[Gems-users] How to collect multicore cache coherence traffic？

[jianhual@xxxxxxxxxxxxxxxx]

Dear All,

    I am a newbie to GEMS framework. Recently, I am trying to use GEMS framework for my simulation. I know that SLICC in GEMS implement the cache coherence protocol. I have several questions on SLICC and can somebody explain these questions for me. Thanks in advance.

1. I want to collect on-chip network traffic trace for a CMP system. If I use MESI_CMP_directory protocol for a mesh-based CMP system, how can I collect the network  traffic, such as cache read, cache write and write-invalidate (multicast) and so on. If I want to store these traffic trace to a file, where can I add code in GEMS? Can such function be implemented in SLICC?

2. I know GEMS simulates serengeti machine that run Solaris, can such platform support 64-core?
 
I have posted the question above for several times, but nobody reply me. I wish somebody can reply me. Thanks very much. I just want to know how can I export the cache coherence traffic into a file from the GEMS multicore platform running multithreaded workloads?

Any suggestions from you will be appreciated. Thanks again!

Regards,
Jianhua