Sujay, But the mcf, lucas, facerec, etc. all work without these changes too. The > benchmarks/cpu2000/data/mcf/input actually contains mcf.in file, and it > does run with this. >
That may be true in your local version of the CPU2000 benchmarks, but it isn't true in general (though it may be the reason the discrepancy wasn't noticed until now). As distributed by SPEC, I believe the filenames correspond with what I sent out, and I think M5 should strive to be generic here. -Vilas > > > ----- Original Message ----- > *From:* Vilas Sridharan <[EMAIL PROTECTED]> > *To:* M5 users mailing list <[email protected]> > *Sent:* Monday, April 21, 2008 12:37 PM > *Subject:* Re: [m5-users] mcf benchmark > > Nate, > Here's an attempt at a diff. I say 'attempt' because (a) I have some > local modifications to the cpu2000.py file, and (b) I haven't tested this > since I'm working around these problems a different way. I generated this > diff using an unmodified version of cpu2000.py, though, and it looks okay, > so hopefully it should apply cleanly. > > -Vilas > > On Sun, Apr 20, 2008 at 12:07 AM, nathan binkert <[EMAIL PROTECTED]> wrote: > > > This doesn't appear to be in our tree right now. Can you send us a > > diff so we're sure to get it right? > > > > Thanks, > > Nate > > > > On Sat, Apr 19, 2008 at 8:38 PM, Vilas Sridharan > > <[EMAIL PROTECTED]> wrote: > > > There was a bug in the cpu2000.py file where mcf (and several of the > > other > > > SPEC2k benchmarks) were getting the wrong input files. From the list > > > archives, this is what I sent out: > > > > > > gzip_source: the second input should be '60', not '1' > > > mcf: should get inp.in as argument, not mcf.in > > > parser: needs ref.in on stdin > > > equake: needs inp.in on stdin > > > facerec: needs ref.in on stdin > > > lucas: needs lucas2.in on stdin > > > sixtrack: needs inp.in on stdin > > > > > > This was as of Jan 7 of this year; I don't know if this got fixed in a > > more > > > recent version (or which version you're using). > > > > > > -Vilas > > > > > > > > > On Sun, Apr 20, 2008 at 8:45 PM, Gabe Black <[EMAIL PROTECTED]> > > wrote: > > > > We use mcf as a regression test so it at least works a little. If it > > only > > > runs a few thousand instructions, it's likely either the standard > > library > > > initialization is failing, or mcf itself is quiting with an error very > > early > > > in execution. If you haven't yet, you should check it's output and see > > if > > > there are any messages. Also, with so few instructions, it should be > > fairly > > > easy to add --trace-flags=Exec --trace-file=trace.out to your command > > line > > > and inspect what it's doing directly. The function it's executing last > > > should be a good clue as to what's going on. > > > > > > > > Gabe > > > > > > > > > > > > > > > > > > > > Shoaib Akram wrote: > > > > > > > > > Did anybody tried running mcf benchmark in SE mode. All of the > > other > > > SPEC CINT benchmarks I am using run for millions of instructions but > > mcf > > > runs for a couple of thousand even under ref input. > > > _______________________________________________ > > > > > 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 > > > > > > > > > > > > > _______________________________________________ > > > 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 > > > > ------------------------------ > > _______________________________________________ > 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 >
_______________________________________________ m5-users mailing list [email protected] http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
