Skip to content
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.

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

[oneMKL][BLAS] add out-of-place TRMM and TRSM #523

Merged
merged 4 commits into from
Mar 8, 2024
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
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
295 changes: 288 additions & 7 deletions source/elements/oneMKL/source/domains/blas/trmm.rst
Original file line number Diff line number Diff line change
Expand Up @@ -19,7 +19,9 @@ the matrices in the multiplication is triangular. The argument
``left_right`` determines if the triangular matrix, ``A``, is on the
left of the multiplication (``left_right`` = ``side::left``) or on
the right (``left_right`` = ``side::right``). Depending on
``left_right``. The operation is defined as:
``left_right``.

There are two operations available, an in-place operation and an out-of-place operation. The in-place operation is defined as:

.. math::

Expand All @@ -31,16 +33,28 @@ or

B \leftarrow alpha*B*op(A)

The out-of-place operation is defined as:

.. math::

C \leftarrow alpha*op(A)*B + beta*C

or

.. math::

C \leftarrow alpha*B*op(A) + beta*C

where:

op(``A``) is one of op(``A``) = *A*, or op(``A``) = ``A``\ :sup:`T`,
or op(``A``) = ``A``\ :sup:`H`,

``alpha`` is a scalar,
``alpha`` and ``beta`` are scalars,

``A`` is a triangular matrix, and ``B`` is a general matrix.
``A`` is a triangular matrix, and ``B`` and ``C`` are general matrices.

Here ``B`` is ``m`` x ``n`` and ``A`` is either ``m`` x ``m`` or
Here ``B`` and ``C`` are ``m`` x ``n`` and ``A`` is either ``m`` x ``m`` or
``n`` x ``n``, depending on ``left_right``.

``trmm`` supports the following precisions.
Expand All @@ -59,12 +73,16 @@ Here ``B`` is ``m`` x ``n`` and ``A`` is either ``m`` x ``m`` or
trmm (Buffer Version)
---------------------

**In-place API**
----------------

.. rubric:: Syntax

.. code-block:: cpp

namespace oneapi::mkl::blas::column_major {
void trmm(sycl::queue &queue,
onemkl::side left_right,
onemkl::uplo upper_lower,
onemkl::transpose transa,
onemkl::diag unit_diag,
Expand All @@ -80,6 +98,7 @@ trmm (Buffer Version)

namespace oneapi::mkl::blas::row_major {
void trmm(sycl::queue &queue,
onemkl::side left_right,
onemkl::uplo upper_lower,
onemkl::transpose transa,
onemkl::diag unit_diag,
Expand All @@ -103,7 +122,7 @@ trmm (Buffer Version)
Specifies whether ``A`` is on the left side of the multiplication
(``side::left``) or on the right side (``side::right``). See :ref:`onemkl_datatypes` for more details.

uplo
upper_lower
Specifies whether the matrix ``A`` is upper or lower triangular. See :ref:`onemkl_datatypes` for more details.

trans
Expand Down Expand Up @@ -181,19 +200,149 @@ trmm (Buffer Version)


:ref:`oneapi::mkl::unimplemented<onemkl_exception_unimplemented>`


**Out-of-place API**
--------------------

.. rubric:: Syntax

.. code-block:: cpp

namespace oneapi::mkl::blas::column_major {
void trmm(sycl::queue &queue,
onemkl::side left_right,
onemkl::uplo upper_lower,
onemkl::transpose trans,
onemkl::diag unit_diag,
std::int64_t m,
std::int64_t n,
T alpha,
sycl::buffer<T,1> &a,
std::int64_t lda,
sycl::buffer<T,1> &b,
std::int64_t ldb,
T beta,
sycl::buffer<T,1> &c,
std::int64_t ldc)
}

.. code-block:: cpp

namespace oneapi::mkl::blas::row_major {
void trmm(sycl::queue &queue,
onemkl::side left_right,
onemkl::uplo upper_lower,
onemkl::transpose trans,
onemkl::diag unit_diag,
std::int64_t m,
std::int64_t n,
T alpha,
sycl::buffer<T,1> &a,
std::int64_t lda,
sycl::buffer<T,1> &b,
std::int64_t ldb,
T beta,
sycl::buffer<T,1> &c,
std::int64_t ldc)
}


.. container:: section

.. rubric:: Input Parameters

queue
The queue where the routine should be executed.

left_right
Specifies whether ``A`` is on the left side of the multiplication
(``side::left``) or on the right side (``side::right``). See :ref:`onemkl_datatypes` for more details.

upper_lower
Specifies whether the matrix ``A`` is upper or lower triangular. See :ref:`onemkl_datatypes` for more details.

trans
Specifies op(``A``), the transposition operation applied to matrix ``A``. See :ref:`onemkl_datatypes` for more details.

unit_diag
Specifies whether ``A`` is assumed to be unit triangular (all diagonal elements are 1). See :ref:`onemkl_datatypes` for more details.

m
Specifices the number of rows of ``B``. The value of ``m`` must be at least zero.

n
Specifies the Number of columns of ``B``. The value of ``n`` must be at least zero.

alpha
Scaling factor for matrix-matrix product.

a
Buffer holding input matrix ``A``. Must have size at least
``lda``\ \*\ ``m`` if ``left_right`` = ``side::left`` or ``lda``\ \*\ ``n`` if ``left_right`` = ``side::right``. See :ref:`matrix-storage` for more details.

lda
Leading dimension of ``A``. Must be at least ``m`` if ``left_right`` = ``side::left`` or at least ``n`` if ``left_right`` = ``side::right``. Must be positive.

b
Buffer holding input matrix ``B``. Must have size at least
``ldb``\ \*\ ``n`` if column major layout or at least ``ldb``\ \*\ ``m`` if row major layout is used. See :ref:`matrix-storage` for more details.

ldb
Leading dimension of matrix ``B``. It must be positive and at least ``m`` if column major layout or at least ``n`` if row major layout is used.

beta
Scaling factor for matrix ``C``.

c
Buffer holding input/output matrix ``C``. Size of the buffer must be at least
``ldc``\ \*\ ``n`` if column major layout or at least ``ldc``\ \*\ ``m`` if row major layout is used. See :ref:`matrix-storage` for more details.

ldc
Leading dimension of matrix ``C``. Must be at least ``m`` if column major layout or at least ``n`` if row major layout is used. Must be positive.

.. container:: section

.. rubric:: Output Parameters

c
Output buffer overwritten by ``alpha``\ \*\ op(``A``)\ \*\ ``B`` + ``beta``\ \*\ ``C`` if ``left_right`` = ``side::left`` or ``alpha``\ \*\ ``B``\ \*\ op(``A``) + ``beta``\ \*\ ``C`` if ``left_right`` = ``side::right``.

.. container:: section

.. rubric:: Throws

This routine shall throw the following exceptions if the associated condition is detected. An implementation may throw additional implementation-specific exception(s) in case of error conditions not covered here.

:ref:`oneapi::mkl::invalid_argument<onemkl_exception_invalid_argument>`


:ref:`oneapi::mkl::unsupported_device<onemkl_exception_unsupported_device>`


:ref:`oneapi::mkl::host_bad_alloc<onemkl_exception_host_bad_alloc>`


:ref:`oneapi::mkl::device_bad_alloc<onemkl_exception_device_bad_alloc>`


:ref:`oneapi::mkl::unimplemented<onemkl_exception_unimplemented>`


.. _onemkl_blas_trmm_usm:

trmm (USM Version)
------------------

**In-place API**
----------------

.. rubric:: Syntax

.. code-block:: cpp

namespace oneapi::mkl::blas::column_major {
sycl::event trmm(sycl::queue &queue,
onemkl::side left_right,
onemkl::uplo upper_lower,
onemkl::transpose transa,
onemkl::diag unit_diag,
Expand All @@ -210,6 +359,7 @@ trmm (USM Version)

namespace oneapi::mkl::blas::row_major {
sycl::event trmm(sycl::queue &queue,
onemkl::side left_right,
onemkl::uplo upper_lower,
onemkl::transpose transa,
onemkl::diag unit_diag,
Expand All @@ -235,7 +385,7 @@ trmm (USM Version)
multiplication (``side::left``) or on the right side
(``side::right``). See :ref:`onemkl_datatypes` for more details.

uplo
upper_lower
Specifies whether the matrix ``A`` is upper or lower
triangular. See :ref:`onemkl_datatypes` for more details.

Expand Down Expand Up @@ -328,5 +478,136 @@ trmm (USM Version)

:ref:`oneapi::mkl::unimplemented<onemkl_exception_unimplemented>`

**Out-of-place API**
--------------------

.. rubric:: Syntax

.. code-block:: cpp

namespace oneapi::mkl::blas::column_major {
void trmm(sycl::queue &queue,
onemkl::side left_right,
onemkl::uplo upper_lower,
onemkl::transpose trans,
onemkl::diag unit_diag,
std::int64_t m,
std::int64_t n,
value_or_pointer<T> alpha,
const T *a,
std::int64_t lda,
const T *b,
std::int64_t ldb,
value_or_pointer<T> beta,
T *c,
std::int64_t ldc,
const std::vector<sycl::event> &dependencies = {})
}

.. code-block:: cpp

namespace oneapi::mkl::blas::row_major {
void trmm(sycl::queue &queue,
onemkl::side left_right,
onemkl::uplo upper_lower,
onemkl::transpose trans,
onemkl::diag unit_diag,
std::int64_t m,
std::int64_t n,
value_or_pointer<T> alpha,
const T *a,
std::int64_t lda,
const T *b,
std::int64_t ldb,
value_or_pointer<T> beta,
T *c,
std::int64_t ldc,
const std::vector<sycl::event> &dependencies = {})
}


.. container:: section

.. rubric:: Input Parameters

queue
The queue where the routine should be executed.

left_right
Specifies whether ``A`` is on the left side of the multiplication
(``side::left``) or on the right side (``side::right``). See :ref:`onemkl_datatypes` for more details.

upper_lower
Specifies whether the matrix ``A`` is upper or lower triangular. See :ref:`onemkl_datatypes` for more details.

trans
Specifies op(``A``), the transposition operation applied to matrix ``A``. See :ref:`onemkl_datatypes` for more details.

unit_diag
Specifies whether ``A`` is assumed to be unit triangular (all diagonal elements are 1). See :ref:`onemkl_datatypes` for more details.

m
Specifices the number of rows of ``B``. The value of ``m`` must be at least zero.

n
Specifies the Number of columns of ``B``. The value of ``n`` must be at least zero.

alpha
Scaling factor for matrix-matrix product. See :ref:`value_or_pointer` for more details.

a
Pointer to input matrix ``A``. Must have size at least
``lda``\ \*\ ``m`` if ``left_right`` = ``side::left`` or ``lda``\ \*\ ``n`` if ``left_right`` = ``side::right``. See :ref:`matrix-storage` for more details.

lda
Leading dimension of ``A``. Must be at least ``m`` if ``left_right`` = ``side::left`` or at least ``n`` if ``left_right`` = ``side::right``. Must be positive.

b
Pointer to input matrix ``B``. Must have size at least
``ldb``\ \*\ ``n`` if column major layout or at least ``ldb``\ \*\ ``m`` if row major layout is used. See :ref:`matrix-storage` for more details.

ldb
Leading dimension of matrix ``B``. It must be positive and at least ``m`` if column major layout or at least ``n`` if row major layout is used.

beta
Scaling factor for matrix ``C``. See :ref:`value_or_pointer` for more details.

c
Pointer to input/output matrix ``C``. Must have size at least
``ldc``\ \*\ ``n`` if column major layout or at least ``ldc``\ \*\ ``m`` if row major layout is used. See :ref:`matrix-storage` for more details.

ldc
Leading dimension of matrix ``C``. Must be at least ``m`` if column major layout or at least ``n`` if row major layout is used. Must be positive.

dependencies
List of events to wait for before starting computation, if any.
If omitted, defaults to no dependencies.

.. container:: section

.. rubric:: Output Parameters

c
Pointer to the output matrix, overwritten by ``alpha``\ \*\ op(``A``)\ \*\ ``B`` + ``beta``\ \*\ ``C`` if ``left_right`` = ``side::left`` or ``alpha``\ \*\ ``B``\ \*\ op(``A``) + ``beta``\ \*\ ``C`` if ``left_right`` = ``side::right``.

.. container:: section

.. rubric:: Throws

This routine shall throw the following exceptions if the associated condition is detected. An implementation may throw additional implementation-specific exception(s) in case of error conditions not covered here.

:ref:`oneapi::mkl::invalid_argument<onemkl_exception_invalid_argument>`


:ref:`oneapi::mkl::unsupported_device<onemkl_exception_unsupported_device>`


:ref:`oneapi::mkl::host_bad_alloc<onemkl_exception_host_bad_alloc>`


:ref:`oneapi::mkl::device_bad_alloc<onemkl_exception_device_bad_alloc>`


:ref:`oneapi::mkl::unimplemented<onemkl_exception_unimplemented>`

**Parent topic:** :ref:`blas-level-3-routines`
Loading
Loading