sage 8.9 crash on older CPU

51 views
Skip to first unread message

Jan Groenewald

unread,
Oct 1, 2019, 5:35:09 AM10/1/19
to sage-devel
Hi

I have compiled sage 8.9 on a KVM used as a master image for a computer lab.
The CPU is presented as model name      : Intel Xeon E3-12xx v2 (Ivy Bridge)

Imaging is using systemimager based on rsync.

On the newer lab machines with Intel(R) Core(TM) i5-3470 CPU @ 3.20GHz
the sage copied to the lab machines works fine.

On the older lab machines with Intel(R) Core(TM)2 Duo CPU     E8400  @ 3.00GHz
there is a crash as attached.

Is there a flag I can set at compile time to compile for the older CPUs by disabling use of some instruction not available on the older CPUs, which is what I suspect is happening?

Regards,
Jan
--
  .~.
  /V\     Jan Groenewald
 /( )\    www.aims.ac.za
 ^^-^^ 

Sage_crash_report.txt

Dima Pasechnik

unread,
Oct 1, 2019, 5:47:37 AM10/1/19
to sage-devel
looks like you need to pass some flags to NTL,
apparently it doesn't know what Core 2 CPUs are any more...

Perhaps reporting this to NTL?
> --
> You received this message because you are subscribed to the Google Groups "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/CAAg%3Dp_1Z_MuZ2r%2Bihgq9BQAhPntVuFboW58xE75rLigm0oYS%3Dg%40mail.gmail.com.

François Bissey

unread,
Oct 1, 2019, 5:50:18 AM10/1/19
to sage-...@googlegroups.com
Have you used “SAGE_FAT_BINARY=yes”? it is the only way to make a somewhat portable
binary that doesn’t have the problem you describe.
> To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/CAAWYfq0n8CrOwO_Jz7YKLJ%2Bc_XSGaVbsYuSf%3D3Yd%2BddE5%2BNP8A%40mail.gmail.com.

Jan Groenewald

unread,
Oct 2, 2019, 1:44:07 PM10/2/19
to sage-devel
Hi

On Tue, 1 Oct 2019 at 11:50, François Bissey <frp.b...@gmail.com> wrote:
Have you used “SAGE_FAT_BINARY=yes”? it is the only way to make a somewhat portable
binary that doesn’t have the problem you describe.

Yes, that works, thanks, I actually just forgot that this time.

Regards,
Jan

Reply all
Reply to author
Forward
0 new messages