Hello everyone,
I have a issue that does not make sense for me. I am using Ruby and I want to
use the bank contentiousness interface gem5 has implemented in
src/mem/ruby/structures/BankArray.
I enable the bank contentiousness in Ruby adding the calling to the functions,
for example L2cache.recordRequestType into my protocol, also I enable the
m_resource_stalls in src/mem/ruby/structures/RubyCache.py. If the resource is
busy I recycle the input in the protocol, In the results I confirmed a lot of
signals have been recycled because the resource is busy.
The issue is that I am testing two algorithms (same intput) one has a lot of
stalls more that the other, I can see that in the stats.txt result file, but
this algorithm has a sim_seconds result smaller than other one. It is like gem5
has not considered the stalls into the sim_seconds calculation.
Algorithm 1 L2cache.num_data_array_stalls: 490142
sim_seconds: 0.001618
Algorithm 2 L2cache.num_data_array_stalls: 61919
sim_seconds: 0.001715
Thanks in advance.
Rodrigo
_______________________________________________
gem5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users
