Why is the timing-report difference between Astro and PT

[blueant]

In one my case, timing check is met in Astro, but have violations in PT.
Anyone can tell me why is the same instance's inc-time close to twice in PT's timint-report than Astro's ?
How can I do to approach the timing-reportbetween Astro and PT?
Thank you veryvery much!

 

[spauls]

AStro works on mesh based analysis and PT is on Arnoldi ROm (Reduced order module) calculations. PT is more accurate than Astro/DC/PC

 

[blueant]

The timing is MET in Astro, but VIOLATE in PT.
How can I do?
Help me! Thanks!

 

[lnc]

You using two tools for your design ?
If you choose PT for test timing before tap-out your project and using another for physical synthesis . You can increase margin when you using your design tool . => you will work harder with new margin . Design meet timing when slack >= 0ps , but using PT slack >=-5ps (assume) => you should design slack >= 5ps => you will meet timing in PT

 

[aravind]

hi spauls
can u explain more on How PT engine works when compare with with cadence/synopsys astro engine.
If u have any document related is good.
please share how PT engine work compare with others

 

[joe2moon]

Why ?
-- Timing engines in Astro and PrimeTime are different.
----------------------------------------------------------------------
How ?
-- If you treat PrimeTime as the sign-off STA tool, then you may need to fix those violations in Astro based on PrimeTime's violation report.
-----------------------------------------------------------------------

As long as the timing engines in implementation tool and sign-off tool are different, then the correlation issue is inevitable.

Over-constrain does not always work.
------------------------------------------------------------------------

 

[blueant]

Thanks everybady!
I'll try it again.

 

[shankarmit]

In Pt u will import Spef also which has wire delay..
and pt report is the actual timing report

Regards
Shankar

 

[aravind]

once again asking friends
Anybody having PT engine and CTgen works?
If any having comparison chart that good to explore more