You have to checkpoint the entire simulation (both systems), as there is TCP connection state on both sides that has to be in sync (and possibly even packets in flight on the wire). It is very useful to do a checkpoint to avoid not just the bootup but the app init time as well.
Unfortunately, in response to you and to Ying Zhang's other recent email, doing more than two systems is pretty constrained simply because we have never gotten around to developing an Ethernet switch model. Without that you can only have point-to-point links. You can have multiple systems if you meet that constraint, e.g. if each client connects to a separate NIC on the server, or if you want to model a system in the middle of a connection doing packet forwarding, but it's not a general solution. That said, writing an Ethernet switch model probably wouldn't be too hard (especially since you could even map ports to Ethernet MAC addrs in the config script), so if you really want to do general multi-system networks I wouldn't rule that out. Of course if you do decide to do that, we'd love to have you contribute it back, too! Steve On Thu, Oct 21, 2010 at 1:39 PM, Lide Duan <[email protected]> wrote: > Hi, > > I was able to simulate a dual-system under ALPHA_FS, using the pre-compiled > Surge as client and Apache as server. After booting the Linux with atomic > CPU, I would like to switch the server CPUs to detailed mode and measure > Apache activities for a certain amount of Surge work. My question is: in > order to repeat the simulation (say with a different Etherlink or server > configuration), should I create a checkpoint for the server system or the > entire dual-system? Put another way, how do you guys usually conduct > simulation for multiple-system? Do you guys restore from a checkpoint or > simply boot the Linux from the very beginning every time? > > One more thing is : is it possible to simulate more than 2 systems? Current > M5 python scripts only support a dual-system connected with a Etherlink. I > wonder if there is an easy way to construct a more complex network topology, > say a server with a few clients (star), or a ring network. From the maillist > archive, looks like M5 used to have a back-to-back-to-back topology, is it > something similar to this? > > > Thanks, > Lide > > _______________________________________________ > 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
