> Korey, the timings for the two memory systems can be different. Sure, the timing will be different, but I'm thinking if you do a small test (say 2 threads) you should be able to capture similar sharing scenarios as soon as the parallel sections start up.
IF you dont want to use the splash2 stuff, I'm not sure you need published tests though, maybe just some microbenchmarks where there is locks are being obtained/released a bunch of times. In terms of supporting threads, there is the "m5threads" extension you can compile into your binary and then the alpha ISA had some limited pthread support. Steve explains a little of what is going on here: http://m5sim.org/Splash_benchmarks -- - Korey _______________________________________________ gem5-dev mailing list [email protected] http://m5sim.org/mailman/listinfo/gem5-dev
