[Gems-users] L2_REQUEST_LATENCY & L2_RESPONSE_LATENCY in MOESI_CMP_directory

["soohong p kim" <spkim@xxxxxxxxxx>]

Hi,

 

I’ve used MOESI_CMP_directory for a while in my CMP simics target with a banked, shared L2 cache.

And I would like to clarify the use of L2_REQUEST/RESPONSE_LATENCY in order to properly set up the L2 latency.

 

I understood and used them as follows:

. L2_REQUEST_LATENCY is the average transit time for the request to get to the proper L2 bank from the CPU/L1

. L2_RESPOSNE_LATENCY is the average transit time for the (data) response to arrive at the CPU (L1 cache) from the proper L2 bank (& control).

 

If my understanding was wrong, please correct me and provide me with definitions/usage of L2_REQUEST_LATENCY and L2_RESPONSE_LATENCY.

 

Thanks,

 

soohong p kim