Currently ADDA in sparse MPI mode (and only when SSE3 is _not_ used), when run
on several processors, incorrectly processes (indexes) material information for
inhomogeneous particles. The problem was in function CcMul in sparse_ops.h.
This bug will be fixed during the cleaning/improvement of the sparse code,
which I am now working on. Hopefully, I will commit it during a week. So this
issue is more for caution with regards to the previous use of the sparse mode.
Original issue reported on code.google.com by yurkin on 4 Feb 2013 at 6:13
Original issue reported on code.google.com by
yurkin
on 4 Feb 2013 at 6:13