You should use the committedInsts statistic. 

Ali

On Jul 19, 2013, at 4:27 PM, Fulya Kaplan <[email protected]> wrote:

> Hi all,
> I am using the precompiled binaries for Parsec, so I decided to follow this 
> procedure to switch to detailed mode at the beginning of ROI:
> 1)Do a simple run.
> 2) Parsec precompiled binaries are instrumented already to dump and reset 
> statistics at the beginning of ROI. Get the number of instructions simulated 
> until ROI from the simple run stats.
> 3)Do the 2nd detailed run with fastforwarding until that number of 
> instructions is reached (in other words, switch to detailed mode at the 
> beginning of ROI).
>  
> My question is what statistics I should use as the fastforward parameter?
> There is a parameter called sim_insts (#Number of instructions simulated). I 
> read in one of your replies that it is meant for analyzing the performance of 
> the simulator. However, I am not sure which instruction count the 
> “–fast-forward” option takes into account. If it is not based on sim_insts, 
> what should I use instead? (I have a 2 core system).
>  
> I tried using the sim_insts parameter that i got from the 1st run and used it 
> as the fastforward parameter. However, when i compare the ticks when the 
> switch cpu occurs (using the info printed on the screen) with the ticks that 
> corresponds to the beginning of ROI (from the simple run stat.txt file), I 
> see a big mismatch. It switches @tick 8.06e12 instead of @tick 7.48e12 which 
> is a lot later than thebeginning of ROI (actually it is almost the end of the 
> whole simulation).
>  
> What am i doing wrong? Should i use another number for fast-forwarding? 
> Please help.
> Best,
> Fulya Kaplan
>  
> _______________________________________________
> gem5-users mailing list
> [email protected]
> http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users

_______________________________________________
gem5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users

Reply via email to