a netlist and runs timing analysis on that netlist.
2. Primetime loads the netlist produced by DC and runs timing analysis on that netlist.

You didn't change the netlist, SDF, or the default wireload models. If they didn't match then I'd be worried.

  • I think DC actually calculates delay based on wireload models (RTL delay) not netlist delay.
  • I loaded both SDF file and netlist file for primetime and not in DC.
 

kirteshmiet

Member level 1
Joined
Feb 16, 2012
Messages
38
Helped
9
Reputation
18
Reaction score
9
Trophy points
1,288
Activity points
1,535
Hi gstekboy
As per my understanding both should match . As you are giving DC output netlist and sdf to PT and your constraints are also same . Since no routing or physical changes had not happened so it should not vary. Both will vary if there are any constraints mismatch or if any environment setting mismatch between them.
if you see between pnr tool and PT then PT will show you good results than pnr tool because pnr tool timing calculation is not as good as primetime. But in case of DC synthesis tool it should not vary much as it is using wireload models.

Hope it will help you.


Thanks
 

gstekboy

Member level 5
Joined
Oct 18, 2013
Messages
87
Helped
1
Reputation
2
Reaction score
1
Trophy points
8
Activity points
512
Hi gstekboy
As per my understanding both should match . As you are giving DC output netlist and sdf to PT and your constraints are also same . Since no routing or physical changes had not happened so it should not vary. Both will vary if there are any constraints mismatch or if any environment setting mismatch between them.
if you see between pnr tool and PT then PT will show you good results than pnr tool because pnr tool timing calculation is not as good as primetime. But in case of DC synthesis tool it should not vary much as it is using wireload models.

Hope it will help you.


Thanks

Both output are same , then how PrimeTime timing analysis will become more accurate than DC?
 

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
Actually in my experience the timing results for PT and DC will vary. This statement is made assuming that PT is not using SDF and uses the same wireload models as DC. I suppose they use different compute algorithms..
 

gstekboy

Member level 5
Joined
Oct 18, 2013
Messages
87
Helped
1
Reputation
2
Reaction score
1
Trophy points
8
Activity points
512
Actually in my experience the timing results for PT and DC will vary. This statement is made assuming that PT is not using SDF and uses the same wireload models as DC. I suppose they use different compute algorithms..

Yeah , I agree with you , but I got same result, why???
 

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
I suspect the reason should be that the module you are testing is pretty simple( a counter as I understand from your first post). Try it for a bigger and more complex block(e.g. an IP).
 

kirteshmiet

Member level 1
Joined
Feb 16, 2012
Messages
38
Helped
9
Reputation
18
Reaction score
9
Trophy points
1,288
Activity points
1,535
Both output are same , then how PrimeTime timing analysis will become more accurate than DC?

Hi gstekboy,
In this case then i would say as design is very small so you are not able to see the difference . try it for some bigger logic. As suggested by sarath:)

Thanks
 

gstekboy

Member level 5
Joined
Oct 18, 2013
Messages
87
Helped
1
Reputation
2
Reaction score
1
Trophy points
8
Activity points
512
I would kind of expect them to be the same...

1. DC synthesizes the design into a netlist and runs timing analysis on that netlist.
2. Primetime loads the netlist produced by DC and runs timing analysis on that netlist.

You didn't change the netlist, SDF, or the default wireload models. If they didn't match then I'd be worried.

I'm feeling some difficulty in your answer because,
  1. DC and PT use different compute algorithms.
  2. DC is based on RTL but PT is based on netlist and SDF.
    So how can I expect them to be same.
 

hoanglongroyal

Member level 1
Joined
Nov 24, 2012
Messages
36
Helped
20
Reputation
40
Reaction score
19
Trophy points
1,288
Activity points
1,480
I'm feeling some difficulty in your answer because,
  1. DC and PT use different compute algorithms.
  2. DC is based on RTL but PT is based on netlist and SDF.
    So how can I expect them to be same.

hi gstekboy,
I need to be cleared about your input database to PT: your netlist and SDF are exported from DC, aren't it ?
if yes, when you read them into PT, it just annotate SDF to netlist (what is DC already calculated) --> should be the same - I think.
to see difference, I think you should load netlist and SDC (which is exported from DC) into PT. then the tool will do compute, and may be bring different result.

P/S: in PT.txt , you can see character "*" --> that tell us that the value is annotated from SDF
 
  • Like
Reactions: gstekboy

    V

    Points: 2
    Helpful Answer Positive Rating

    gstekboy

    Points: 2
    Helpful Answer Positive Rating

gstekboy

Member level 5
Joined
Oct 18, 2013
Messages
87
Helped
1
Reputation
2
Reaction score
1
Trophy points
8
Activity points
512
hi gstekboy,
- I think.
to see difference, I think you should load netlist and SDC (which is exported from DC) into PT. then the tool will do compute, and may be bring different result.

P/S: in PT.txt , you can see character "*" --> that tell us that the value is annotated from SDF

So, to get accurate result, instead of SDF we must use SDC along with netlist. isn't it?
 

hoanglongroyal

Member level 1
Joined
Nov 24, 2012
Messages
36
Helped
20
Reputation
40
Reaction score
19
Trophy points
1,288
Activity points
1,480
So, to get accurate result, instead of SDF we must use SDC along with netlist. isn't it?

yes gstekboy,
- when you use SDF --> annotated, not calculate
- when you use SDC --> PT will calculate timing.
but I'm not sure about "accurate" :D
 
Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Top