I imagine you're seeing a combination of things. The main memory image is compressed with gzip, so that reduces the file size significantly. Also, I would check what input set you're using and if you've actually selected the reference input set. The others do not use 800MB, and running the reference input set to half way through the benchmark will likely take many days of execution.
Thanks, Ali On 23.05.2012 09:57, Kiyeon Lee wrote: > Hi. > > I am running ARM full system simulations using the checkpoints generated from SPEC CPU 2006 benchmarks. > I am using the VExpress_EMM platform with a 2GB main memory. > To generate a checkpoint, I fast-forward a program using atomicSimpleCPU until the program reaches the region of my interest. > > I have a question about the size of the checkpoint generated by the gem5 simulator. > I recently realized that the size of the checkpoint is much smaller than I expected. > > For instance, as far as I know, the mcf benchmark from the SPEC CPU 2006 benchmark suite dynamically allocates around 800MB of memory space at the start of the program execution. > However, the files created after generating a checkpoint in the middle of mcf program execution, only sums up to around 20MB. > The largest file (13MB) is a file named "system.physmem.physmem". > > I was expecting the checkpoint file size to be much much larger than 20MB, since the checkpoint has to capture the memory image of mcf execution. > I apologize for my lack of knowledge, but can some one help me understand how the simulator keeps the checkpoint file size so small? > > Thanks! > > Kiyeon
_______________________________________________ gem5-users mailing list [email protected] http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users
