CodeQL / Analyze (actions) (push) Canceled after 0s
CodeQL / Analyze (python) (push) Canceled after 0s
pre-commit / pre-commit (push) Canceled after 0s
Multi-Arch CI / setup (push) Canceled after 0s
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Setup (Multi-Arch) / setup (push) Canceled after 0s
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Multi-Arch CI / CI Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
189 lines
9.0 KiB
ReStructuredText
189 lines
9.0 KiB
ReStructuredText
.. meta::
|
|
:description: hipBLAS documentation and API reference library
|
|
:keywords: hipBLAS, rocBLAS, BLAS, ROCm, API, Linear Algebra, documentation
|
|
|
|
.. _deprecations:
|
|
|
|
********************************************************************
|
|
Deprecations by version
|
|
********************************************************************
|
|
|
|
This topic lists the deprecated hipBLAS API elements by release, from newest to oldest.
|
|
|
|
Removed in hipBLAS 3.0
|
|
**********************
|
|
``hipblasDatatype_t`` was replaced with ``hipDataType`` in hipBLAS 3.0. All functions which previously used ``hipblasDatatype_t`` now
|
|
use ``hipDataType`` for every instance, except for as the ``compute_type`` in GEMM functions, in which ``hipblasComputeType_t`` is now used.
|
|
|
|
``hipblasComplex`` and ``hipblasDoubleComplex`` were replaced with ``hipComplex`` and ``hipDoubleComplex`` in hipBLAS 3.0.
|
|
|
|
Announced in hipBLAS 2.0
|
|
************************
|
|
|
|
hipblasDatatype_t will be replaced with hipDataType
|
|
=====================================================
|
|
Use of ``hipblasDatatype_t`` will give deprecation warnings telling you to compile with ``-DHIPBLAS_V2``
|
|
and to use ``hipDataType`` instead. All functions which currently use ``hipblasDatatype_t`` are therefore deprecated as well
|
|
and will be replaced with functions which use ``hipDataType`` in the place of ``hipblasDatatype_t``. These functions include:
|
|
``hipblasTrsmEx``, ``hipblasGemmEx``, ``hipblasGemmExWithFlags``, ``hipblasAxpyEx``, ``hipblasDot(c)Ex``, ``hipblasNrm2Ex``, ``hipblasRotEx``, ``hipblasScalEx``,
|
|
and the batched and strided-batched variants of these. See the documentation for each function for more information.
|
|
|
|
Note that there are no deprecation warnings for the hipBLAS Fortran API.
|
|
|
|
``hipblasDatatype_t`` will be removed in a future release, and the use of this type in the API will be replaced with ``hipDataType``.
|
|
|
|
hipblasComplex and hipblasDoubleComplex will be replaced by hipComplex and hipDoubleComplex
|
|
===========================================================================================
|
|
|
|
Use of these datatypes will give deprecation warnings telling you to compile with ``-DHIPBLAS_V2`` and to use HIP complex types
|
|
instead. All functions which currently use ``hipblasComplex`` and ``hipblasDoubleComplex`` are therefore deprecated as well
|
|
and will be replaced with functions which use ``hipComplex`` and ``hipDoubleComplex`` in their place.
|
|
|
|
Note that there are no deprecation warnings for the hipBLAS Fortran API.
|
|
|
|
``hipComplex`` and ``hipDoubleComplex`` will be removed in a future release, and the use of this type in the API will be replaced by
|
|
``hipComplex`` and ``hipDoubleComplex``.
|
|
|
|
``ROCM_MATHLIBS_API_USE_HIP_COMPLEX`` is also deprecated as the behavior provided by defining it will be the default in the future.
|
|
|
|
Removed in hipBLAS 2.0
|
|
**********************
|
|
|
|
Legacy BLAS in-place trmm is removed
|
|
====================================
|
|
|
|
The legacy BLAS in-place ``hipblasXtrmm`` that calculates ``B <- alpha * op(A) * B`` is removed and replaced with the
|
|
out-of-place ``hipblasXtrmm`` that calculates ``C <- alpha * op(A) * B``.
|
|
|
|
The prototype for the removed legacy BLAS in-place functionality was
|
|
|
|
::
|
|
|
|
hipblasStatus_t hipblasStrmm(hipblasHandle_t handle,
|
|
hipblasSideMode_t side,
|
|
hipblasFillMode_t uplo,
|
|
hipblasOperation_t transA,
|
|
hipblasDiagType_t diag,
|
|
int m,
|
|
int n,
|
|
const float* alpha,
|
|
const float* A,
|
|
int lda,
|
|
float* B,
|
|
int ldb);
|
|
|
|
The prototype for the replacement in-place and out-of-place functionality is
|
|
|
|
::
|
|
|
|
hipblasStatus_t hipblasStrmm(hipblasHandle_t handle,
|
|
hipblasSideMode_t side,
|
|
hipblasFillMode_t uplo,
|
|
hipblasOperation_t transA,
|
|
hipblasDiagType_t diag,
|
|
int m,
|
|
int n,
|
|
const float* alpha,
|
|
const float* A,
|
|
int lda,
|
|
const float* B,
|
|
int ldb,
|
|
float* C,
|
|
int ldc);
|
|
|
|
The legacy BLAS in-place functionality can be obtained with the new function if you set pointer ``C`` equal to pointer ``B`` and
|
|
``ldc`` equal to ``ldb``.
|
|
|
|
The out-of-place functionality is from setting pointer ``B`` so it is distinct from pointer ``C``.
|
|
|
|
|
|
Announced in hipBLAS 1.0
|
|
************************
|
|
|
|
Legacy BLAS in-place trmm functions will be replaced with trmm functions that support both in-place and out-of-place functionality
|
|
==================================================================================================================================
|
|
|
|
Use of the deprecated Legacy BLAS in-place trmm functions will give deprecation warnings telling
|
|
you to compile with ``-DHIPBLAS_V1`` and use the new in-place and out-of-place trmm functions.
|
|
|
|
Note that there are no deprecation warnings for the hipBLAS Fortran API.
|
|
|
|
The Legacy BLAS in-place trmm calculates ``B <- alpha * op(A) * B``. Matrix B is overwritten by
|
|
triangular matrix A multiplied by matrix B. The prototype in the include file ``rocblas-functions.h`` is:
|
|
|
|
::
|
|
|
|
hipblasStatus_t hipblasStrmm(hipblasHandle_t handle,
|
|
hipblasSideMode_t side,
|
|
hipblasFillMode_t uplo,
|
|
hipblasOperation_t transA,
|
|
hipblasDiagType_t diag,
|
|
int m,
|
|
int n,
|
|
const float* alpha,
|
|
const float* AP,
|
|
int lda,
|
|
float* BP,
|
|
int ldb);
|
|
|
|
The above is replaced by an in-place and out-of-place trmm that calculates ``C <- alpha * op(A) * B``. The prototype is:
|
|
|
|
::
|
|
|
|
hipblasStatus_t hipblasStrmmOutofplace(hipblasHandle_t handle,
|
|
hipblasSideMode_t side,
|
|
hipblasFillMode_t uplo,
|
|
hipblasOperation_t transA,
|
|
hipblasDiagType_t diag,
|
|
int m,
|
|
int n,
|
|
const float* alpha,
|
|
const float* AP,
|
|
int lda,
|
|
const float* BP,
|
|
int ldb,
|
|
float* CP,
|
|
int ldc);
|
|
|
|
The new API provides the legacy BLAS in-place functionality if you set pointer ``C`` equal to pointer ``B`` and
|
|
``ldc`` equal to ``ldb``.
|
|
|
|
There are similar deprecations for the ``_batched`` and ``_strided_batched`` versions of trmm.
|
|
|
|
Removed in hipBLAS 1.0
|
|
**********************
|
|
|
|
HIPBLAS_INT8_DATATYPE_PACK_INT8x4 hipblasGemmEx support removed
|
|
===============================================================
|
|
|
|
Packed int8x4 is removed as support for arbitrary dimensioned ``int8_t data`` is a superset of this functionality:
|
|
|
|
* enum ``hipblasInt8Datatype_t`` is removed
|
|
* function ``hipblasSetInt8Datatype`` is removed
|
|
* function ``hipblasGetInt8Datatype`` is removed
|
|
|
|
Announced in hipBLAS 0.53
|
|
*************************
|
|
|
|
packed_int8x4 datatype will be removed
|
|
======================================
|
|
|
|
The ``packed_int8x4`` datatype will be removed in hipBLAS 1.0. There are two ``int8`` datatypes:
|
|
|
|
* ``int8_t``
|
|
* ``packed_int8x4``
|
|
|
|
``int8_t`` is the C99 unsigned 8-bit integer. ``packed_int8x4`` has 4 consecutive ``int8_t`` numbers
|
|
in the k dimension packed into 32 bits. ``packed_int8x4`` is only used in ``hipblasGemmEx``.
|
|
``int8_t`` will continue to be available in ``hipblasGemmEx``.
|
|
|
|
Announced in hipBLAS 0.49
|
|
*************************
|
|
|
|
Inplace hipblasXtrmm will be replaced with out-of-place hipblasXtrmm
|
|
====================================================================
|
|
|
|
The ``hipblasXtrmm`` API, along with batched versions, will be changing in hipBLAS 1.0
|
|
release to allow in-place and out-of-place behavior. This change will introduce an
|
|
output matrix ``C``, matching the ``rocblas_xtrmm_outofplace`` API and the ``cublasXtrmm`` API.
|