Re: [gem5-users] Questions about gem5 system configurations

2019-01-17 Thread Andrea Mondelli
Hi Weizheng, Have you tries to explore the parameters offered by the se.py script? It could be a good starting point if you want to play with a meaningful configuration. Bests, A. Il giorno 17 gen 2019, alle ore 16:13, Weizheng Xu mailto:weizhengxu2...@u.northwestern.edu>> ha scritto: Hello

Re: [gem5-users] A build issue on RISCV gem5

2019-02-07 Thread Andrea Mondelli
Can you specify the command, the commit and the system used? It might help to reproduce the problem. Bests, A. > Il giorno 7 feb 2019, alle ore 02:12, Shimomura Shigeki > ha scritto: > > Hello, > > I tried the suggested patch, and now I got the following error... > > Thanks, > Shigeki > >

Re: [gem5-users] A build issue on RISCV gem5

2019-02-05 Thread Andrea Mondelli
Hi Can you try to apply the following patch? https://gem5-review.googlesource.com/c/public/gem5/+/16103 Bests, A. Il giorno 5 feb 2019, alle ore 11:22, Shimomura Shigeki mailto:sshimomu...@gmail.com>> ha scritto: Hello, I just got gem5 source codes over the weekend and trying to build it,

Re: [gem5-users] Instrumenting ROI in Syscall emulation mode

2019-06-11 Thread Andrea Mondelli
Hi Victor, I use checkpoints with SE.py , can you describe your problem? Is it related to the various “—work-{begin | end}-*” parameters in Options.py? Il giorno 10 giu 2019, alle ore 12:43, Victor Kariofillis mailto:vickariofil...@gmail.com>> ha scritto: Hi, I have already implemented pseudo

Re: [gem5-users] Trace file capture.

2019-06-18 Thread Andrea Mondelli
Hi Ghalib “Debug-flags” is a gem5.opt parameter. The error is related to the script (se.py). If you want to execute your a.out , you have to pass the correct cmd parameter. More info with: gem5.debug configs/example/se.py --help Il giorno 18 giu 2019, alle ore 09:01, Ghalib Khan

Re: [gem5-users] Switching off AVX-512 for running SPEC 2017 in SE mode

2019-07-31 Thread Andrea Mondelli
Hi A solution could be to compile SPEC in a docker with Debian Jessie. Sometimes the problem comes from unsupported instructions used by system libraries, and not the benchmark itself. Another way is to compile and use an old libc* version, but I found this option time-consuming. Il giorno