Check benchmark differences on non-envinf1 runner
2D_Adaptive_dt_ThermalConvection_constviscosityStaggeredScheme
was disabled in !3228 (merged) when not running on envinf1
. The following happens when the benchmarks is enabled and the CI runner is not envinf1
:
Comparing data array `p_ref' from file
`/datadrive/gitlab/builds/ogs/ogs/Tests/Data/Parabolic/HT/StaggeredCoupling/ConstViscosity/square_5500x5500.vtu'
to data array `p' from file
`/datadrive/gitlab/builds/ogs/ogs/build/Tests/Data/Parabolic/HT/StaggeredCoupling/ConstViscosity/ConstViscosityThermalConvectionStaggeredAdaptive_dt_ts_141_t_50000000000.000000.vtu'.
Computed difference between data arrays:
abs l1 norm = [9.658762992946957e+03]
abs l2-norm^2 = [2.357046747437178e+07]
abs l2-norm = [4.854942581985062e+03]
abs maximum norm = [4.509302689506032e+03]
rel l1 norm = [1.948025696557644e-02]
rel l2-norm^2 = [1.567145865233648e-04]
rel l2-norm = [1.567145865233648e-04]
rel maximum norm = [1.128721724847964e-02]
Absolute and relative error (maximum norm) are larger than the
corresponding thresholds 1.000000000000000e-03 and 2.000000000000000e-03.
Similar problem for LARGE_2D_ReactiveMassTransport_Phreeqc_KineticReactantBlockTest_AllAsComponents
:
Comparing data array `Synthetica' from file
`/datadrive/gitlab/builds/ogs/ogs/Tests/Data/Parabolic/ComponentTransport/ReactiveTransport/KineticReactant_AllAsComponents/KineticReactant2_2d_ts_8_t_800.000000_expected.vtu'
to data array `Synthetica' from file
`/datadrive/gitlab/builds/ogs/ogs/build/Tests/Data/Parabolic/ComponentTransport/ReactiveTransport/KineticReactant_AllAsComponents/KineticReactant2_2d_ts_8_t_800.000000.vtu'.
Computed difference between data arrays:
abs l1 norm = [2.603111029932603e-03]
abs l2-norm^2 = [1.285736677413423e-09]
abs l2-norm = [3.585717051599893e-05]
abs maximum norm = [1.158152757196884e-06]
rel l1 norm = [5.245739331803091e-03]
rel l2-norm^2 = [5.221383271381814e-09]
rel l2-norm = [5.221383271381814e-09]
rel maximum norm = [2.334025993049136e-06]
Absolute and relative error (maximum norm) are larger than the
corresponding thresholds 1.000000000000000e-10 and 1.000000000000000e-16.
-
Check what is wrong and re-enable the benchmark for all runners.