Hello, everyone

Now I am using M5 to simulate 4-core CPU. When I saw the stats.txt reports,
here are some questions showing below.

1. For some L2 statistic parameters, there are 5 or 6 different values, for
example

system.l2.occ_blocks::0 4296.306219 # Average occupied blocks per context
system.l2.occ_blocks::1 4219.322094 # Average occupied blocks per context
system.l2.occ_blocks::2 4222.702362 # Average occupied blocks per context
system.l2.occ_blocks::3 4224.841879 # Average occupied blocks per context
system.l2.occ_blocks::4 7387.035504 # Average occupied blocks per context

system.l2.occ_percent::0 0.131113 # Average percentage of cache occupancy
system.l2.occ_percent::1 0.128763 # Average percentage of cache occupancy
system.l2.occ_percent::2 0.128867 # Average percentage of cache occupancy
system.l2.occ_percent::3 0.128932 # Average percentage of cache occupancy
system.l2.occ_percent::4 0.225434 # Average percentage of cache occupancy

Usually, the fifth value is the summary of the first four. However for the
"occ_blocks", and "occ_persent", it is not the summary. Does anybody know
what the fifth value means?

2. For 6 different values case, for example

system.l2.overall_hits::0 32916188 # number of overall hits
system.l2.overall_hits::1 32991021 # number of overall hits
system.l2.overall_hits::2 30251740 # number of overall hits
system.l2.overall_hits::3 32917713 # number of overall hits
system.l2.overall_hits::4 0 # number of overall hits
system.l2.overall_hits::total 129076662 # number of overall hits

Where is "system.l2.overall_hits::4" coming from?

3. I have read all stats.txt, and I didn't find any main memory access
statistic. So can I deduce the memory read and write from L2 statistics? For
example,

# main memory read requests == l2.replacement
# main memory write requests == l2.writeback

is that correct?


Many thanks in advance.

Bests
Dawei
_______________________________________________
gem5-users mailing list
gem5-users@gem5.org
http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users

Reply via email to