Oh, also, which CPUs? I'm planning to do at most 5 or 6 runs for run time and compile time.
Gabe On 11/26/11 00:06, Ali Saidi wrote: > I think running 1 hours worth would be fine... Just run a billion cycles. > > Ali > > On Nov 25, 2011, at 9:45 PM, Gabe Black wrote: > >> Doesn't that android benchmark take two days to run? I'm limited to my >> laptop right now so I don't think I can do that. If you want to I can >> tell you what to grab. You'd just need to clone my repository and use or >> not use a patch I have. That is, unless the compiler is *really* smart >> and figures out that it's still really a constant. >> >> Gabe >> >> On 11/25/11 22:14, Ali Saidi wrote: >>> Android disk image running bbench >>> A couple of SE workloads (perhaps one from each architecture) >>> Something else running in FS mode (trouble is that booting isn't great >>> because there is a lot of idle time) >>> >>> Ali >>> >>> >>> On Nov 25, 2011, at 8:35 PM, Gabe Black wrote: >>> >>>> I'm planning to measure how simulator performance changes when SE/FS is >>>> a runtime decision vs. compile time. There are a lot of possible cases >>>> to check (SE/FS, CPU, ISA) and getting reasonable results, so I wanted >>>> to see what variants would be most interesting to you guys. Any >>>> suggestions? >>>> >>>> Gabe >>>> _______________________________________________ >>>> gem5-dev mailing list >>>> [email protected] >>>> http://m5sim.org/mailman/listinfo/gem5-dev >>>> >>> _______________________________________________ >>> gem5-dev mailing list >>> [email protected] >>> http://m5sim.org/mailman/listinfo/gem5-dev >> _______________________________________________ >> gem5-dev mailing list >> [email protected] >> http://m5sim.org/mailman/listinfo/gem5-dev >> > _______________________________________________ > gem5-dev mailing list > [email protected] > http://m5sim.org/mailman/listinfo/gem5-dev _______________________________________________ gem5-dev mailing list [email protected] http://m5sim.org/mailman/listinfo/gem5-dev
