Merge branch 'master' into alby/tfactor-optim/bulk

eth-cscs · Dec 20, 2024 · 2f90f23 · 2f90f23
2 parents ffd3135 + b82a100
commit 2f90f23
Show file tree

Hide file tree

Showing 110 changed files with 397 additions and 69 deletions.
diff --git a/.inshpect.toml b/.inshpect.toml
@@ -152,6 +152,7 @@ patterns = [
     { pattern = '\bstd::error_(code|condition|category)\b', include = 'system_error' },
     { pattern = '\bstd::system_error\b', include = 'system_error' },
     { pattern = '\bstd::chrono\b', include = 'chrono' },
+    { pattern = '\bstd::complex\b', include = 'complex' },
 ]
 
 [copyright]

diff --git a/BUILD.md b/BUILD.md
@@ -55,7 +55,7 @@ You can then install the master version with:
 spack install dla-future@master <variants>
 ```
 
-## Build the good old way
+## Manual build
 
 ### Dependencies
 
@@ -64,14 +64,18 @@ spack install dla-future@master <variants>
 - [umpire](https://github.com/LLNL/Umpire)
 - [blaspp](https://bitbucket.org/icl/blaspp/src/default/)
 - [lapackpp](https://bitbucket.org/icl/lapackpp/src/default/)
-- Intel MKL or other LAPACK implementation
+- BLAS/LAPACK implementations (e.g. Intel MKL, OpenBLAS, NVPL)
 - A ScaLAPACK implementation (optional, ScaLAPACK-like C API only)
 - [whip](https://github.com/eth-cscs/whip) (optional, GPU only)
 - [CUDA](https://developer.nvidia.com/cuda) (optional, NVidia GPUs only)
 - [HIP/ROCm](https://github.com/RadeonOpenCompute/ROCm) (optional, AMD GPUs only)
 - [GoogleTest](https://github.com/google/googletest) (optional; bundled) - unit testing
 - Doxygen (optional) - documentation
 
+### Version constraints
+
+Please refer to the [Spack package](https://packages.spack.io/package.html?name=dla-future) for requirements and conflicts.
+
 ### Get DLA-Future
 
 You can download the [releases](https://github.com/eth-cscs/DLA-Future/releases).
@@ -94,9 +98,10 @@ CMake option | Values | Note
 `pika_DIR` | CMAKE:PATH | Location of the pika CMake-config file
 `blaspp_DIR` | CMAKE:PATH | Location of the blaspp CMake-config file
 `lapackpp_DIR` | CMAKE:PATH | Location of the lapackpp CMake-config file
+`DLAF_LAPACK_LIBRARY` | CMAKE:STRING | Full link line for BLAS and LAPACK
+`DLAF_SCALAPACK_LIBRARY` | CMAKE:STRING | Full link line for ScaLAPACK
 `DLAF_WITH_MKL` | `{ON,OFF}` (default: `OFF`) | if blaspp/lapackpp is built with oneMKL
 `DLAF_WITH_SCALAPACK` | `{ON,OFF}` (default: `OFF`) | Enable ScaLAPACK-like API.
-`MKL_ROOT` | CMAKE:PATH | Location of the MKL library
 `DLAF_ASSERT_ENABLE` | `{ON,OFF}` (default: `ON`) | enable/disable cheap assertions
 `DLAF_ASSERT_MODERATE_ENABLE` | `{ON,OFF}` (default: `ON` in Debug, `OFF` otherwise) | enable/disable moderate assertions
 `DLAF_ASSERT_HEAVY_ENABLE` | `{ON,OFF}` (default: `ON` in Debug, `OFF` otherwise) | enable/disable heavy assertions

diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,5 +1,42 @@
 # Changelog
 
+## DLA-Future 0.7.3
+
+### Bug fixes
+
+- Changed C ScaLAPACK API indexing convention to 1-based for partial eigenspectrum. (#1248)
+
+## DLA-Future 0.7.2
+
+Not released.
+
+## DLA-Future 0.7.1
+
+### Bug fixes
+
+- Fixed a compilation issue with ROCm. (#1241)
+- Fixed missing includes of `<complex>`. (#1243)
+
+## DLA-Future 0.7.0
+
+### Changes
+
+- Added (generalized) eigensolver which computes only a part of the eigenspectrum. (#1194)
+- `Norm` is now fully asynchronous. (#1221)
+
+### Performance improvements
+
+- Refactored communication to use pika's `transform_mpi` and polling support. (#1125)
+- Use custom coalescing heuristic for memory pools. (#1183)
+- Added configuration option for number of CUDA streams and cuBLAS/SOLVER handles. (#1222, #1182)
+- Some algorithmic clean-ups and improvements. (#1213, #1219, #1232)
+
+### Bug fixes
+
+- Fixed builds with CUDA and stdexec enabled. (#1188)
+- Work around buggy HIP complex operator overloads. (#1195)
+- Namespace (SCA)LAPACK CMake modules with DLAF to avoid conflicts with other packages. (#1178)
+
 ## DLA-Future 0.6.0
 
 ### Changes

diff --git a/CITATION.cff b/CITATION.cff
@@ -1,5 +1,5 @@
 cff-version: 1.2.0
-title: DLA-Future 0.6.0
+title: DLA-Future 0.7.3
 repository-code: https://github.com/eth-cscs/DLA-Future
 message: >-
   If you use this software, please cite it using the
@@ -54,5 +54,5 @@ keywords:
   - mpi
 license: BSD-3-Clause
 license-url: https://opensource.org/license/bsd-3-clause
-version: 0.6.0
-date-released: '2024-06-24'
+version: 0.7.3
+date-released: '2024-12-20'
diff --git a/CMakeLists.txt b/CMakeLists.txt
@@ -10,7 +10,7 @@
 
 cmake_minimum_required(VERSION 3.22)
 
-project(DLAF VERSION 0.6.0)
+project(DLAF VERSION 0.7.3)
 
 # ---------------------------------------------------------------------------
 # CMake configurations

diff --git a/DOCUMENTATION.md b/DOCUMENTATION.md
@@ -3,6 +3,9 @@
 ## API documentation
 
 - [Documentation of `master` branch](https://eth-cscs.github.io/DLA-Future/master/)
+- [Documentation of `v0.7.3`](https://eth-cscs.github.io/DLA-Future/v0.7.3/)
+- [Documentation of `v0.7.1`](https://eth-cscs.github.io/DLA-Future/v0.7.1/)
+- [Documentation of `v0.7.0`](https://eth-cscs.github.io/DLA-Future/v0.7.0/)
 - [Documentation of `v0.6.0`](https://eth-cscs.github.io/DLA-Future/v0.6.0/)
 - [Documentation of `v0.5.0`](https://eth-cscs.github.io/DLA-Future/v0.5.0/)
 - [Documentation of `v0.4.1`](https://eth-cscs.github.io/DLA-Future/v0.4.1/)

diff --git a/README.md b/README.md
@@ -60,7 +60,7 @@ target_link_libraries(<your_target> PRIVATE DLAF::DLAF)
 ### Documentation
 
 - [Documentation of `master` branch](https://eth-cscs.github.io/DLA-Future/master/)
-- [Documentation of `v0.6.0`](https://eth-cscs.github.io/DLA-Future/v0.6.0/)
+- [Documentation of `v0.7.3`](https://eth-cscs.github.io/DLA-Future/v0.7.3/)
 
 See [DOCUMENTATION.md](DOCUMENTATION.md) for the documentation of older versions, or for the instructions to build it.
 

diff --git a/ci/.gitlab-ci.yml b/ci/.gitlab-ci.yml
@@ -17,3 +17,4 @@ include:
   - local: 'ci/cuda/gcc13_release_stdexec.yml'
   - local: 'ci/rocm/clang14_release.yml'
   - local: 'ci/rocm/clang14_release_stdexec.yml'
+  - local: 'ci/rocm/clang15_release_stdexec.yml'
diff --git a/ci/cpu/asan_ubsan_lsan.yml b/ci/cpu/asan_ubsan_lsan.yml
@@ -13,7 +13,7 @@ cpu asan ubsan lsan deps:
 cpu asan ubsan lsan build:
   extends:
     - .build_common
-    - .build_for_eiger
+    - .build_for_daint-mc
   needs:
     - cpu asan ubsan lsan deps
   variables:

diff --git a/ci/cpu/clang15_release.yml b/ci/cpu/clang15_release.yml
@@ -13,7 +13,7 @@ cpu clang15 release deps:
 cpu clang15 release build:
   extends:
     - .build_common
-    - .build_for_eiger
+    - .build_for_daint-mc
   needs:
     - cpu clang15 release deps
   variables:

diff --git a/ci/cpu/clang15_release_cxx20.yml b/ci/cpu/clang15_release_cxx20.yml
@@ -14,7 +14,7 @@ cpu clang15 cxx20 release deps:
 cpu clang15 cxx20 release build:
   extends:
     - .build_common
-    - .build_for_eiger
+    - .build_for_daint-mc
   needs:
     - cpu clang15 cxx20 release deps
   variables:

diff --git a/ci/cpu/clang15_release_stdexec.yml b/ci/cpu/clang15_release_stdexec.yml
@@ -14,7 +14,7 @@ cpu clang15 stdexec release deps:
 cpu clang15 stdexec release build:
   extends:
     - .build_common
-    - .build_for_eiger
+    - .build_for_daint-mc
   needs:
     - cpu clang15 stdexec release deps
   variables:

diff --git a/ci/cpu/clang16_release.yml b/ci/cpu/clang16_release.yml
@@ -13,7 +13,7 @@ cpu clang16 release deps:
 cpu clang16 release build:
   extends:
     - .build_common
-    - .build_for_eiger
+    - .build_for_daint-mc
   needs:
     - cpu clang16 release deps
   variables:

diff --git a/ci/cpu/clang18_release.yml b/ci/cpu/clang18_release.yml
@@ -13,7 +13,7 @@ cpu clang18 release deps:
 cpu clang18 release build:
   extends:
     - .build_common
-    - .build_for_eiger
+    - .build_for_daint-mc
   needs:
     - cpu clang18 release deps
   variables:

diff --git a/ci/cpu/gcc11_debug_stdexec.yml b/ci/cpu/gcc11_debug_stdexec.yml
@@ -14,7 +14,7 @@ cpu gcc11 stdexec debug deps:
 cpu gcc11 stdexec debug build:
   extends:
     - .build_common
-    - .build_for_eiger
+    - .build_for_daint-mc
   needs:
     - cpu gcc11 stdexec debug deps
   variables:

diff --git a/ci/cpu/gcc11_release_stdexec.yml b/ci/cpu/gcc11_release_stdexec.yml
@@ -14,7 +14,7 @@ cpu gcc11 stdexec release deps:
 cpu gcc11 stdexec release build:
   extends:
     - .build_common
-    - .build_for_eiger
+    - .build_for_daint-mc
   needs:
     - cpu gcc11 stdexec release deps
   variables:

diff --git a/ci/cpu/gcc12_release_cxx20.yml b/ci/cpu/gcc12_release_cxx20.yml
@@ -14,7 +14,7 @@ cpu gcc12 cxx20 release deps:
 cpu gcc12 cxx20 release build:
   extends:
     - .build_common
-    - .build_for_eiger
+    - .build_for_daint-mc
   needs:
     - cpu gcc12 cxx20 release deps
   variables:

diff --git a/ci/cpu/gcc13_codecov.yml b/ci/cpu/gcc13_codecov.yml
@@ -12,7 +12,7 @@ cpu gcc13 codecov deps:
 cpu gcc13 codecov build:
   extends:
     - .build_common
-    - .build_for_eiger
+    - .build_for_daint-mc
   needs:
     - cpu gcc13 codecov deps
   variables:

diff --git a/ci/cpu/gcc13_release.yml b/ci/cpu/gcc13_release.yml
@@ -12,7 +12,7 @@ cpu gcc13 release deps:
 cpu gcc13 release build:
   extends:
     - .build_common
-    - .build_for_eiger
+    - .build_for_daint-mc
   needs:
     - cpu gcc13 release deps
   variables:

diff --git a/ci/docker/release-rocm602-stdexec.yaml b/ci/docker/release-rocm602-stdexec.yaml
@@ -0,0 +1,70 @@
+#
+# Distributed Linear Algebra with Future (DLAF)
+#
+# Copyright (c) 2018-2024, ETH Zurich
+# All rights reserved.
+#
+# Please, refer to the LICENSE file in the root directory.
+# SPDX-License-Identifier: BSD-3-Clause
+#
+
+spack:
+  include:
+  - /spack_environment/common.yaml
+
+  view: false
+  concretizer:
+    unify:
+      true
+
+  specs:
+  - dla-future@master +rocm amdgpu_target=gfx90a:xnack- +miniapps +ci-test
+
+  packages:
+    all:
+      variants:
+      - build_type=Release
+      - cxxstd=20
+      - amdgpu_target=gfx90a:xnack-
+    pika:
+      require:
+        - '+stdexec'
+    blas:
+      require:: openblas
+    lapack:
+      require:: openblas
+    mpich:
+      require:
+        - '~rocm'
+        - 'device=ch3'
+        - 'netmod=tcp'
+    llvm-amdgpu:
+      externals:
+      - spec: llvm-amdgpu@6.0.2 ~rocm-device-libs
+        prefix: /opt/rocm-6.0.2/llvm
+      buildable: false
+    rocm-device-libs:
+      externals:
+      - spec: rocm-device-libs@6.0.2
+        prefix: /opt/rocm-6.0.2
+      buildable: false
+    hip:
+      externals:
+      - spec: hip@6.0.2
+        prefix: /opt/rocm-6.0.2
+      buildable: false
+    rocblas:
+      externals:
+      - spec: rocblas@6.0.2
+        prefix: /opt/rocm-6.0.2
+      buildable: false
+    rocsolver:
+      externals:
+      - spec: rocsolver@6.0.2
+        prefix: /opt/rocm-6.0.2
+      buildable: false
+    hsa-rocr-dev:
+      externals:
+      - spec: hsa-rocr-dev@6.0.2
+        prefix: /opt/rocm-6.0.2
+      buildable: false
diff --git a/ci/rocm/clang15_release_stdexec.yml b/ci/rocm/clang15_release_stdexec.yml
@@ -0,0 +1,21 @@
+include:
+  - local: 'ci/common-ci.yml'
+
+rocm clang15 stdexec release deps:
+  extends: .build_deps_common
+  variables:
+    BASE_IMAGE: docker.io/rocm/dev-ubuntu-22.04:6.0.2
+    EXTRA_APTGET: "clang-15 libomp-15-dev rocblas rocblas-dev rocsolver rocsolver-dev llvm-amdgpu rocm-device-libs"
+    COMPILER: clang@15
+    USE_ROCBLAS: "ON"
+    SPACK_ENVIRONMENT: ci/docker/release-rocm602-stdexec.yaml
+    DEPS_IMAGE: $CSCS_REGISTRY_PATH/rocm-clang15-stdexec-release/deps
+
+rocm clang15 stdexec release build:
+  extends:
+    - .build_common
+  after_script: null
+  needs:
+    - rocm clang15 stdexec release deps
+  variables:
+    DLAF_IMAGE: $CSCS_REGISTRY_PATH/rocm-clang15-stdexec-release/dlaf:$CI_COMMIT_SHA
diff --git a/include/dlaf/auxiliary/norm/api.h b/include/dlaf/auxiliary/norm/api.h
@@ -9,6 +9,8 @@
 //
 #pragma once
 
+#include <complex>
+
 #include <pika/execution.hpp>
 
 #include <dlaf/communication/communicator_grid.h>

diff --git a/include/dlaf/eigensolver/band_to_tridiag/api.h b/include/dlaf/eigensolver/band_to_tridiag/api.h
@@ -10,6 +10,8 @@
 
 #pragma once
 
+#include <complex>
+
 #include <dlaf/matrix/matrix.h>
 #include <dlaf/types.h>
 

diff --git a/include/dlaf/eigensolver/bt_band_to_tridiag/api.h b/include/dlaf/eigensolver/bt_band_to_tridiag/api.h
@@ -10,6 +10,8 @@
 
 #pragma once
 
+#include <complex>
+
 #include <dlaf/communication/communicator_grid.h>
 #include <dlaf/matrix/matrix.h>
 #include <dlaf/matrix/matrix_ref.h>

diff --git a/include/dlaf/eigensolver/bt_reduction_to_band/api.h b/include/dlaf/eigensolver/bt_reduction_to_band/api.h
@@ -9,6 +9,8 @@
 //
 #pragma once
 
+#include <complex>
+
 #include <dlaf/matrix/matrix.h>
 #include <dlaf/matrix/matrix_ref.h>
 #include <dlaf/types.h>

diff --git a/include/dlaf/eigensolver/eigensolver/api.h b/include/dlaf/eigensolver/eigensolver/api.h
@@ -10,6 +10,8 @@
 
 #pragma once
 
+#include <complex>
+
 #include <dlaf/blas/tile.h>
 #include <dlaf/common/vector.h>
 #include <dlaf/communication/communicator_grid.h>

diff --git a/include/dlaf/eigensolver/gen_eigensolver/api.h b/include/dlaf/eigensolver/gen_eigensolver/api.h
@@ -9,6 +9,8 @@
 //
 #pragma once
 
+#include <complex>
+
 #include <dlaf/communication/communicator_grid.h>
 #include <dlaf/eigensolver/eigensolver/api.h>
 #include <dlaf/matrix/matrix.h>

diff --git a/include/dlaf/eigensolver/gen_to_std/api.h b/include/dlaf/eigensolver/gen_to_std/api.h
@@ -9,6 +9,8 @@
 //
 #pragma once
 
+#include <complex>
+
 #include <dlaf/matrix/matrix.h>
 #include <dlaf/types.h>