Hi, I am running some multicore, multithreaded benchmarks (1 thread/core) on a gem5 full-system simulation, and I am having trouble interpreting the data cache misses that I am seeing.
As far as I am concerned, the threads do not share any data. However, I am seeing an increase in data cache misses concentrated at a certain address with more threads. I looked into the packet requests that cause the cache misses, and the virtual and phyisical address of the requests are the same. Because of this, I am wildly guessing that the misses might be due to some OS interaction, but I might be completely off. This might be a long shot, but has anyone seen similar behavior with multithreaded full-system simulations? I would really appreciate any advice or help. Thanks in advance, Jiwon
_______________________________________________ gem5-users mailing list firstname.lastname@example.org http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users