Every time a set of stats is created (E.g. if you execute m5 dumpresetstats
1000 1000) every 1000 cycles a new begin/end pair should be emitted.
Ali
On Nov 16, 2010, at 6:41 PM, Stevenson Jian wrote:
> In case you don't see the underline, here is the line that I added:"
> if((*i)->name.compare("system.l2.overall_hits")==0 ||
> (*i)->name.compare("system.l2.overall_accesses")==0)
> "
>
> Thanks,
> Steve
> On Tue, Nov 16, 2010 at 6:36 PM, Stevenson Jian <[email protected]>
> wrote:
> Hi,
> I figured out an easy way of doing this. I went into Text.cc and changed the
> output() routine to the following:
> ############
> void
> Text::output()
> {
> ccprintf(*stream, "\n---------- Begin Simulation Statistics
> ----------\n");
> list<Info *>::const_iterator i, end = statsList().end();
> for (i = statsList().begin(); i != end; ++i){
> if((*i)->name.compare("system.l2.overall_hits")==0 ||
> (*i)->name.compare("system.l2.overall_accesses")==0){
> (*i)->visit(*this);
> }
> }
> ccprintf(*stream, "\n---------- End Simulation Statistics
> ----------\n");
> stream->flush();
> }
> ###############
> Essentially, I check for the name of the info before allowing it to be
> visited.
> While this works, I am encountering a very peculiar behavior. In the
> Stats.txt in the M5out output folder, I am getting double amount of output.
> For example, before changing Text.cc, I get
> "-------------------Begin Simulation Statistics -------------"
> (useful stats here...)
> "-------------------End Simulation Statistics ---------------"
>
> But after changing Text.cc by adding the underlined lines, I get the
> following:
> "-------------------Begin Simulation Statistics -------------"
> (useful stats here...(A))
> "-------------------End Simulation Statistics ---------------"
> "-------------------Begin Simulation Statistics -------------"
> (useful stats here...(B))
> "-------------------End Simulation Statistics ---------------"
>
> Where A and B are not the same. While I verified that the numbers in B are
> what I want, I am very curious what the numbers for A represent and how to
> prevent A from getting into the output file.
>
> Does anyone have any ideas?
> Thanks,
> Steve
>
> _______________________________________________
> m5-users mailing list
> [email protected]
> http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
_______________________________________________
m5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
