Multiple agents in tetbench

Hi,

My Design is PIPE for serial interface which supports multiple lanes.So we have per lane interface signals in my design.

Which is the best way to develop UVM testbench,from the below two ways?

  1. single agent which drives and monitors for all lanes in parallel threads. ??
    OR 2) having separate agents for each lane ??

Please elaborate.

Thanks

In reply to chandu_jyothi:

Hi ,

Can Anyone clarify my question.

Thanks

In reply to chandu_jyothi:

In my understanding it is the best solution with respect to reuse to use 1 agent for each seoerate virtual interface, i.e. lane.