Thanks for the reply Hossein. The applications I am running the simulation for are sufficiently tuned to the cluster hardware that I think I need cycle-accurate predictions for changes we make. That is, I don't think we can get away with much abstraction.
I am also looking at how to incorporate gem5 into SST, which is my plan for scaling up past 2 nodes. From: [email protected] [mailto:[email protected]] On Behalf Of Hossein Nikoonia Sent: Friday, October 26, 2012 11:57 PM To: gem5 users mailing list Subject: Re: [gem5-users] cluster simulation gem5 uses interconnection network within a "system" (i.e. CPU+OS+Caches+DirecrotyControllers+ ...) I guess what you want is a TCP/IP-network-connected systems. This is also supported; at least for two systems. For more than that, you have to add software-simulated swtich/routers between. remember! this might not be the best way to simulate a cluster. you should think about the level of abstraction you need. On Thu, Oct 25, 2012 at 5:03 PM, Payne, Benjamin <[email protected]> wrote: Hello, I'm interested in a cycle-accurate simulation of a cluster of nodes, each with their own CPU/RAM+OS and assocaited network connections. It looks like gem5 can do this according to http://www.m5sim.org/Interconnection_Network and https://lists.cs.wisc.edu/archive/gems-users/2012-March/msg00009.shtml but I don't see any mention of it in the paper https://research.cs.wisc.edu/multifacet/papers/can11_gem5.pdf What is the status of networking and gem5? Is a cluster simulation a realistic objective? Kindly, Ben Payne 5520 Research Park Drive Catonsville, MD 21228-4870 Laboratory for Physical Sciences http://www.lps.umd.edu/ _______________________________________________ gem5-users mailing list [email protected] http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users _______________________________________________ gem5-users mailing list [email protected] http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users
