Scan Tenant Number with airSlate SignNow

Scan assignee ssn

hello and welcome to mentor at itc my name is guyer ida and in this video i'm going to show you our new no compromise approach to dft the tescent streaming scan network first let's review some of the challenges we see in soc designs today the most common way of connecting core level scan channels to chip oil pins is through what we call a penmoxed approach where a mux network decides which cores that are directly connected to the chip level pins this is straightforward for a small design like this but when the number of course grows and when the complexity of the course grows and when we get multiple levels of course this becomes uh more and more complex and in sometimes sub-optimal and the reason for that is a lot of the constraints that we have to deal with for example how many cores you can test concurrently is determined by how many ios you have available for a scan test also which course that are tested concurrently or which course can be tested concurrently is a hardwired decision it's something you have to decide during the sign process there's also dependencies related to the number of channels at core level and that also again determines which course you can can test in parallel and now you can build in some more flexibility in this approach with a more complex mux network but that also increases the the potential for for routing congestion related to routing all of the scan channels all across the design now there are also different ways you can you can deal with this now you can implement what we call a bottom-up flow where you allocate a fixed number of scan channels like early in the flow let's say you decide to have the same number of scan channels for all of the course and kind of make a lot of decisions early that's the easiest approach the least effort but a side effect of that you might end up wasting bandwidth because the different cores that you group together end up having different scan chain length a different pattern count now you can address that then by reallocating the scan resources for instance allocating fewer scan channels to the cores that require less data that means reconfiguring compression rerouting the scan channels regenerating patterns and this is a lot of effort so you're dealing often with a you have to make a compromise between effort and test time furthermore we also see that the pinmax approach is not very effective when we deal with large numbers of identical cores you can broadcast scan inputs to multiple cores but typically have to observe all of the scan the scan outputs for all of the cores and there's also some challenges associated with multiple different pipeline stages different number of pipelines for different cores the throughput also is determined by the core shift speed so even though you have a tester...