Hi Vilas,

Are the diffs complete? The max_inst option never appears in Options.py but it's used in Simulation.py

Thanks,
ali

On Feb 15, 2008, at 2:00 PM, Vilas Sridharan wrote:


As an aside, if you want to fast forward 300M instructions and not
necessarily 300M cycles, it would be good to be able to specify that
directly... didn't we just have a thread on the mailing list about
that?  I don't think it's in the code right now, but did someone out
there add it?  If so, would you mind contributing the code back?

I added this functionality to Simulation.py and Options.py. I use -- fast_forward on its own to create checkpoints, and then with -- standard_switch to warmup caches. I've attached the diffs.

A word of caution: I had to back out some of my other changes to Simulation.py to create this diff, so I haven't tested this on its own. But it should be really close to what is needed.

   -Vilas


Steve

On Jan 26, 2008 8:31 AM, dean michael ancajas <[EMAIL PROTECTED]> wrote:
> hi everybody,
>    this is my first time to fast forward instructions using
> AtomicSimpleCPU. I want to fast forward to 300M instructions (for gzip > graphic-ref) so I fast forward to (1Ghz * 300M= 3x10^17 ticks), its now > been simulating for 29hrs now and I decided to attach 'm5term' and it
> says "113180285494000: system_remote.gdb: remote gdb attached". My
> question is, is that number the tick number? If it is and assuming
> constant ticks/second execution of the simulator, if I divide
> (3X10^17)/1131180285494000=265. Does this mean the whole simulation will
> take 265*29hrs?
>
> this is the command that i invoked:
> [EMAIL PROTECTED] ./build/ALPHA_SE/m5.opt
> --stats-file=statsdir/gzip_graphic_stats.txt configs/example/se.py
> --cmd=/home/m5user/m5/spec2000/binaries/alpha/tru64/gzip --input=
> /home/m5user/m5/spec2000/data/gzip/ref/input/input.graphic
> --max_checkpoints=1 --take_checkpoints=300000000000000000,10000
> --checkpoint_dir=cpt_gzip_gra
>
> btw,
>   i adjusted my clockrate in se.py to be 1Ghz.
>
> thanks in advance,
> -dean
> _______________________________________________
> m5-users mailing list
> m5-users@m5sim.org
> http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
>
_______________________________________________
m5-users mailing list
m5-users@m5sim.org
http://m5sim.org/cgi-bin/mailman/listinfo/m5-users

< Simulation .py .diff><Options.py.diff>_______________________________________________
m5-users mailing list
m5-users@m5sim.org
http://m5sim.org/cgi-bin/mailman/listinfo/m5-users

_______________________________________________
m5-users mailing list
m5-users@m5sim.org
http://m5sim.org/cgi-bin/mailman/listinfo/m5-users

Reply via email to