The solution that Somayeh is working on will allow us to remove the ruby event queue API. Ruby already uses the main event queue underneath for all operations except cache warmup. When executing a warmup trace, the event queue must move forward. If Ruby is the first SimObject unserialized, then the event queue can be moved forward when the trace is executed. Subsequently the rest of the system is unserialized and the curTick is set the appropriate value.
Brad > -----Original Message----- > From: [email protected] [mailto:[email protected]] On > Behalf Of nathan binkert > Sent: Saturday, June 18, 2011 12:52 PM > To: gem5 Developer List > Cc: M5 Developer List > Subject: Re: [gem5-dev] [m5-dev] Ruby Cache Warmup > > > As I mentioned to you over the phone, I think the best way to > manipulate the order of the simObjectList is to modify the SimObject > constructor to push the associated SimObject pointer on the front of > the list instead of the back. You could do this by simply adding a > Boolean parameter to SimObject that indicates whether to push it on the > front or the back of the list. Eventually, we may want to make that > some sort of integer for finer grain order, but for now a simple bool > is probably easiest. > > > > Others, please speak up if you disagree. Otherwise I believe this is > the direction Somayeh is going to take. > > This sounds super sketchy. Why not just get rid of the ruby event > queue and start using the main event queue? It's really not that much > work. All you need to do is convert cycles to ticks. > When we did this for M5 it was actually pretty quick. > > We could even create a ClockedObject that has a clock and some > functions for scheduling based on cycles. This would make it even > easier. > > Nate > _______________________________________________ > 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
