|
|
|
@ -601,7 +601,7 @@ An integer value; 0 means no error has been detected.
|
|
|
|
the assembled matrix may substantially improve performance in the
|
|
|
|
the assembled matrix may substantially improve performance in the
|
|
|
|
matrix build phase, as it will reduce or eliminate the need for
|
|
|
|
matrix build phase, as it will reduce or eliminate the need for
|
|
|
|
(potentially multiple) data reallocations;
|
|
|
|
(potentially multiple) data reallocations;
|
|
|
|
\item Using \verb|psb_matbld_remote_| will cause an overhead at
|
|
|
|
\item Using \verb|psb_matbld_remote_| is likely to cause a runtime overhead at
|
|
|
|
assembly time;
|
|
|
|
assembly time;
|
|
|
|
\end{enumerate}
|
|
|
|
\end{enumerate}
|
|
|
|
|
|
|
|
|
|
|
|
@ -712,8 +712,7 @@ An integer value; 0 means no error has been detected.
|
|
|
|
\item The input data can be passed in either COO or CSR formats;
|
|
|
|
\item The input data can be passed in either COO or CSR formats;
|
|
|
|
\item In COO format the coefficients to be inserted are represented by
|
|
|
|
\item In COO format the coefficients to be inserted are represented by
|
|
|
|
the ordered triples $ia(i),ja(i),val(i)$, for $i=1,\dots,nz$;
|
|
|
|
the ordered triples $ia(i),ja(i),val(i)$, for $i=1,\dots,nz$;
|
|
|
|
these triples should belong to the current process, i.e. $ia(i)$
|
|
|
|
these triples are arbitrary;
|
|
|
|
should be one of the local indices, but are otherwise arbitrary;
|
|
|
|
|
|
|
|
\item In CSR format the coefficients to be inserted for each input row $i=1,nr$
|
|
|
|
\item In CSR format the coefficients to be inserted for each input row $i=1,nr$
|
|
|
|
are represented by the ordered triples $(i+irw-1),ja(j),val(j)$, for
|
|
|
|
are represented by the ordered triples $(i+irw-1),ja(j),val(j)$, for
|
|
|
|
$j=irp(i),\dots,irp(i+1)-1$;
|
|
|
|
$j=irp(i),\dots,irp(i+1)-1$;
|
|
|
|
@ -724,8 +723,11 @@ An integer value; 0 means no error has been detected.
|
|
|
|
may be split into as many calls as desired (even in the CSR format);
|
|
|
|
may be split into as many calls as desired (even in the CSR format);
|
|
|
|
\item Coefficients from different rows may also be mixed up freely
|
|
|
|
\item Coefficients from different rows may also be mixed up freely
|
|
|
|
in a single call, according to the application needs;
|
|
|
|
in a single call, according to the application needs;
|
|
|
|
\item Any coefficients from matrix rows not owned by the calling
|
|
|
|
\item Coefficients from matrix rows not owned by the calling
|
|
|
|
process are silently ignored;
|
|
|
|
process are treated according to the value of \verb|buildmode|
|
|
|
|
|
|
|
|
specified at allocation time; if
|
|
|
|
|
|
|
|
\verb|buildmode| was chosen as \verb|psb_matbld_remote_| the library
|
|
|
|
|
|
|
|
will keep track of them, otherwise they are silently ignored;
|
|
|
|
\item If the descriptor is in the assembled state, then any entries in
|
|
|
|
\item If the descriptor is in the assembled state, then any entries in
|
|
|
|
the sparse matrix that would generate additional communication
|
|
|
|
the sparse matrix that would generate additional communication
|
|
|
|
requirements are ignored;
|
|
|
|
requirements are ignored;
|
|
|
|
@ -739,7 +741,7 @@ An integer value; 0 means no error has been detected.
|
|
|
|
\clearpage\subsection{psb\_spasb --- Sparse matrix assembly routine}
|
|
|
|
\clearpage\subsection{psb\_spasb --- Sparse matrix assembly routine}
|
|
|
|
|
|
|
|
|
|
|
|
\begin{verbatim}
|
|
|
|
\begin{verbatim}
|
|
|
|
call psb_spasb(a, desc_a, info, afmt, upd, mold)
|
|
|
|
call psb_spasb(a, desc_a, info [, afmt, upd, mold])
|
|
|
|
\end{verbatim}
|
|
|
|
\end{verbatim}
|
|
|
|
|
|
|
|
|
|
|
|
\begin{description}
|
|
|
|
\begin{description}
|
|
|
|
@ -808,8 +810,9 @@ An integer value; 0 means no error has been detected.
|
|
|
|
at allocation time, contributions defined on the current process but
|
|
|
|
at allocation time, contributions defined on the current process but
|
|
|
|
belonging to a remote process will be handled accordingly. This is
|
|
|
|
belonging to a remote process will be handled accordingly. This is
|
|
|
|
most likely to occur in finite element applications, with
|
|
|
|
most likely to occur in finite element applications, with
|
|
|
|
\verb|dupl=psb_dupl_add_|; note that the descriptor will be updated,
|
|
|
|
\verb|dupl=psb_dupl_add_|; it is necessary to check for possible
|
|
|
|
and there will be a runtime overhead.
|
|
|
|
updates needed in the descriptor, hence there will be a runtime
|
|
|
|
|
|
|
|
overhead.
|
|
|
|
\end{enumerate}
|
|
|
|
\end{enumerate}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@ -1007,7 +1010,7 @@ Scope: {\bf global}.\\
|
|
|
|
Type: {\bf optional}.\\
|
|
|
|
Type: {\bf optional}.\\
|
|
|
|
Intent: {\bf in}.\\
|
|
|
|
Intent: {\bf in}.\\
|
|
|
|
Specified as: integer, possible values: \verb|psb_dupl_ovwrt_|,
|
|
|
|
Specified as: integer, possible values: \verb|psb_dupl_ovwrt_|,
|
|
|
|
\verb|psb_dupl_add_|, \verb|psb_dupl_err_|.
|
|
|
|
\verb|psb_dupl_add_|; \verb|psb_dupl_err_| has no effect.
|
|
|
|
\item[buildmode] Whether to keep track of matrix entries that do not
|
|
|
|
\item[buildmode] Whether to keep track of matrix entries that do not
|
|
|
|
belong to the current process.\\
|
|
|
|
belong to the current process.\\
|
|
|
|
Scope: {\bf global}.\\
|
|
|
|
Scope: {\bf global}.\\
|
|
|
|
@ -1032,6 +1035,11 @@ Intent: {\bf out}.\\
|
|
|
|
An integer value; 0 means no error has been detected.
|
|
|
|
An integer value; 0 means no error has been detected.
|
|
|
|
\end{description}
|
|
|
|
\end{description}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
{\par\noindent\large\bfseries Notes}
|
|
|
|
|
|
|
|
\begin{enumerate}
|
|
|
|
|
|
|
|
\item Using \verb|psb_matbld_remote_| is likely to cause a runtime overhead at
|
|
|
|
|
|
|
|
assembly time;
|
|
|
|
|
|
|
|
\end{enumerate}
|
|
|
|
|
|
|
|
|
|
|
|
%
|
|
|
|
%
|
|
|
|
%% psb_ins %%
|
|
|
|
%% psb_ins %%
|
|
|
|
@ -1140,7 +1148,22 @@ Type: {\bf required} \\
|
|
|
|
Intent: {\bf out}.\\
|
|
|
|
Intent: {\bf out}.\\
|
|
|
|
An integer value; 0 means no error has been detected.
|
|
|
|
An integer value; 0 means no error has been detected.
|
|
|
|
\end{description}
|
|
|
|
\end{description}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
{\par\noindent\large\bfseries Notes}
|
|
|
|
|
|
|
|
\begin{enumerate}
|
|
|
|
|
|
|
|
\item On entry to this routine the descriptor must be in the
|
|
|
|
|
|
|
|
assembled state, i.e. \verb|psb_cdasb| must already have been called.
|
|
|
|
|
|
|
|
\item If the \verb|buildmode=psb_matbld_remote_| value was specified
|
|
|
|
|
|
|
|
at allocation time, contributions defined on the current process but
|
|
|
|
|
|
|
|
belonging to a remote process will be handled accordingly. This is
|
|
|
|
|
|
|
|
most likely to occur in finite element applications, with
|
|
|
|
|
|
|
|
\verb|dupl=psb_dupl_add_|.
|
|
|
|
|
|
|
|
\end{enumerate}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
%
|
|
|
|
%
|
|
|
|
|
|
|
|
|
|
|
|
%% psb_free %%
|
|
|
|
%% psb_free %%
|
|
|
|
%
|
|
|
|
%
|
|
|
|
\clearpage\subsection{psb\_gefree --- Frees a dense matrix}
|
|
|
|
\clearpage\subsection{psb\_gefree --- Frees a dense matrix}
|
|
|
|
|
Salvatore Filippone
3 years ago