Functional Coverage At Subsystem or SOC Level

[Thirumalesh]

HI All,

At subsystem or SOC verification level,
let’s say there are three IPs integrated : IP1, IP2 and IP3.

Individually from block level verification, all IPs functional coverage is achieved 100%.
IP1 FC 100%
IP2 FC 100%
IP3 FC 100%

But at the subsystem level, when I run some tests or regression, functional coverage is still 60 to 70%. What would be the understanding here?

case_1 : We need to still add tests to make 100% coverage even at SOC level
case_2 : Getting coverage (code or FC) 70% at SOC level is still acceptable, no need to wait for 100% coverage since individually all IPs are already verified.

So, Now I need to sign-off. Can I proceed with just 70% coverage at SOC level,
requesting everyone to share thoughts here.
I am interesting in measuring sign off factors at SOC or sub system level.

Kindly provide your inputs.

Similar query :

Motive behind functional coverage at SOC level and motive behind code coverage at SOC level

Could you please clarify the following. at the SOC level verification we need to see whether pins are exercised. for this toggle coverage of pins at the SOC level is enough. please let me know is there any other motive is there behind toggle coverage at SOC level. or any other code coverage is required. and why we need functional coverage at the SoC level. can we reuse the code coverage at IP level,at the SoC level.similarly can we reuse functional coverage at IP level at the SoC level.

[Surya_prakash.Kapireddy]

Hi,
As of my knowledge 100% IP-level coverage doesn’t guarantee the correctness of subsystem. 70% coverage is insufficient for signoff. you might missed some integration bugs. if you should Sign-off at the SoC level, you should have 100% functional coverage, code coverage and assertions.

[chr_sue]

The question is how did you specify the SoC level functional coverage?. And you should have in mind even reaching 100% IP-level coverage might be wrong because you have poor coverage model.