Skip to content

typos #2

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
langou wants to merge 1 commit into
base: main
Choose a base branch
from
Open

typos #2

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 4 additions & 4 deletions sparseblas.tex
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -766,7 +766,7 @@ \subsection{Review of existing API designs SpMV }
y += beta*A*x
GrB_mxm (y, GrB_NULL, GrB_PLUS_FP64. GrB_PLUS_TIMES_SEMIRING_FP64, A, x, GrB_NULL);
\end{lstlisting}
\item ALP/GraphBLAS \cplusplus (minimum number of containers, optimised by nonblocking execution)
\item ALP/GraphBLAS \cplusplus (minimum number of containers, optimized by nonblocking execution)
\begin{lstlisting}
grb::Semiring< grb::operators::add<double>, grb::operators::mul<double>,
grb::identities::zero, grb::identities::one > plusTimes;
Expand Down Expand Up @@ -1091,7 +1091,7 @@ \subsubsection{Multi-Stage API for Operations with unknown Output}
\todo[inline]{C is not an obj, but the CSR view}
\begin{minted}{c++}
using namespace spblas;
// csr_wrapper<float> A,B,D filled elsewhwere
// csr_wrapper<float> A,B,D filled elsewhere
// csr_wrapper<float> C(c_nrows, c_ncols);
// auto A_obj = matrix_opt(A, allocator);
// auto B_obj = matrix_opt(B, allocator);
Expand Down Expand Up @@ -1124,7 +1124,7 @@ \subsubsection{Multi-Stage API for Operations with unknown Output}
\todo[inline]{C is not an obj, but the CSR view}
\begin{minted}{c++}
using namespace spblas;
// csr_wrapper<float> A,B,D filled elsewhwere
// csr_wrapper<float> A,B,D filled elsewhere
// csr_wrapper<float> C(c_nrows, c_ncols);
// auto A_obj = matrix_opt(A, allocator);
// auto B_obj = matrix_opt(B, allocator);
Expand Down Expand Up @@ -1575,7 +1575,7 @@ \subsection{Handling Asynchronous $\alpha$/$\beta$ Scalars}

Suppose that for SpMV operation $y=\alpha op(A)x + \beta y$, that the $\alpha$ and/or $\beta$ parameters are computed as part of the higher level algorithm, (e.g. as part of a conjugate gradient iteration) and hence may also be asynchronously defined on the device. We should provide a method of supporting such use cases in our designs. A simple approach is to switch from a $DataType \alpha$ argument to a $DataType * p\_alpha$ argument.

Each extension language provides a method of determining where such a pointer lives (must be device accessible) and our APIs should internally query this, we should assume it is dynamically being set and only de-reference the value once all dependencies are satisfied, ie during the kernel execution itself.
Each extension language provides a method of determining where such a pointer lives (must be device accessible) and our APIs should internally query this, we should assume it is dynamically being set and only de-reference the value once all dependencies are satisfied, i.e. during the kernel execution itself.

Using the \verb|scaled(alpha, matrix_view)| approach for setting scalar factors on a matrix or other object, we may choose to have overloads for a constant scalar or asynchronous pointers to a scalar which should be handled appropriately internally. In the out-of-order asynchronous approach, the user has the responsibility to ensure any events related to the setting of alpha value asynchronously should be provided as input to the function call. For in-order asynchronous cases, there are no additional requirements on the user's side. In all cases, the library has the responsibility to ensure that the pointer to alpha is only de-referenced in the asynchronous kernel.

Expand Down