The function setup violations and scan mode hold violations in Astro

Dear All

I'm p&r a design with scan chain.
First , I fixed all the timing violations in function mode,
then I switched to scan mode and fixed the timing .
But when I turned back to function mode, there were
setup violations .

It seems an iteration when design with function mode and scan mode.
Does anyone has a suggestion about this?
How can I made this 1-pass or less timing consuming ?
Any advice will help .

Thanks

Re: Timing closure issue

I'm still struggling between the function setup violations and scan mode hold violations in Astro . Although I can't offer a good solution, I'm glad to share
my useless experience : )

Following are ways I tried :

1. Used PPO2 to fix timing violations. But Astro(2005) doesn't handle multimode .
I'd rather fix timing by my own hand. I used PostPlace > Change Netlist to add and remove buffers manually . But I still got 7 setup violations in function mode which I can't fix .

2. First , I fix all timing violations in function mode. Then I use astSetDontTouch to all cells. And then I turn to Scan mode and fix hold time violations. Doesn't Worked.

3. I use "create_scenario" to create scenario for each SDC file. The I use PPO2 to fix timing. But it turn out to be useless either . It seems Astro can only handle multimode in sequence not concurrently .

4. There is a SNUG paper on this issue. "Hold Time Fixing for MultiMode Design".
You can find it in EDAboard.com.

If Astro can't handle multimode design, what should I do ??

Re: Timing closure issue

Hi leeguoxian,

This is a common problem in multi scenario design, but could you explain me why

2. First , I fix all timing violations in function mode. Then I use astSetDontTouch to all cells. And then I turn to Scan mode and fix hold time violations. Doesn't Worked.


does not worked.
I assume your scan frequency is less than functional mode & once you fix the setup and hold in functional mode, scan mode problem should converge mainly to hold fixes :|
May be very few setups are there. Which should be easy to fix manually by considering functional mode. If fixing these disturbs your functional mode hold than you need to consider reduced SCAN frequency or resynthesis with both the modes constrained properly

I do not know if your schedule permits that.

Re: Timing closure issue

i think detaching and attaching scan back can solve this problem, isnt it?
detach it while doing functional mode...and after CTS attach it back ....let me know on that.........


thanks,
Prasad