cutt -> librett

.gitlab-ci.yml

@@ -7,28 +7,29 @@ default:
 
 variables:
   MAD_NUM_THREADS : 2
-  TA_TARGETS : "tiledarray examples ta_test check-tiledarray"
+  TA_TARGETS : "tiledarray examples-tiledarray ta_test check-tiledarray"
   # Debug builds with ScaLAPACK=ON need increased TA_UT_CTEST_TIMEOUT
   TA_CONFIG : >
     CMAKE_BUILD_TYPE=${BUILD_TYPE}
     TA_ASSERT_POLICY=TA_ASSERT_THROW
-    TA_UT_CTEST_TIMEOUT=2000
+    TA_UT_CTEST_TIMEOUT=3000
     ${TA_PYTHON}
     ${ENABLE_CUDA}
     ${BLA_VENDOR}
     ${BLA_THREADS}
     ${ENABLE_SCALAPACK}
 
 before_script:
-  # NB: below tag parsing is not robust
-  - echo "CI_RUNNER_TAGS=$CI_RUNNER_TAGS"
-  - CMAKE_BUILD_PARALLEL_LEVEL=$(echo $CI_RUNNER_TAGS | sed -n 's/CMAKE_BUILD_PARALLEL_LEVEL=\([0-9]\+\).*/\1/p')
+  # NB: if CMAKE_BUILD_PARALLEL_LEVEL is not set (i.e. using shared runner), use 1 to ensure we have enough memory
+  # TODO optimize ta_test build memory consumption
   - export CMAKE_BUILD_PARALLEL_LEVEL=${CMAKE_BUILD_PARALLEL_LEVEL:=1}
   - echo "CMAKE_BUILD_PARALLEL_LEVEL=$CMAKE_BUILD_PARALLEL_LEVEL"
 
 ubuntu:
   stage: build
-  tags: [ docker ]
+  tags:
+    - docker
+    - ${RUNNER_TAGS}
   timeout: 3h
   image: valeevgroup/${IMAGE}
   variables:
@@ -62,12 +63,15 @@ ubuntu:
         BLA_THREADS : [ "IntelMKL_THREAD_LAYER=tbb" ]
         # ENABLE_SCALAPACK : [ "ENABLE_SCALAPACK=ON", "ENABLE_SCALAPACK=OFF" ]
         TA_PYTHON : [ "TA_PYTHON=OFF" ] # needs to be fixed for MKL
+        RUNNER_TAGS: [ linux ]
       - IMAGE : [ "ubuntu:18.04", "ubuntu:20.04" ]
         CXX: [ g++, clang++-9 ]
         BUILD_TYPE : [ "Release", "Debug" ]
         ENABLE_SCALAPACK : [ "ENABLE_SCALAPACK=ON", "ENABLE_SCALAPACK=OFF" ]
+        RUNNER_TAGS: [ linux ]
       - IMAGE : [ "ubuntu:18.04", "ubuntu:20.04" ]
         CXX: [ g++ ]
         BUILD_TYPE : [ "Release", "Debug" ]
         ENABLE_CUDA : [ "ENABLE_CUDA=ON" ]
-        TA_TARGETS : [ "tiledarray examples" ]
+        TA_TARGETS : [ "tiledarray examples-tiledarray check_serial-tiledarray" ]
+        RUNNER_TAGS: [ cuda ]

CMakeLists.txt

@@ -374,14 +374,18 @@ add_subdirectory(doc)
 ##########################
 include(CTest)
 if (BUILD_TESTING)
-  set(_ctest_args -V -R "tiledarray/unit")
+  set(_ctest_args -V -R "tiledarray/unit/run-np.*")
+  set(_ctest_args_serial -V -R "tiledarray/unit/run-np-1")
   if (DEFINED TA_UT_CTEST_TIMEOUT)
     list(APPEND _ctest_args --timeout ${TA_UT_CTEST_TIMEOUT})
+    list(APPEND _ctest_args_serial --timeout ${TA_UT_CTEST_TIMEOUT})
   endif(DEFINED TA_UT_CTEST_TIMEOUT)
   add_custom_target_subproject(tiledarray check USES_TERMINAL COMMAND ${CMAKE_CTEST_COMMAND} ${_ctest_args})
+  add_custom_target_subproject(tiledarray check_serial USES_TERMINAL COMMAND ${CMAKE_CTEST_COMMAND} ${_ctest_args_serial})
   add_subdirectory(tests)
 else()
   add_custom_target_subproject(tiledarray check USES_TERMINAL COMMAND echo "WARNING: unit testing disabled. To enable, give -DBUILD_TESTING=ON to cmake")
+  add_custom_target_subproject(tiledarray check_serial USES_TERMINAL COMMAND echo "WARNING: unit testing disabled. To enable, give -DBUILD_TESTING=ON to cmake")
 endif()
 
 ##########################

INSTALL.md

@@ -64,11 +64,11 @@ Compiling BTAS requires the following prerequisites:
 
 Optional prerequisites:
 - [CUDA compiler and runtime](https://developer.nvidia.com/cuda-zone) -- for execution on CUDA-enabled accelerators. CUDA 11 or later is required. Support for CUDA also requires the following additional prerequisites, both of which will be built and installed automatically if missing:
-  - [cuTT](github.com/ValeevGroup/cutt) -- CUDA transpose library; note that our fork of the [original cuTT repo](github.com/ap-hynninen/cutt) is required to provide thread-safety (tag 0e8685bf82910bc7435835f846e88f1b39f47f09).
+  - [LibreTT](github.com/victor-anisimov/LibreTT) -- free tensor transpose library for CUDA, HIP, and SYCL platforms that is based on the [original cuTT library](github.com/ap-hynninen/cutt) extended to provide thread-safety improvements (via github.com/ValeevGroup/cutt) and extended to non-CUDA platforms by [@victor-anisimov](github.com/victor-anisimov) (tag 68abe31a9ec6fd2fd9ffbcd874daa80457f947da).
   - [Umpire](github.com/LLNL/Umpire) -- portable memory manager for heterogeneous platforms (tag f9640e0fa4245691cdd434e4f719ac5f7d455f82).
 - [Doxygen](http://www.doxygen.nl/) -- for building documentation (version 1.8.12 or later).
 - [ScaLAPACK](http://www.netlib.org/scalapack/) -- a distributed-memory linear algebra package. If detected, the following C++ components will also be sought and downloaded, if missing:
-  - [scalapackpp](https://github.com/wavefunction91/scalapackpp.git) -- a modern C++ (C++17) wrapper for ScaLAPACK (tag 28433942197aee141cd9e96ed1d00f6ec7b902cb); pulls and builds the following additional prerequisite
+  - [scalapackpp](https://github.com/wavefunction91/scalapackpp.git) -- a modern C++ (C++17) wrapper for ScaLAPACK (tag 711ef363479a90c88788036f9c6c8adb70736cbf); pulls and builds the following additional prerequisite
     - [blacspp](https://github.com/wavefunction91/blacspp.git) -- a modern C++ (C++17) wrapper for BLACS
 - Python3 interpreter -- to test (optionally-built) Python bindings
 - [Range-V3](https://github.com/ericniebler/range-v3.git) -- a Ranges library that served as the basis for Ranges component of C++20; only used for some unit testing of the functionality anticipated to be supported by future C++ standards.
@@ -330,7 +330,7 @@ Support for execution on CUDA-enabled hardware is controlled by the following va
 * `ENABLE_CUDA`  -- Set to `ON` to turn on CUDA support. [Default=OFF].
 * `CMAKE_CUDA_HOST_COMPILER`  -- Set to the path to the host C++ compiler to be used by CUDA compiler. CUDA compilers used to be notorious for only being able to use specific C++ host compilers, but support for more recent C++ host compilers has improved. The default is determined by the CUDA compiler and the user environment variables (`PATH` etc.).
 * `ENABLE_CUDA_ERROR_CHECK` -- Set to `ON` to turn on assertions for successful completion of calls to CUDA runtime and libraries. [Default=OFF].
-* `CUTT_INSTALL_DIR` -- the installation prefix of the pre-installed cuTT library. This should not be normally needed; it is strongly recommended to let TiledArray build and install cuTT.
+* `LIBRETT_INSTALL_DIR` -- the installation prefix of the pre-installed LibreTT library. This should not be normally needed; it is strongly recommended to let TiledArray build and install LibreTT.
 * `UMPIRE_INSTALL_DIR` -- the installation prefix of the pre-installed Umpire library. This should not be normally needed; it is strongly recommended to let TiledArray build and install Umpire.
 
 For the CUDA compiler and toolkit to be discoverable the CUDA compiler (`nvcc`) should be in the `PATH` environment variable. Refer to the [FindCUDAToolkit module](https://cmake.org/cmake/help/latest/module/FindCUDAToolkit.html) for more info.

README.md

@@ -1,4 +1,3 @@
-[![Travis Build Status](https://travis-ci.com/ValeevGroup/tiledarray.svg?branch=master)](https://travis-ci.com/ValeevGroup/tiledarray)
 [![Gitlab Pipeline Status](https://gitlab.com/ValeevGroup/tiledarray/badges/master/pipeline.svg)](https://gitlab.com/ValeevGroup/tiledarray/-/pipelines?page=1&scope=all&ref=master)
 [![codecov](https://codecov.io/gh/ValeevGroup/tiledarray/branch/master/graph/badge.svg)](https://codecov.io/gh/ValeevGroup/tiledarray)
 

bin/admin/dependency-versions-update-hook.py

@@ -106,11 +106,11 @@ def replace_dep_id(topsrc, file_ext, dep_name, old_id, new_id, search_prefix = '
                 btas_old_tag = tokens[2]
             else:
                 btas_new_tag = tokens[2]
-        elif tokens[1].find('CUTT') != -1:
+        elif tokens[1].find('LIBRETT') != -1:
             if tokens[1].find('PREVIOUS') != -1:
-                cutt_old_tag = tokens[2]
+                librett_old_tag = tokens[2]
             else:
-                cutt_new_tag = tokens[2]
+                librett_new_tag = tokens[2]
         elif tokens[1].find('UMPIRE') != -1:
             if tokens[1].find('PREVIOUS') != -1:
                 umpire_old_tag = tokens[2]
@@ -146,8 +146,8 @@ def replace_dep_id(topsrc, file_ext, dep_name, old_id, new_id, search_prefix = '
 # BTAS tag in INSTALL.md
 any_files_changed |= replace_dep_id(topsrc, 'md', 'BTAS', btas_old_tag, btas_new_tag, 'ValeevGroup/BTAS), tag ', '')
 
-# cuTT tag in INSTALL.md
-any_files_changed |= replace_dep_id(topsrc, 'md', 'cuTT', cutt_old_tag, cutt_new_tag, '', '')
+# LibreTT tag in INSTALL.md
+any_files_changed |= replace_dep_id(topsrc, 'md', 'LibreTT', librett_old_tag, librett_new_tag, '', '')
 
 # Umpire tag in INSTALL.md
 any_files_changed |= replace_dep_id(topsrc, 'md', 'Umpire', umpire_old_tag, umpire_new_tag, '', '')