CosmoCoffee

[Ian Sollom]

I've been trying to get the simlens demo to run on Cosmos but it keeps crashing. If Anthony, or anybody else can help I'd be most grateful!

Hopefully all the information that might help is below.

At Andrey's advice I alter the MakeFile to compile:

F90C = ifort
FFLAGS = -g -O2 -fpp -DMPIPIX -DMPI -heap-arrays -I
LAPACKL = -lhealpix -lcfitsio -lmpi -lmkl_lapack -lmkl -lguide -lpthread

On running the job I get the following feedback (with the debug messages turned on in spin_alm_tools.f90) before it dies:

mpi_division = 3
0 row end = 1525
1 row end = 2260
2 row end = 3178
3 row end = 4096
Random seeds: 28735, 25107 rand_inst: 0
ALM2GRADIENTMAP: Sending to farm
Send messages ALM2GRADIENTMAP
Got message 1 ALM2GRADIENTMAP
Got message 2 ALM2GRADIENTMAP
Got message 3 ALM2GRADIENTMAP
ALM2GRADIENTMAP Got alm 0 6.528219999745488E−002
ALM2GRADIENTMAP Got alm 1 6.464845000300556E−002
ALM2GRADIENTMAP Got alm 2 7.294410001486540E−002
ALM2GRADIENTMAP Got alm 3 7.287950001773424E−002
ALM2GRADIENTMAP Gather 3
ALM2GRADIENTMAP Gather 2
ALM2GRADIENTMAP Gather 1
ALM2GRADIENTMAP Gather 0
ALM2GRADIENTMAP Wait at gather 1 7.53443310002331
ALM2GRADIENTMAP Wait at gather 3 10.4921001500043
ALM2GRADIENTMAP Wait at gather 2 10.4067222500162
ALM2GRADIENTMAP Wait at gather 0 0.437246449990198
ALM2GRADIENTMAP Time: 77.7274020499899
SCALALM2LENSEDMAPINTERPCYL: Sending to farm
Send messages SCALALM2LENSEDMAPINTERPCYL
Got message 1 SCALALM2LENSEDMAPINTERPCYL
Got message 2 SCALALM2LENSEDMAPINTERPCYL
Got message 3 SCALALM2LENSEDMAPINTERPCYL
SCALALM2LENSEDMAPINTERPCYL: Got alm 0 4.447034999611787E−002
cyl_interp_init interpolation with nside = 3072
SCALALM2LENSEDMAPINTERPCYL: Got alm 1 4.443184999399818E−002
SCALALM2LENSEDMAPINTERPCYL: Got alm 2 4.443914999137633E−002
SCALALM2LENSEDMAPINTERPCYL: Got alm 3 4.461139999330044E−002
cyl_interp_init Scattered grad_phi 1 0.186948149988893
cyl_interp_init Scattered grad_phi 3 0.190261549985735
cyl_interp_init Scattered grad_phi 0 0.204508549999446
cyl_interp_init Scattered grad_phi 2 0.204464100010227
1 borders = 4 3
0 borders = 3 2
3 borders = 2 3
2 borders = 4 4


Finally, Andrey has also looked at the crash data:

Reading symbols from /local/cosmos-tmp/ak591/ifs24/LensPix/simlens...done.
Core file produced from executable simlens
Initial part of arglist: ./simlens params.ini
Thread terminated at PC 0x4000000000099c21 by signal SIGSEGV
(idb) where
#0 0x4000000000099c21 in SPINALM_TOOLS::scalalm2lensedmapinterpcyl (h=<no value>, inlmax=0,
alm=Info: symbol alm is defined but not allocated (optimized away)
<no value>, grad_phi_map=(...), map=(...), nside_factor=2.54353118) at
/local/cosmos-tmp/ak591/ifs24/LensPix/spin_alm_tools.F90:5063
#1 0x4000000000042db0 in SPINALM_TOOLS::messageloop (h=<no value>) at
/local/cosmos-tmp/ak591/ifs24/LensPix/spin_alm_tools.F90:7465
#2 0x40000000000f93e0 in SPINALM_TOOLS::healpixinit (h=<no value>, nside=0, lmax=1082673323,
haspol=.FALSE., w8dir=(...), method=1076021559, .tmp.W8DIRlen_V374=−9223372036854775808) at
/local/cosmos-tmp/ak591/ifs24/LensPix/spin_alm_tools.F90:484
#3 0x400000000013f090 in simlenscmb () at
/local/cosmos-tmp/ak591/ifs24/LensPix/SimLens.f90:80
#4 0x4000000000005f10 in main () in /local/cosmos-tmp/ak591/ifs24/LensPix/simlens
#5 0x2000000008d23c20 in __libc_start_main () in /lib/libc−2.4.so
#6 0x4000000000005d40 in _start () in /local/cosmos-tmp/ak591/ifs24/LensPix/simlens

(some of the line references above may be different to the original as I've inserted a few comments in spin_alm_tools.f90 - though they won't be very far off)

[Antony Lewis]

I don't know, but a good first check might be to compile without optimizations in case it is a compiler bug.