Yes, I had also previously observed that debug symbols make the huge majority of the executable's size, and for some reason much more so in .opt than in debug (presumably it takes more information to map back optimized code to source).
You can confirm that by using strip gem5.opt or gcc- s during build, which leads to binaries of the order of 60Mb. Related thread: https://gem5.atlassian.net/browse/GEM5-572 ________________________________ From: Liao Xiongfei via gem5-users <gem5-users@gem5.org> Sent: Friday, October 30, 2020 4:38 PM To: gem5-users@gem5.org <gem5-users@gem5.org> Cc: Liao Xiongfei <liao.xiong...@huawei.com> Subject: [gem5-users] File sizes of simulators generated by test script are much larger than those built by manual scons commands Hi all, When I ran the test script under $GEM5/tests using command “python main.py run” on a machine previously without any gem5 simulator built, the script built 5 different gem5.opt simulators before running test suites. However, the file sizes of the simulators are pretty large, shown below. The gem5.opt for RISC-V is 4.4GB and failed with error message “cannot allocate memory”. 1.8G build/ARM/gem5.opt 858M build/NULL/gem5.opt 4.4G build/RISCV/gem5.opt 1.5G build/X86/gem5.opt 1.5G build/X86_MSI/gem5.opt On the other hand, the gem5.debug simulation was built using “scons build/RISCV/gem5.debug –j 8” and its size is much smaller. 522M build/RISCV/gem5.debug Are these caused by different compiler or linker settings? I was surprised by a large number of failed tests because RISCV gem5.opt could not run. Hopefully, this could be fixed somehow. Thanks.
_______________________________________________ gem5-users mailing list -- gem5-users@gem5.org To unsubscribe send an email to gem5-users-le...@gem5.org %(web_page_url)slistinfo%(cgiext)s/%(_internal_name)s
