Hi Nilay, Thanks for the pointer. I have added some comments to that patch (after essentially having created a similar one myself just now :-)
I'm still puzzled as to why the switcheroo regressions changed with the cache line size patches. The configurations are identical, yet they diverge. Any ideas? Andreas On 06/08/2013 12:41, "Nilay Vaish" <[email protected]> wrote: >Lena Olson posted a patch some time back that corrects the idle fraction >for atomic and timing cpus. > >http://reviews.gem5.org/r/1953 > >-- >Nilay > >On Tue, 6 Aug 2013, Andreas Hansson wrote: > >> Hi all, >> >> A bit of an update here. It turns out I also had changed values for the >> switcheroo regressions (but not the pc ruby one). On a closer inspection >> it is my changes that move the cache line to the system that caused it. >>I >> still don't know why, but I'm trying to iron that down. >> >> On a much more exciting note, it turns out that the reference stats for >> these regressions have some rather entertaining idle_fractions and >> num_busy_cycles, with fractions having values like -116 etc. I will >>post a >> patch that fixes this and also bump the stats. >> >> Andreas >_______________________________________________ >gem5-dev mailing list >[email protected] >http://m5sim.org/mailman/listinfo/gem5-dev > -- IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you. _______________________________________________ gem5-dev mailing list [email protected] http://m5sim.org/mailman/listinfo/gem5-dev
