Hi, I am currently using M5 running uniprocessor workload. I find that
there is some unwanted coherence overhead in L2 cache (ReadExReq is always
a miss). Because my current work does not require any coherence, is there
any way to bypass the coherence protocol?

_______________________________________________
m5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/m5-users

Reply via email to