Hi Reyad, I am not sure what might be going wrong in your case then. But, in our experience, when we ran almost the same number of gem5 jobs with similar hardware resources we saw some bad_alloc() errors on stderr suggesting insufficient memory.
-Ayaz On Fri, Feb 7, 2020 at 5:22 PM Hafizul Islam Reyad <reyad...@gmail.com> wrote: > Hi Ayaz, > > > > Thank you for your response. I have checked if any of the gem5 process was > actually killed because of being Out of memory but it’s not the case. The > simulations doesn’t use that much memory and I have checked occasionally to > see the system resource usage and never found memory usage more than 50%. > Could there be any other reason beside this? > > > > Regards, > > Reyad > > > > *From: *Ayaz Akram <yazak...@ucdavis.edu> > *Sent: *Friday, February 7, 2020 8:19 PM > *To: *gem5 users mailing list <gem5-users@gem5.org> > *Subject: *Re: [gem5-users] Multiple simultaneous benchmark run on same > diskimage/kernel > > > > Hi Reyad, > > > > Sharing the disk image and kernel should not be a problem. I will suggest > you to make sure that the OOM (out of memory) killer is not getting > activated on the system (usually logged in /var/log/messages). > > > > Regards, > > Ayaz > > > > On Fri, Feb 7, 2020 at 4:53 PM Hafizul Islam Reyad <reyad...@gmail.com> > wrote: > > > > Hi, > > > > I am new to gem5 and I was trying to run multiple benchmarks on gem5 to > get some results for my research. I face a issue that whenever I run > multiple gem5 simulations on one disk image, some of the gem5 process exits > without giving any results or errors. > > > > For example, I have about 20 microbenchmarks and 3 gem5 configurations > (e.g. different cachesize/ memory size etc.). So when I start a benchmark > run, I use one disk image and one kernel to run all these simulations (20 > microbenchmarks * 3 gem configs = 60 simulations). Out of these 60 > simulations, some of gem process (about 10) stops without giving any error > or exit message. When I run the incomplete benchmarks again (using the same > command), they finish successfully. > > > > My question is that if I share the same disk image and kernel across a lot > of simultaneous gem5 simulation, does this cause any issue? If not, what > might be the possible reason for this? > > > > I ran my simulations in a bare-metal AWS server that had 72 threads and > 512 GB memory, so I don't think resource was an issue. > > > > Example gem5 script: > > > > build/X86_MESI_Two_Level/gem5.opt --outdir=x configs/example/fs.py > --kernel=binaries/x86_64-vmlinux-3.4.112.smp > --disk-image=disks/x86_benchmark.img --script=test.rcS > --cpu-type=AtomicSimpleCPU --num-cpus=4 --caches --l1d_size=64kB > --l1i_size=64kB --l1d_assoc=2 --l1i_assoc=2 --l2cache --l2_size=16MB > --num-l2caches=4 --l2_assoc=8 --mem-channels=2 --mem-ranks=2 --mem-size=2GB > --ruby --num-dirs=2 --smt --network=garnet2.0 --topology=Mesh_XY > --mesh-rows=2 > > > > > > Thank you > > With regards, > > Reyad > > > > _______________________________________________ > gem5-users mailing list > gem5-users@gem5.org > http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users > > > _______________________________________________ > gem5-users mailing list > gem5-users@gem5.org > http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users
_______________________________________________ gem5-users mailing list gem5-users@gem5.org http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users
