Somehow I missed that in 2021 support for asymmetric EO-precon was added to QUDA for all Wilson(-clover)-type operators.
We could therefore simplify the logic that is currently used and unify all EO-preconditioned solves to be performed consistently with tmLQCD.
Somehow I missed that in 2021 support for asymmetric EO-precon was added to QUDA for all Wilson(-clover)-type operators.
We could therefore simplify the logic that is currently used and unify all EO-preconditioned solves to be performed consistently with tmLQCD.