rocm-5.4.0

ROCm release v5.4.0

rocm-5.3.3

ROCm release v5.3.3

AOMP Release 16.0-2

These are the release notes for AOMP 16.0-2. This release uses modifications to the LLVM development trunk called the "amd-stg-open" branch. This is found at https://github.com/RadeonOpenCompute/llvm-project. The amd-stg-open branch is constantly changing as AMD merges upstream development trunk with its internal open development efforts. Some AMD modifications are experimental and/or under review for the LLVM upstream mono-repo. The AOMP release is a snapshot of amd-stg-open and supporting repositories to build various components.

For AOMP 16.0-2, the last trunk commit is 0ccff030f3b4145bd658e362a63db9aae2942bee on Oct 15 2022. This is the 3rd AOMP release for LLVM 16 development. The last amd-only commit is 863a830a66c8bdb5371e56030961449df24d5c48 on Nov 2, 2022. This forms a frozen branch now called "aomp-16.0-2". See https://github.com/RadeonOpenCompute/llvm-project/tree/aomp-16.0-2. Currently, the amd-only content differs from the trunk by 62480 lines in 459 files.

AOMP is a "standalone" build of all necessary ROCm components with the exception of the kernel module and libdrm. The non llvm-project components for this release were built with ROCM 5.3.x sources.

These are the changes from 16.0-1 to 16.0-2 include:

• Dropped support for Ubuntu 18.0-4.
• Fix for early USM failure - openfoam does not hang
• Enhance xteam reductions (with codegen). Support for integer data types.
• Fix for double _Complex scalars in target region, test/smoke-fails/double_complex_scalar now works. This test will move to smoke in next release.
• Support for target teams loop directive
• Force synchronous execution of regions controlled via OMPX_FORCE_SYNC_REGIONS
• New environment variable GPU_MAX_HW_QUEUES controls number of HSA queues created, default is 4. This change actually occurred in 16.0-1 but was not listed in release notes.

Errata:
(potential regressions from 16.0-1):

• Smoke test failures:
  clang-ifaces: core dump (gfx908)
  clang-337336: core dump gfx908)
  clang-325070: core dump (gfx908)

(potential regressions from 16.0-0):

• Performance decrease with lulesh
• Performance decrease with Nekbone
• Smoke test failures:
  flang-315870: (resolved by building this test case with cov5)
  managed_memory: segfault, when 2+ devices are present
• Hip example failure:
  device-lib

rocm-5.3.2

ROCm release v5.3.2

rocm-5.3.1

ROCm release v5.3.1

AOMP Release 16.0-1

These are the release notes for AOMP 16.0-1. This release uses modifications to the LLVM development trunk called the "amd-stg-open" branch. This is found at https://github.com/RadeonOpenCompute/llvm-project. The amd-stg-open branch is constantly changing as AMD merges upstream development trunk with its internal open development efforts. Some AMD modifications are experimental and/or under review for the LLVM upstream mono-repo. The AOMP release is a snapshot of amd-stg-open and supporting repositories to build various components.

For AOMP 16.0-1, the last trunk commit is aa89f08afad7ee0581c39638abd8ee0df9ba1c65 on Oct 17 2022. This is the 2nd AOMP release for LLVM 16 development. The last amd-only commit is 16791f61b04f07a7968a67c18ed41388279018d5 on Oct 13, 2022. This forms a frozen branch now called "aomp-16.0-1". See https://github.com/RadeonOpenCompute/llvm-project/tree/aomp-16.0-1 . Currently, the amd-only content differs from the trunk by 63,997 lines in 467 files.

AOMP is a "standalone" build of all necessary ROCm components with the exception of the kernel module and libdrm. The non llvm-project components for this release were built with ROCM 5.3.x sources.

The changes from 16.0-0 to 16.0-1 include:

• Enhanced xteam reductions, no codegen.
• Optimized wait for signals (perf gain).
• Fix aompcc and mark for deprecation.
• Added switch for code object version 5. Version 4 is still default.
• Support for gfx1100 - gfx1103.
• Support for order(concurrent).
• Build OpenMP warnings cleaned up.
• Support atomic min/max on MI200.
• Support for device new and delete.
• Bumped cmake version to 3.18.5.
• Switch to ROCm 5.3 sources.

Errata: (potential regressions from 16.0-0)

• Performance decrease with lulesh
• Performance decrease with Nekbone
• Smoke test failure: flang-315870
• Hip example failure: device-lib

rocm-5.3.0

ROCm release v5.3.0

AOMP Release 16.0-0

These are the release notes for AOMP 16.0-0. This release uses modifications to the LLVM development trunk called the "amd-stg-open" branch. This is found at https://github.com/RadeonOpenCompute/llvm-project. The amd-stg-open branch is constantly changing as AMD merges upstream development trunk with its internal open development efforts. Some AMD modifications are experimental and/or under review for the LLVM upstream mono-repo. The AOMP release is a snapshot of amd-stg-open and supporting repositories to build various components.

For AOMP 16.0-0, the last trunk commit is 1b56b2b2678cde21f7c20e83f881ded9b96518e4 on Sep 14 2022. This is the first AOMP release for LLVM 16 development. The last amd-only commit is 0018e8ab17297453e971ea1867d085eba5ea3f9d on Sep 14 2022. This forms a frozen branch now called "aomp-16.0-0". See https://github.com/RadeonOpenCompute/llvm-project/tree/aomp-16.0-0

AOMP is a "standalone" build of all necessary ROCm components with the exception of the kernel module. The non llvm-project components for this release were built with ROCM 5.2.x sources.

The changes from 15.0-3 to 16.0-0 include:

• Adds new flag -fopenmp-target-fast to group enable a set of OpenMP target optimizations.
• Enhancements and bug fixes for No-Loop and cross-team reduction support.

AOMP Release 15.0-3

These are the release notes for AOMP 15.0-3. This release uses modifications to the LLVM development trunk called the "amd-stg-open" branch. This is found at https://github.com/RadeonOpenCompute/llvm-project. The amd-stg-open branch is constantly changing as AMD merges upstream development trunk with its internal open development efforts. Some AMD modifications are experimental and/or under review for the LLVM upstream mono-repo. The AOMP release is a snapshot of amd-stg-open and supporting repositories to build various components.

For AOMP 15.0-3, the last trunk commit is 1f8ae9d7e7e4afcc4e76728b28e64941660ca3eb on Jul 26 2022. This is the fourth AOMP release for LLVM 15 development. The last amd-only commit is b745843ebcb77f55de887b5741197184e7d0dcbd on Aug 01 2022. This forms a frozen branch now called "aomp-15.0-3". See https://github.com/RadeonOpenCompute/llvm-project/tree/aomp-15.0-3

AOMP is a "standalone" build of all necessary ROCm components with the exception of the kernel module. The non llvm-project components for this release were built with ROCM 5.2.x sources.

The changes from 15.0-2 to 15.0-3 include:

• Use the new openmp DeviceRTL by default.
• New DeviceRTL APIs for optimized cross-team reduction.
• Clang codegen changes to use the optimized cross-team reduction APIs for a reduction clause in a device construct.
• Added support for classic flang to use the new DeviceRTL.

Known Issues:

• Flang has issues at -O0 when using the new DeviceRTL on GPUs other than gfx90a.

rocm-5.2.1

ROCm release v5.2.1