[SOLVED] Do we need to change the CTL model after Scan chain re-ordering?

[assud]

DFT: Do we need to change the CTL model after Scan chain re-ordering?

DFT Compiler provided ScanDEF and CTL model.
In P&R, scan chain reordering is done.
Can we still use the old CTL model for top level stitching? Do we need to change CTL model for top level stitching? Please help.

 

[kornukhin]

Not sure, but... You can check CTL content. If it doesn't contain information about cell order, then TMAX will not know that scan-chain reordered.

 

[assud]

Got clarification for the issue. Let me share it.
In HSS flow, CTL model will be used for top level scan stitching.
But for P&R scan re-ordering, P&R will be done for the entire design not in module level and hence we will be using SCANDEF file in back-end tools in design level not in module level. SCANDEF may be generated at block level depending on the flow selected. But this is to support chip level DFT insertion, not for P&R tool

 

[englishdogg]

@assud

Pls help understand what you mean by HSS flow?
My understanding tells me that CTL has more detailed info than a scandef - though both have its own usage. On elik eyou mentioned that SCANDEF can be used for block level or bottom up linking - that being said a CTL file can also be used. (?)

PnR tool not being able to read a CTL file may be a case as well to consider. But i usually look more in detail into a CTL file over a scandef file to get inputs or post process DFT info.

 

[assud]

ed,

Your understanding is correct. Let me provide you some more details:
HSS means hierarchichal scan synthesis or bottom-up synthesis flow. This makes use of the scan information of DFT inserted blocks as CTL model (core test language). For chip level scan integration, instead of netlist we can use the CTL models.
SCANDEF files are used for scan chain reordering by back-end tools. It consists of stub chains for reordering. In HSS flow by using CTL models, SCANDEF has subblock scan segments as black boxes. Here we would only be able to perform repartitioning considering segment as a whole. For reordering the cells in the segment, we can use '-expand_elements' option of 'write_scan_def' command.

 

[englishdogg]

So both can be used, but PnR would use Scandef. thank you very much.