I mentioned the enable_periodic_dump in the IPC just as an example. You need to do the same thing to whatever StatObject you want. Not sure what you mean by "plot Cache miss By using Marss". Marss does not plot anything. Marss dump the stats as text (csv in this case) and you can use your favorite plotting software to plot it (R in Avadh's case) -Furat
On Fri, Jun 27, 2014 at 11:30 PM, Arsalan Shahid < [email protected]> wrote: > Thanks for your reply Furat. I had used enable_periodic_dump() which was > commented in ooo.cpp in the code call > > thread_stats.commit.ipc.enable_periodic_dump(); > > Then also I got the stats in csv file. > > > sim_cycle,time_ns,base_machine.ooo_0_0.cycles,base_machine.ooo_0_0.thread0.commit.insns,base_machine.ooo_0_0.thread0.commit.ipc,base_machine.ooo_0_0.thread1.commit.insns,base_machine.ooo_0_0.thread1.commit.ipc > 100000,50000,100000,1109,0.01109,30187,0.30187 > 200000,100000,100000,5753,0.05753,18864,0.18864 > 300000,150000,100000,24873,0.24873,999,0.00999 > 400000,200000,100000,31533,0.31533,1110,0.0111 > 500000,250000,100000,46878,0.46878,1110,0.0111 > 600000,300000,100000,47405,0.47405,1110,0.0111 > > But actually Furat I want to plot Cache miss By using Marss. Or atleast > Cache Miss rate which is already plotted by Avadh. But in the above data of > csv file there is not data for Cache. My question is > 1) Can we get data for cache in csv file ? > 2) Also can we plot a linear quantity i.e cache miss in Marss > > please suggest me ways to get Cache miss data in that csv file. > > Thanks > Arsalan Shahid > On Saturday, June 28, 2014 1:55 AM, Furat Afram <[email protected]> > wrote: > > > The changes you have in the cfg file enable the time stats. you need to > use enable_periodic_dump() in the source code to specify which stat you > want to be dumped. > one or two stats (I think IPC) has a commented call to > "enable_periodic_dump()" > you can check it out to see how it is used. > -Furat > > > On Wed, Jun 25, 2014 at 8:34 AM, Arsalan Shahid < > [email protected]> wrote: > > Yes, you are right Tyler. But Avadh has plotted Cache Miss rate and Miss > Ratio along with sim cycle in his blog. For that particular purpose we need > to have Cache Data in csv file to be used in R-project. > > I requested you to tell me about getting that particular data so that I > can plot Cache miss and miss rate. > Here is the link where Avadh has shows Plot of L2 miss ratio with cycles > > > https://3.bp.blogspot.com/-OqAKZAhM83o/T5__xpYYIJI/AAAAAAAABa0/i133eMpOe5k/s1600/l2_miss_ratio.png > > As told my Avadh that we have to make few changes in cfg file given to > Marss to obtain stats in csv file. > I made the following cfg file > > -machine ooo_2_th > -stopinsns 100m > -kill-after-run > -stats results.stats > -logfile results.log > -time-stats-logfile time_stats.csv > -time-stats-period 100K > > But The csv file obtained as a result don't have Cache Data. Is there any > solution for this ? > > Thanks and Regards > Arsalan Shahid > > On Wednesday, June 25, 2014 8:02 PM, Arsalan Shahid < > [email protected]> wrote: > > > Yes you are right Tyler. But Avadh has plotted Cache Miss rate and Miss > Ratio along with sim cycle in his blog. For that particular purpose we need > to have Cache Data in csv file to be used in R-project. > > I requested you to tell me about getting that particular data so that I > can plot Cache miss and miss rate. Attached is that plot. > As told my Avadh that we have to make few changes in cfg file given to > Marss to obtain stats in csv file. > I made the following cfg file > -machine ooo_2_th > > -stopinsns 100m > -kill-after-run > -stats results.stats > -logfile results.log > -time-stats-logfile time_stats.csv > -time-stats-period 100K > > But The csv file obtained as a result don't have Cache Data. Is there any > solution for this ? > > Thanks and Regards > Arsalan Shahid > > > On Wednesday, June 25, 2014 7:49 PM, "[email protected]" < > [email protected]> wrote: > > > The purpose of this mailing list is not to request new features or code > additions. Please limit your posts to specific MARSS-related questions. > > Thanks, > Tyler > > > > Hi > > > > I was having a problem regarding MARSS time stats which are obtained in > > .csv file after running a simulation. > > I am having the following data in csv file > > > > > sim_cycle,time_ns,base_machine.ooo_0_0.cycles,base_machine.ooo_0_0.thread0.commit.insns,base_machine.ooo_0_0.thread0.commit.ipc,base_machine.ooo_1_1.cycles,base_machine.ooo_1_1.thread0.commit.insns,base_machine.ooo_1_1.thread0.commit.ipc > > 100000,29481.1,100000,21636,0.21636,100000,1109,0.01109 > > 200000,58962.3,100000,52764,0.52764,100000,1110,0.0111 > > 300000,88443.4,100000,111788,1.11788,100000,1110,0.0111 > > 400000,117925,100000,112067,1.12067,100000,1110,0.0111 > > 500000,147406,100000,82724,0.82724,100000,1110,0.0111 > > > > but I want to have data of L1 and L2 time stats. Kindly tell me what > > changes should be made to get cache stats and where to make changes. > > > > Thanks and Regards > > Arsalan Shahid_______________________________________________ > > > http://www.marss86.org > > Marss86-Devel mailing list > > [email protected] > > https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel > > > > > > > > > > > _______________________________________________ > http://www.marss86.org > Marss86-Devel mailing list > [email protected] > https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel > > > > >
_______________________________________________ http://www.marss86.org Marss86-Devel mailing list [email protected] https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel
