Hi Sue,
I merely want to highlight that clang seems to use far less memory than gcc
during both compilation and linking. If you have the possibility, using clang
might mitigate the issue until the decoder is split.
Andreas
From: [email protected] [mailto:[email protected]] On
Behalf Of Steve Reinhardt
Sent: 24 July 2012 23:21
To: gem5 users mailing list
Subject: Re: [gem5-users] [EXTERNAL] Re: Gem5 Red Hat build not completing
Thanks for the heads up. I see in the mailing list archives where others have
run into this... I don't know what's unique about Red Hat that causes these
problems.
As I mentioned before, I did spend some time looking into how we could split up
the x86 decoder.cc into chunks, and it's very feasible; it just needs someone
with the time and initiative to make it happen:
http://www.mail-archive.com/[email protected]/msg04744.html
Steve
On Tue, Jul 24, 2012 at 2:00 PM, Kelly, Suzanne M
<[email protected]<mailto:[email protected]>> wrote:
Thank you for the suggestion. We'll give it a try later this week or next.
Sue
-----Original Message-----
From: [email protected]<mailto:[email protected]>
[mailto:[email protected]<mailto:[email protected]>] On
Behalf Of Nilay Vaish
Sent: Monday, July 23, 2012 11:15 AM
To: gem5 users mailing list
Subject: [EXTERNAL] Re: [gem5-users] Gem5 Red Hat build not completing
On Mon, 23 Jul 2012, Kelly, Suzanne M wrote:
> Greetings, Based on a web page review, we (the Sandia SST team) suspect
> you had/are experiencing a similar problem when trying to build GEM5 on
> some Red Hat systems. For us, the problem manifests itself when
> compiling decoder.cc. The compilation consumes large amounts of CPU time
> and never finishes because gcc requires over 15 GB of memory for the
> compilation. decoder.cc is a generated file and is about 119,000 lines.
> Our bailing wire is to recognize that the compile options being selected
> are -g and -O3. With these options, the resulting object file (on a
> system with sufficient memory) is 3.8 _GB_, With just -O3 the object
> file is 12 MB. With just -g the object file is 16 MB. Based on some
> testing, the problem is not unique to a specific version of gcc.
>
> We are now running a local patch to remove the -g option. Unfortunately,
> the change is applied to all objects generated as part of the gem5
> build. Have you found a better solution? If you have any suggestions, we
> would be happy to try them out.
>
> Regards,
> Sue Kelly
>
> PS Our local GEM5 copy is changeset: 8642:549b72de8f72
>
>
Sue, you might want to take a look at how -Werror option can be set
specifically for a c/c++ file. Grep for Werror in the file src/SConscript.
It is plausible that something similar can be worked out for the -g
option.
--
Nilay
_______________________________________________
gem5-users mailing list
[email protected]<mailto:[email protected]>
http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users
_______________________________________________
gem5-users mailing list
[email protected]<mailto:[email protected]>
http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users
-- IMPORTANT NOTICE: The contents of this email and any attachments are
confidential and may also be privileged. If you are not the intended recipient,
please notify the sender immediately and do not disclose the contents to any
other person, use it for any purpose, or store or copy the information in any
medium. Thank you.
_______________________________________________
gem5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users
