Closed dmcdougall closed 2 months ago
@jxy Just to clarify -- the error reported here is not the error message you were seeing with your failures on MI300A, correct?
Thanks @dmcdougall. This is the issue I mentioned on the call the other day that sneaked thru CI.
Ok. Sounds like this is separate to XY's issue, then. Thanks.
@dmcdougall Right. The issues I saw on mi300a seem to be real failures.
These tests
dslash_domain-wall_policytune dslash_domain-wall-4d_sym_policytune dslash_domain-wall-4d_asym_policytune dslash_mobius-eofa_sym_policytune dslash_mobius-eofa_asym_policytune invert_test_domain_wall invert_test_domain_wall_4d_sym invert_test_domain_wall_4d_asym invert_test_mobius_eofa_sym invert_test_mobius_eofa_asym eigensolve_test_ndeg_twisted_mass_sym eigensolve_test_domain_wall eigensolve_test_domain_wall_4d_sym eigensolve_test_domain_wall_4d_asym eigensolve_test_mobius_eofa_sym eigensolve_test_mobius_eofa_asym
all failed with an error message that looks like
--dslash-type: not in {wilson,clover,clover-hasenbusch-twist,twisted-mass,twisted-clover,staggered,asqtad,hisq,domain_wall,domainwall 4d,mobius,mobius_eofa,laplace}
This commit fixed up the expected name (dash versus underscore) for the dslash-type parameter.