Closed djabaay closed 4 years ago
Is this the branch that was reverted that now needs to be remerged? Our pipeline is running DBC now.
@aarograh and @stimpsonsg want me to turn on DBC in the travis-ci for this?
Seems like a prudent move. I think we always want DBC on for automated pre-push testing.
Yep...it needs to be on here and both the ORNL and UM CIs.
@stimpsonsg ok. I'll try and knock that out tonight and create a separate PR. We'll hold this up, until that's merged.
Ran heavies with DBC enabled:
99% tests passed, 6 tests failed out of 777
Subproject Time Summary:
CTeuchos = 1.21 sec*proc (4 tests)
ForTeuchos = 1.42 sec*proc (4 tests)
Futility = 29.74 sec*proc (60 tests)
MPACT_API = 67.41 sec*proc (4 tests)
MPACT_Drivers = 0.94 sec*proc (4 tests)
MPACT_exe = 267780.13 sec*proc (522 tests)
MPACT_libs = 998.45 sec*proc (179 tests)
Label Time Summary:
BASIC = 29.74 sec*proc (60 tests)
CONTINUOUS = 29.74 sec*proc (60 tests)
HEAVY = 29.74 sec*proc (60 tests)
NIGHTLY = 29.74 sec*proc (60 tests)
Total Test time (real) = 10461.97 sec
The following tests FAILED:
322 - MPACT_exe_testValid_excore_detector_2D_core (Failed)
405 - MPACT_exe_testValid_transient_4-mini_3D (Failed)
406 - MPACT_exe_testValid_transient_4-mini_1GAccel (Failed)
407 - MPACT_exe_testValid_transient_4-mini_3D_SCRAM (Failed)
412 - MPACT_exe_testValid_steadystate_4mini_crw (Failed)
413 - MPACT_exe_testValid_transient_4mini_crw (Failed)
These failures are from enabling DBC where during the solve there are negative fluxes encountered for the 4-mini cases.
DBC Failure: .NOT. ANY(tmpTS%phis(:,ig) < 0.0_SRK) in /home/djabaay/MPACT/MPACT_libs/CoreSolvers/src/FixedSrcSolver.f90 on line 953: process 14 of 16
#0 0x7FDB12B1ABF7
#1 0xEE3F9E in __futility_dbc_MOD_dbc_fail
#2 0xC289EC in __fixedsrcsolver_MOD_step_fss_default
#3 0xC296AF in __fixedsrcsolver_MOD_solve_fss_default
#4 0xC3D0B5 in __eigensolver_MOD_step_eigensolvertype
#5 0xC41327 in __feedbacksolver_MOD_step_feedbacksolvertype
#6 0xC41555 in __feedbacksolver_MOD_solve_feedbacksolvertype
#7 0xC52147 in __multistatesolver_MOD_solve_multistatesolvertype
#8 0x417ADC in __mpact_driver_MOD_solve_mpact at Driver.f90:?
#9 0x42A482 in MPACT_DRIVER_solve
Note: The following floating-point exceptions are signalling: IEEE_INVALID_FLAG
STOP 2
--------------------------------------------------------------------------------
TEST_2: Return code = 2
TEST_2: Pass criteria = Zero return code [FAILED]
TEST_2: Result = FAILED
================================================================================
TEST_3
Running: "/home/djabaay/MPACT_BIN/MPACT_exe/src/MPACTdiff.exe" "/home/djabaay/MPACT/MPACT_exe/tests/regression_tests/solution_verification/transient/VERA/3D/transient_4-mini_3D.h5.gold"↪› "transient_4-mini_3D.h5" "/EXPECTED/" "/" "pin_isotopes_U-235|pin_isotopes_B-10"
The excore detector test is also a known failing test on the detector response that needs to be updated.
honestly, I'm not sure that the phis > 0.0 should actually be DBC...it can happen and not fail everything.
@stimpsonsg I agree. @djabaay did you build and run under VERA or MPACT? For this I don't think its necessary to run under VERA.
given that the excore_detector_2D_core case ran, seems indicative it was standalone MPACT not in VERA
Correct, MPACT + CTF
Description: Adding %wtfactor attribute used to scale the %wts in the input. The default is 1.0. Adding optional argument to %metrics() routine for the max number of source regions ratio for the given decomposition.