[Nek5000-users] Nek5000-users Digest, Vol 50, Issue 6
nek5000-users at lists.mcs.anl.gov
nek5000-users at lists.mcs.anl.gov
Thu Apr 18 20:17:04 CDT 2013
Paul,
Thanks for your help; both of these worked... Last real machine I ran
on was a Cray YMP, so I'm learning a lot about MPI. I ended up
installing MPI on an SMP machine; compiled NEK for MPI with the
wrappers; and ran it on multiple cores. I assume not quite as good as
OpenMP, but worked great for my purposes...
Thanks,
Murph
On 4/16/2013 1:00 PM, nek5000-users-request at lists.mcs.anl.gov wrote:
> On Mon, Apr 15, 2013 at 11:14 AM, <nek5000-users at lists.mcs.anl.gov
> <mailto:nek5000-users at lists.mcs.anl.gov>> wrote:
>
>
> Hi Murph,
>
> A couple of options here:
>
> 1) Try recompiling with G = "-mcmodel=medium" in makenek.
>
> 2) Use more than one core so that lelt can remain the same
> but lelg can be as large as lelt*P, where P is the number
> of cores you'd like to use.
>
> Raw email is the standard interface as far as I know...
>
> hth,
>
> Paul
>
>
> On Mon, 15 Apr 2013, nek5000-users at lists.mcs.anl.gov
> <mailto:nek5000-users at lists.mcs.anl.gov> wrote:
>
> Hi All,
>
> Trying to run some initial test cases of a problems I'm
> working on:
> - Currently on single node (SMP Linux)
> - MPI turned off (IFMPI="false")
> - DOF at 5 (lx1=6, lxd=9, lelx=8)
>
> If I run with lp=1,lelt=2160, lelg=2160, for a 12x12x15 mesh,
> it compiles and runs fine.
>
> If I up it to lp=1,lelt=4500,lelg=4500, for a 15x15x20 mesh,
> it crashes the compile with:
>
> obj/drive2.o: In function `plan4_vol_':
> drive2.f:(.text+0x82f): relocation truncated to fit:
> R_X86_64_32S against symbol `scrvh_' defined in COMMON section
> in obj/hsmg.o
> drive2.f:(.text+0x837): relocation truncated to fit:
> R_X86_64_32S against symbol `scrvh_' defined in COMMON section
> in obj/hsmg.o
> obj/drive2.o: In function `plan3_vol_':
> drive2.f:(.text+0xac5): relocation truncated to fit:
> R_X86_64_32S against symbol `scrvh_' defined in COMMON section
> in obj/hsmg.o
> drive2.f:(.text+0xacd): relocation truncated to fit:
> R_X86_64_32S against symbol `scrvh_' defined in COMMON section
> in obj/hsmg.o
> obj/drive2.o: In function `plan2_vol_':
> drive2.f:(.text+0xeb8): relocation truncated to fit:
> R_X86_64_32S against symbol `scrvh_' defined in COMMON section
> in obj/hsmg.o
> drive2.f:(.text+0xec0): relocation truncated to fit:
> R_X86_64_32S against symbol `scrvh_' defined in COMMON section
> in obj/hsmg.o
> obj/drive2.o: In function `dofcnt_':
> drive2.f:(.text+0x170f): relocation truncated to fit:
> R_X86_64_PC32 against symbol `scrns_' defined in COMMON
> section in obj/convect.o
> obj/drive2.o: In function `settime_':
> drive2.f:(.text+0x7cc9): relocation truncated to fit:
> R_X86_64_PC32 against symbol `cprint_' defined in COMMON
> section in obj/navier1.o
> drive2.f:(.text+0x7cef): relocation truncated to fit:
> R_X86_64_PC32 against symbol `cprint_' defined in COMMON
> section in obj/navier1.o
> drive2.f:(.text+0x7d02): relocation truncated to fit:
> R_X86_64_PC32 against symbol `cprint_' defined in COMMON
> section in obj/navier1.o
> obj/plan4.o: In function `crespsp_':
> plan4.f:(.text+0xb23): additional relocation overflows omitted
> from the output
> collect2: ld returned 1 exit status
> make: *** [nek5000] Error 1
>
> Just wondering if this is due to a true limit with the
> platform/word size, or if I'm screwing something up...
>
> Thanks,
> Murph
>
> PS - Is there any other interface to this list other than raw
> email?
>
>
>
> --
>
> *Murphy Leo O'Dea, PE"Murph"*
>
> PhD Student, Mechanical Engineering
>
> /School Of Engineering And Computer Science/
>
> /Oakland University/
>
> mlodea at oakland.edu <mailto:mlodea at oakland.edu>
> <mailto:mlodea at oakland.edu <mailto:mlodea at oakland.edu>>
>
> http://www.secs.oakland.edu/~mlodea/
> <http://www.secs.oakland.edu/%7Emlodea/>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/nek5000-users/attachments/20130418/5ae46c9a/attachment.html>
More information about the Nek5000-users
mailing list