You would basically undo this change: https://gem5-review.googlesource.com/c/public/gem5/+/40815
except there should also be an option to re-enable LTO for the other builds. Like I said, for some users (myself included) LTO actually speeds up the build. Gabe On Tue, Apr 20, 2021 at 9:55 PM Bobby Bruce <bbr...@ucdavis.edu> wrote: > I did create a patch for this: > https://gem5-review.googlesource.com/c/public/gem5/+/44565. I _think_ > this is the correct way to make LTO for .fast only. I won't submit until > the community agrees this is the best course of action. > > Kind regards, > Bobby > -- > Dr. Bobby R. Bruce > Room 2235, > Kemper Hall, UC Davis > Davis, > CA, 95616 > > web: https://www.bobbybruce.net > > > On Tue, Apr 20, 2021 at 5:07 PM Gabe Black via gem5-dev <gem5-dev@gem5.org> > wrote: > >> I think that's reasonable. I didn't think the performance difference >> would be as much as it is, since my machine has more cores than that, 12 >> with 2 threads per core. It actually speeds up the build for people who >> have enough cores which offset the extra work the linker has to do, but I >> suspect those people are not the common case. We should still make it >> possible to use a flag to enable LTO explicitly for a given build, so >> basically switch the polarity of the default and the override flag. >> >> Gabe >> >> On Tue, Apr 20, 2021 at 7:54 AM Jason Lowe-Power via gem5-dev < >> gem5-dev@gem5.org> wrote: >> >>> Hi all, >>> >>> It seems the recent change to enable LTO by default is causing more harm >>> than good. Especially for people debugging, it's a large overhead to wait >>> 2-5 minutes for gem5 to link in opt mode. I even have encountered this with >>> a 6 core 3.5+GHz Ryzen. It's quite annoying. We've gotten at least 3 >>> different posts on gem5-users and gem5-dev complaining about this. I think >>> this is a strong sign we need to change something. >>> >>> I suggest *only enabling* LTO for fast builds and not for any other. >>> >>> Thoughts? >>> >>> Jason >>> _______________________________________________ >>> gem5-dev mailing list -- gem5-dev@gem5.org >>> To unsubscribe send an email to gem5-dev-le...@gem5.org >>> %(web_page_url)slistinfo%(cgiext)s/%(_internal_name)s >> >> _______________________________________________ >> gem5-dev mailing list -- gem5-dev@gem5.org >> To unsubscribe send an email to gem5-dev-le...@gem5.org >> %(web_page_url)slistinfo%(cgiext)s/%(_internal_name)s > >
_______________________________________________ gem5-dev mailing list -- gem5-dev@gem5.org To unsubscribe send an email to gem5-dev-le...@gem5.org %(web_page_url)slistinfo%(cgiext)s/%(_internal_name)s
