However, for the purpose of comparing data serially generated by our regression tests (i.e. accepted results) with data generated by a parallel run it'd be good to be able to do something like
gkyl comparefiles -da directoryA/ -db directoryB/
and it'll compare files just like our regression system does, reporting relative differences. This is a useful step on the way to checking results with MPI.
A more specific example, suppose I generate data for the vm-lbo/rt-lboRelax-1x1v-p1 test with
gkyl runregression run create -r vm-lbo/rt-lboRelax-1x1v-p1.lua
and make some changes that primarily affect MPI, then i'd like to check if a parallel run produces the same results with
At the moment comparefiles works like
gkyl comparefiles -a fileA.bp -b fileB.bpHowever, for the purpose of comparing data serially generated by our regression tests (i.e. accepted results) with data generated by a parallel run it'd be good to be able to do something like
gkyl comparefiles -da directoryA/ -db directoryB/and it'll compare files just like our regression system does, reporting relative differences. This is a useful step on the way to checking results with MPI.
A more specific example, suppose I generate data for the vm-lbo/rt-lboRelax-1x1v-p1 test with
gkyl runregression run create -r vm-lbo/rt-lboRelax-1x1v-p1.luaand make some changes that primarily affect MPI, then i'd like to check if a parallel run produces the same results with