On Mon, February 11, 2013 6:39 pm, Beckmann, Brad wrote: > Hi Nilay, > > Do you know why the stat values changed? The values seem relatively > close, but I would have expected no changes in the values.
Hi Brad, There are three different cases in which we can divide the changes to stats: a. changes to the only fs ruby regression: ruby.stats changed due to recent changes to FSConfig.py in which SimpleDRAM was replaced with SimpleDDR3. Andreas updated the stats.txt file but not ruby.stats. b. all protocols except MOESI_CMP_directory: I think there are absolutely no changes to stats as such. Only a line related to Total_nonPF_delay_cycle has been removed. I removed it because we do not record any such statistic. This seems to be a remnant from the past. c. changes to regressions related to MOESI CMP directory protocol: the changeset 28d88a0fda74 made a single line change to the protocol. The ordering of files in MOESI_CMP_directory.slicc was changed. I checked it twice and surprisingly this change in ordering is causing the statistics to change. I personally do not know as to why this should cause the statistics to change. -- Nilay _______________________________________________ gem5-dev mailing list [email protected] http://m5sim.org/mailman/listinfo/gem5-dev
