Updating IM list for SCE-MI 2.0

Brian,

 

I would like to suggest adding IMs for the following topics. Some topics
have been discussed and progress is made, others need to be resolved.
But I think we should track these issues and close them once the
resolution is reflected in an updated spec, reviewed and agreed to.

 

a.       Creating a Pipe object instead of using Pipe ID as an argument
to a function (Jason's proposal)

 

b.       Removing any calls to SV DPI utility functions and DPI exported
functions outside an imported context function

 

c.       Implementing implicit synchronization on EOM and the default
during bring-up mode.

 

d.       Defining clearly buffer length (HW only or HW SW)

 

e.       HW/SW side synchronization when buffer is smaller than a
transaction.

 

f.         Defining SCE-MI 2.0 terminology (transaction/message and
element including EOM/EOT).

 

g.       Synchronization semantics of FLUSH for non-blocking. 

 

h.       Rename pipes function calls with sceme as opposed to dpi
prefix.

 

i.         What happens if the consumer doesn't listen when the producer
issues a FLUSH or when its being blocked. For example what if monitor is
not listened to at the decision of the test. Will this scenario create a
deadlock?

 

j.         The scope of using pipes in their current form from Verilog
2001 and VHDL. Can pipes be called from Verilog 2001 and VHDL? 

 

Thanks,

 

Shabtay

 

 

-------------------------------------

 

Shabtay Matalon

Solution Architect

R&D, CVA

Phone: (408) 428 5081

email: shabtay@cadence.com