Internal compiler error - CCSM / PGI7.0-3 x86_64

Accelerated Computing HPC Compilers Legacy PGI Compilers
1

Hi Mat,

I had been testing out CCSM (v3.0.1beta14) with PGI 7.0-3 on x86-64 and when I compiled for my first test everything worked fine, but I wanted to reduce the processor count down to something smaller and I’ve ended up with an internal compiler error. The error is the following:

mpif90 -I. -I/opt/new/netcdf/include -I/system/brian//smallrun/lib/include -I/opt/new/mpich2-1.0.5p4/include -I/opt/new/netcdf/include/ -I. -I/system/brian/smallrun/SourceMods/src.clm -I/system/brian/ccsm/ccsm3_0_1_beta14/models/lnd/clm2/src/main -I/system/brian/ccsm/ccsm3_0_1_beta14/models/lnd/clm2/src/biogeophys -I/system/brian/ccsm/ccsm3_0_1_beta14/models/lnd/clm2/src/ecosysdyn -I/system/brian/ccsm/ccsm3_0_1_beta14/models/lnd/clm2/src/riverroute -I/system/brian/ccsm/ccsm3_0_1_beta14/models/lnd/clm2/src/biogeochem -I/system/brian/ccsm/ccsm3_0_1_beta14/models/lnd/clm2/src/mksrfdata -I/system/brian/ccsm/ccsm3_0_1_beta14/models/csm_share/cpl -I/system/brian/ccsm/ccsm3_0_1_beta14/models/csm_share/shr -I/system/brian/ccsm/ccsm3_0_1_beta14/models/utils/timing -c -r8 -i4 -Kieee -Mrecursive -Mdalign -Mextend -Mfree -DLINUX -DPGF90 -DNO_SHR_VMATH -DLINUX /system/brian/ccsm/ccsm3_0_1_beta14/models/lnd/clm2/src/main/time_manager.F90
PGF90-S-0000-Internal compiler error. sym_of_ast: unexpected ast 11 (/system/brian/ccsm/ccsm3_0_1_beta14/models/lnd/clm2/src/main/time_manager.F90: 106)
PGF90-S-0000-Internal compiler error. sym_of_ast: unexpected ast 11 (/system/brian/ccsm/ccsm3_0_1_beta14/models/lnd/clm2/src/main/time_manager.F90: 106)
Lowering Error: array extnt is not a symbol for datatype 59
Lowering Error: array extnt is not a symbol for datatype 62
PGF90-F-0000-Internal compiler error. Errors in Lowering 2 (/system/brian/ccsm/ccsm3_0_1_beta14/models/lnd/clm2/src/main/time_manager.F90: 106)
PGF90/x86-64 Linux 7.0-3: compilation aborted
gmake: *** [time_manager.o] Error 2


As I said, it worked fine for the other processor configuration, and the line in question has nothing to do with processor counts as far as I can tell. (It’s a ‘contains’ line, actually. )

I can certainly work around this for now, but I wanted to bring it to your attention – what other information can I provide you with to help you guys figure out why this is happening?

Thanks,

  • Brian
2

Hi Brian,

Thanks for the report. What we need to do next is try to recreate the problem here. Do you think this issue only occurs in the 3.0.1 beta version or could I recreate it using the 3.0 version? What are the exact configuration and build steps you did to get this error?

Thanks,
Mat