Cycles number increased when data cache latency is decreased.

[Guerra260267]

I tried to simulate a data cache with latency = 1 so I can see the impact caused in the number of cycles after the simulation. When I opened the out file, the final number of cycles used by the only thread, was almost ten thousand cycles greater than the number of cycles used by the same test using a data cache with latency = 4. Do anyone knows how can I fix it? There's no mistakes in the test file. Thanks for the help!

[fbmoreira]

This is likely related to what has been reported in issues #66 and #104. The access latency is set to zero for terminal caches in init.cpp, and a macro is used to define an access latency (4 cycles) in the ooo_core.cpp file. Therefore, you only changed the invalidation access latency.

It is still curious, however, that the same test, executed with a smaller invalidation latency, made the core stall more cycles. Anyone got any ideas on what could have happened?