Stucking with sequence of sequences

UVM
, ,
1

Hi folks,

i’m got lost while build up my first sequence of sequences in combination with different configurations. My scenario looks like this:

Test 1: One sequence (SINGLE_SEQ)

  • Config object SINGLE_CONF contains two filenames A_SEQ for input stimuli and A_SCB for scoreboard.
  • Sequence SINGLE_SEQ fetches config object with filename A_SEQ from config db and fill some transactions items with data from the file
  • Scoreboard fetches config object as well and generates some expected transactions with respect to the file A_SCB

Test 2: Sequence of sequences (MULTI_SEQ)

  • A new config object MULTI_CONF contains a queue of SINGLE_CONF objects: ((A_SEQ,A_SCB), (B_SEQ,B_SCB), …)
  • Sequence MULTI_SEQ fetches the new config object from the config db and setup a SINGLE_SEQ for every entry in the queue
  • Sequence order can be random, for instance A->B->C or A->C->B

But how it is possible to pass the corresponding filename *_SCB to the scoreboard? Should I setup a connection between the sequencer and scoreboard?
In my former (non-UVM but even class based) system verilog testbench i instantiated a new environment object for every entry in the queue.

2

In reply to hbeck:

I solved my problem with an analysis port between sequencer and scoreboard. Looks quite complicated in comparison to my old testbench, especially if you want to explain this architecture to other colleagues.