.gitlab/test_openmp.yml

@@ -2,9 +2,7 @@
 .test_openmp:
   extends: .test_script_linux
   needs: [build_openmp]
-  interruptible: true
   variables:
-    SYSTEM: linux
     VERSION: openmp
     CATEGORY: shm
     LOGNAME: "chameleon-test-${SYSTEM}-${VERSION}-${CATEGORY}-${PRECISION}"

.gitlab/test_parsec.yml

@@ -2,9 +2,7 @@
 .test_parsec:
   extends: .test_script_linux
   needs: [build_parsec]
-  interruptible: true
   variables:
-    SYSTEM: linux
     VERSION: parsec
     CATEGORY: shm
     LOGNAME: "chameleon-test-${SYSTEM}-${VERSION}-${CATEGORY}-${PRECISION}"

.gitlab/test_quark.yml

@@ -2,9 +2,7 @@
 .test_quark:
   extends: .test_script_linux
   needs: [build_quark]
-  interruptible: true
   variables:
-    SYSTEM: linux
     VERSION: quark
     CATEGORY: shm
     LOGNAME: "chameleon-test-${SYSTEM}-${VERSION}-${CATEGORY}-${PRECISION}"

.gitlab/test_starpu.yml

 ---
-.test_starpu:
+.test_script_starpu:
   extends: .test_script_linux
   needs: [build_starpu]
-  interruptible: true
   variables:
-    SYSTEM: linux
     VERSION: starpu
     LOGNAME: "chameleon-test-${SYSTEM}-${VERSION}-${CATEGORY}-${PRECISION}"
-    TESTS_RESTRICTION: "-R _${CATEGORY}_${PRECISION}|example -E sytrf|sysv"
+    TESTS_RESTRICTION: "-R _${CATEGORY}_${PRECISION}|example"
 
 test_starpu_master:
-  extends: .test_starpu
+  extends: .test_script_starpu
   rules:
     - if: ($CI_COMMIT_BRANCH == $CI_DEFAULT_BRANCH || $CI_COMMIT_BRANCH =~ /^ci-.*$/) && $CI_PIPELINE_SOURCE != "schedule"
   parallel:
@@ -19,7 +17,7 @@ test_starpu_master:
         CATEGORY: [shm, mpi]
 
 test_starpu_mr:
-  extends: .test_starpu
+  extends: .test_script_starpu
   rules:
     - if: ($CI_PIPELINE_SOURCE == "merge_request_event" && $CI_MERGE_REQUEST_SOURCE_BRANCH_NAME !~ /^notest-.*$/)
   parallel:
@@ -43,7 +41,8 @@ test_starpu_macosx:
     - bash .gitlab/test.sh | tee ${LOGNAME}.log
   artifacts:
     name: "$CI_JOB_NAME-$CI_COMMIT_REF_SLUG"
-    expire_in: 180 minutes
+    expire_in: 1 week
+    when: always
     paths:
       - ${LOGNAME}.log
     reports:
@@ -69,7 +68,8 @@ test_starpu_macosx:
     - bash -lc .gitlab/test.sh | tee "$env:LOGNAME.log"
   artifacts:
     name: "$CI_JOB_NAME-$CI_COMMIT_REF_SLUG"
-    expire_in: 180 minutes
+    expire_in: 1 week
+    when: always
     paths:
       - ${LOGNAME}.log
     reports:

.gitlab/test_starpu_plafrim.yml

+---
+.test_script_starpu_plafrim:
+  tags: ['plafrim']
+  stage: test
+  variables:
+    SYSTEM: guix
+  script:
+    - guix describe -f channels | tee guix-channels.scm
+    - guix shell --pure --preserve=SYSTEM --preserve=VERSION --preserve=LOGNAME --preserve=BUILD_OPTIONS --preserve=TESTS_RESTRICTION --preserve=SLURM_CONSTRAINTS --preserve=LD_PRELOAD
+                 -D chameleon-${GPU_BACKEND} ${BLAS} slurm coreutils inetutils bash-minimal --
+                 bash .gitlab/sbatch.sh ${LOGNAME} | tee ${LOGNAME}.log
+  artifacts:
+    name: "$CI_JOB_NAME-$CI_COMMIT_REF_SLUG"
+    expire_in: 1 week
+    when: always
+    paths:
+      - guix-channels.scm
+      - ${LOGNAME}.log
+    reports:
+      junit: ${LOGNAME}-junit.xml
+
+.test_script_starpu_plafrim_cuda:
+  extends: .test_script_starpu_plafrim
+  needs: [build_starpu_cuda]
+  variables:
+    VERSION: starpu_cuda
+    GPU_BACKEND: cuda
+    BLAS: "--with-input=openblas=intel-oneapi-mkl"
+    LOGNAME: "chameleon-${SYSTEM}-${VERSION}"
+    BUILD_OPTIONS: "-DCHAMELEON_USE_CUDA=ON -DCHAMELEON_USE_MPI=OFF -DBLA_VENDOR=Intel10_64lp_seq"
+    SLURM_CONSTRAINTS: "sirocco"
+    LD_PRELOAD: "/usr/lib64/libcuda.so"
+
+test_starpu_cuda_master:
+  extends: .test_script_starpu_plafrim_cuda
+  rules:
+    - if: ($CI_COMMIT_BRANCH == $CI_DEFAULT_BRANCH || $CI_COMMIT_BRANCH =~ /^ci-.*$/) && $CI_PIPELINE_SOURCE != "schedule"
+  variables:
+    TESTS_RESTRICTION: "-R test_shm_gpu"
+
+test_starpu_cuda_mr:
+  extends: .test_script_starpu_plafrim_cuda
+  rules:
+    - if: ($CI_PIPELINE_SOURCE == "merge_request_event" && $CI_MERGE_REQUEST_SOURCE_BRANCH_NAME !~ /^notest-.*$/)
+  variables:
+    TESTS_RESTRICTION: "-R test_shm_gpu_d|test_shm_gpu_c"

.gitlab/test_starpu_simgrid.yml

@@ -7,7 +7,6 @@ test_starpu_simgrid_simu_all:
   variables:
     STARPU_WORKERS_NOBIND: 1
     CHAM_CI_ENV_ARG: simgrid
-    SYSTEM: linux
     VERSION: starpu_simgrid
     CATEGORY: simu
     PRECISION: all

.gitlab/validate.sh

-#!/usr/bin/env bash
-###
-#
-#  @file validate.sh
-#  @copyright 2023-2024 Bordeaux INP, CNRS (LaBRI UMR 5800), Inria,
-#                       Univ. Bordeaux. All rights reserved.
-#
-#  @version 1.2.0
-#  @author Mathieu Faverge
-#  @author Florent Pruvost
-#  @date 2023-09-22
-#
-###
-
-# Check some metrics on sonarqube (https://sonarqube.inria.fr/sonarqube/)
-# and depending on the value return 0 (success) or 1 (failure).
-
-if [ $# -gt 0 ]; then
-  METRIC=$1
-fi
-METRIC=${METRIC:-BUG}
-
-if [[ -z $CI_MERGE_REQUEST_IID || -z $CI_PROJECT_NAMESPACE || -z $CI_PROJECT_NAME ]]; then
-  echo "One of the variables CI_MERGE_REQUEST_IID, CI_PROJECT_NAMESPACE,
-  CI_PROJECT_NAME is empty. This script must be used during a gitlab merge
-  request only -> Failure."
-  exit 1
-fi
-
-if [[ -z $SONARQUBE_LOGIN ]]; then
-  echo "SONARQUBE_LOGIN is empty, please give a valid sonarqube user's token,
-  with permissions set on the project -> Failure."
-  exit 1
-fi
-
-if [[ $METRIC == "BUG" ]]; then
-  BUG=`curl -u $SONARQUBE_LOGIN: -X GET "https://sonarqube.inria.fr/sonarqube/api/measures/component?component=${CI_PROJECT_NAMESPACE}%3A${CI_PROJECT_NAME}&pullRequest=${CI_MERGE_REQUEST_IID}&metricKeys=new_bugs" |jq '.component.measures[0].period.value' | sed -e "s#\"##g"`
-  echo "BUG=$BUG"
-  if [[ $BUG -gt 0 ]]; then
-    echo "%{BUG} new bugs detected by Sonarqube -> Failure."
-    exit 1
-  else
-    echo "No new bugs detected by Sonarqube -> Success."
-    exit 0
-  fi
-elif [[ $METRIC == "COVERAGE" ]]; then
-  COV=`curl -u $SONARQUBE_LOGIN: -X GET "https://sonarqube.inria.fr/sonarqube/api/measures/component?component=${CI_PROJECT_NAMESPACE}%3A${CI_PROJECT_NAME}&pullRequest=${CI_MERGE_REQUEST_IID}&metricKeys=new_coverage" |jq '.component.measures[0].period.value' | sed -e "s#\"##g" | cut -d "." -f 1`
-  echo "COV=$COV"
-  if [[ $COV == "null" || -z $COV  ]]; then
-    echo "Coverage is empty, certainly that there are no lines of new code (considered during the analysis) to compare -> Success."
-  else
-    if [[ $COV -lt 80 ]]; then
-      echo "Coverage on new lines is ${COV}%, which is < 80% -> Failure."
-      exit 1
-    else
-      echo "Coverage on new lines is ${COV}%, which is >= 80% -> Success."
-      exit 0
-    fi
-  fi
-fi

.gitlab/validate.yml

----
-validate:
-  stage: validate
-  tags: ["docker", "large"]
-  extends: .only-mr
-  needs: [sonarqube_mr]
-  parallel:
-    matrix:
-      - METRIC: [BUG, COVERAGE]
-  script:
-    - ./.gitlab/validate.sh $METRIC
-  allow_failure: true
\ No newline at end of file

CMakeLists.txt

@@ -4,7 +4,7 @@
 #
 # @copyright 2009-2014 The University of Tennessee and The University of
 #                      Tennessee Research Foundation. All rights reserved.
-# @copyright 2012-2024 Bordeaux INP, CNRS (LaBRI UMR 5800), Inria,
+# @copyright 2012-2025 Bordeaux INP, CNRS (LaBRI UMR 5800), Inria,
 #                      Univ. Bordeaux. All rights reserved.
 #
 ###
@@ -31,7 +31,7 @@
 #  @author Loris Lucido
 #  @author Nathan Précigout
 #  @author Abel Calluaud
-#  @date 2024-07-17
+#  @date 2024-12-06
 #
 ###
 cmake_minimum_required(VERSION 3.5)
@@ -349,7 +349,7 @@ endif()
 # Add math library (libm), searched with MORSE module FindM.cmake
 # it normally exists on all common systems provided with a C compiler
 find_package(M REQUIRED)
-morse_export_imported_target(MORSE M m chameleon)
+morse_export_imported_target(MORSE::M m chameleon)
 
 # If simulation we don't enter in kernel functions so that we don't need to link
 # with concerned libraries
@@ -366,12 +366,12 @@ if(NOT CHAMELEON_SIMULATION)
     # CHAMELEON depends on CBLAS
     #---------------------------
     find_package(CBLAS REQUIRED)
-    morse_export_imported_target(MORSE CBLAS cblas chameleon)
+    morse_export_imported_target(MORSE::CBLAS cblas chameleon)
 
     # CHAMELEON depends on LAPACKE
     #-----------------------------
     find_package(LAPACKE REQUIRED)
-    morse_export_imported_target(MORSE LAPACKE lapacke chameleon)
+    morse_export_imported_target(MORSE::LAPACKE lapacke chameleon)
 
     # CHAMELEON depends on CUDA/CUBLAS
     #----------------------------------
@@ -418,8 +418,8 @@ if(NOT CHAMELEON_SIMULATION)
                 "\n   CUDA_LIBRARIES to be found. Be sure you have"
                 "\n   libcuda with your distribution of CUDA.")
             endif()
-            morse_export_imported_target(CUDA CUDA cuda chameleon)
-            morse_export_imported_target(CUDA CUBLAS cublas chameleon)
+            morse_export_imported_target(CUDA::CUDA cuda chameleon)
+            morse_export_imported_target(CUDA::CUBLAS cublas chameleon)
         endif (CUDA_FOUND)
 
     endif(CHAMELEON_USE_CUDA)
@@ -427,10 +427,16 @@ if(NOT CHAMELEON_SIMULATION)
     # CHAMELEON depends on HIP/HIPBLAS
     #----------------------------------
     if (CHAMELEON_USE_HIP)
+
         if (CHAMELEON_USE_HIP_ROC)
 
-            find_package(HIP REQUIRED)
-            find_package(hipblas REQUIRED)
+            # should define targets: hip::host, hip::device
+            find_package(hip REQUIRED CONFIG)
+            message(STATUS "hip target found")
+
+            # should define targets: roc::hipblas
+            find_package(hipblas REQUIRED CONFIG)
+            message(STATUS "hipblas target found")
 
         elseif(CHAMELEON_USE_HIP_CUDA)
 
@@ -466,56 +472,49 @@ if(NOT CHAMELEON_SIMULATION)
             if(CHAMELEON_HIPBLAS_PATH)
                 list(APPEND HIPBLAS_INCLUDE_DIRS "${CHAMELEON_HIPBLAS_PATH}/include")
                 list(APPEND HIPBLAS_LIBRARIES "${CHAMELEON_HIPBLAS_PATH}/lib/libhipblas.so")
+                message(STATUS "HIPBLAS_INCLUDE_DIRS: ${HIP_INCLUDE_DIRS}")
+                message(STATUS "HIPBLAS_LIBRARIES: ${HIPBLAS_LIBRARIES}")
             else()
                 message(FATAL_ERROR "Please set CHAMELEON_HIPBLAS_PATH to your HIPBLAS installation.")
             endif()
-        endif()
 
-        if (HIP_FOUND)
-            message("-- ${Blue}Add definition CHAMELEON_USE_HIP"
-                " - Activate HIP in Chameleon${ColourReset}")
             # create imported target because not provided with old cmake
-            add_library(HIP::HIP INTERFACE IMPORTED)
-            add_library(HIP::HIPBLAS INTERFACE IMPORTED)
-
-            if (CHAMELEON_USE_HIP_CUDA)
-                target_compile_definitions(HIP::HIP INTERFACE "__HIP_PLATFORM_NVIDIA__")
-                set(HIP_INCLUDE_DIRS "${HIP_INCLUDE_DIRS};${CUDA_INCLUDE_DIRS}")
-                set(HIP_LIBRARIES "${HIP_LIBRARIES};${CUDA_LIBRARIES}")
-            endif()
+            add_library(hip::hip INTERFACE IMPORTED)
+            add_library(roc::hipblas INTERFACE IMPORTED)
+            target_compile_definitions(hip::hip INTERFACE "__HIP_PLATFORM_NVIDIA__")
+            set(HIP_INCLUDE_DIRS "${HIP_INCLUDE_DIRS};${CUDA_INCLUDE_DIRS}")
+            set(HIP_LIBRARIES "${HIP_LIBRARIES};${CUDA_LIBRARIES}")
 
             if (HIP_INCLUDE_DIRS)
-                set_target_properties(HIP::HIP PROPERTIES INTERFACE_INCLUDE_DIRECTORIES "${HIP_INCLUDE_DIRS}")
+                set_target_properties(hip::hip PROPERTIES INTERFACE_INCLUDE_DIRECTORIES "${HIP_INCLUDE_DIRS}")
                 if (HIPBLAS_INCLUDE_DIRS)
-                    set_target_properties(HIP::HIPBLAS PROPERTIES INTERFACE_INCLUDE_DIRECTORIES "${HIPBLAS_INCLUDE_DIRS}")
+                    set_target_properties(roc::hipblas PROPERTIES INTERFACE_INCLUDE_DIRECTORIES "${HIPBLAS_INCLUDE_DIRS}")
                 else()
-                    message(WARNING "CHAMELEON_USE_HIP requires"
+                    message(WARNING "CHAMELEON_USE_HIP_CUDA requires"
                         "\n   HIPBLAS_INCLUDE_DIRS to be found. Be sure you have"
                         "\n   hipblas headers with your distribution of HIP.")
                 endif()
             else()
-                message(WARNING "CHAMELEON_USE_HIP requires"
+                message(WARNING "CHAMELEON_USE_HIP_CUDA requires"
                     "\n   HIP_INCLUDE_DIRS to be found. Be sure you have"
                     "\n   hip headers with your distribution of HIP.")
             endif()
             if (HIP_LIBRARIES)
-                set_target_properties(HIP::HIP PROPERTIES INTERFACE_LINK_LIBRARIES "${HIP_LIBRARIES}")
+                set_target_properties(hip::hip PROPERTIES INTERFACE_LINK_LIBRARIES "${HIP_LIBRARIES}")
                 if (HIPBLAS_LIBRARIES)
-                    set_target_properties(HIP::HIPBLAS PROPERTIES INTERFACE_LINK_LIBRARIES "${HIPBLAS_LIBRARIES}")
-                    target_link_libraries(HIP::HIPBLAS INTERFACE HIP::HIP)
-                    message("-- ${Blue}Add definition CHAMELEON_USE_HIPBLAS"
-                        " - Use GPU kernels from hipblas${ColourReset}")
+                    set_target_properties(roc::hipblas PROPERTIES INTERFACE_LINK_LIBRARIES "${HIPBLAS_LIBRARIES}")
+                    target_link_libraries(roc::hipblas INTERFACE hip::hip)
                 else()
-                    message(FATAL_ERROR "CHAMELEON_USE_HIP requires"
+                    message(FATAL_ERROR "CHAMELEON_USE_HIP_CUDA requires"
                         "\n   HIPBLAS_LIBRARIES to be found. Be sure you have"
                         "\n   libhipblas with your distribution of HIP.")
                 endif()
             else()
-                message(FATAL_ERROR "CHAMELEON_USE_HIP requires"
+                message(FATAL_ERROR "CHAMELEON_USE_HIP_CUDA requires"
                     "\n   HIP_LIBRARIES to be found. Be sure you have"
                     "\n   libamdhip64 with your distribution of ROCm.")
             endif()
-            if (CHAMELEON_USE_HIP_CUDA AND NOT CUDA_LIBRARIES)
+            if (NOT CUDA_LIBRARIES)
                 message(FATAL_ERROR "CHAMELEON_USE_HIP_CUDA requires"
                 "\n   CUDA_LIBRARIES to be found. Be sure you have"
                 "\n   libcuda with your distribution of CUDA.")
@@ -531,17 +530,13 @@ if(NOT CHAMELEON_SIMULATION)
             message("-- HIPBLAS include path:  ${HIPBLAS_INCLUDE_DIRS}")
             message("-- HIPBLAS libraries:     ${HIPBLAS_LIBRARIES}")
 
-            morse_export_imported_target(HIP HIP hip chameleon)
-            morse_export_imported_target(HIP HIPBLAS hipblas chameleon)
-            if (CHAMELEON_USE_HIP_ROC)
-                morse_export_imported_target(roc hipblas rochipblas chameleon)
-                morse_export_imported_target(hip host hiphost chameleon)
-                morse_export_imported_target(hip device hipdevice chameleon)
-                morse_export_imported_target(hip amdhip64 hipamdhip64 chameleon)
-            endif()
-        else(HIP_FOUND)
-            message(FATAL_ERROR "CHAMELEON_USE_HIP requires HIP to be found")
-        endif (HIP_FOUND)
+            morse_export_imported_target(hip::hip hip chameleon)
+            morse_export_imported_target(roc::hipblas hipblas chameleon)
+
+        endif()
+
+        message("-- ${Blue}Add definition CHAMELEON_USE_HIP"
+        " - Activate HIP in Chameleon${ColourReset}")
 
     endif(CHAMELEON_USE_HIP)
 
@@ -562,7 +557,7 @@ if(NOT CHAMELEON_SIMULATION)
             " - Activate MPI in Chameleon${ColourReset}")
         endif (MPI_C_FOUND)
 
-        morse_export_imported_target(MPI MPI_C mpi chameleon)
+        morse_export_imported_target(MPI::MPI_C mpi chameleon)
 
         if (CHAMELEON_SCHED STREQUAL "PARSEC")
           message(FATAL_ERROR "CHAMELEON does not support PARSEC with MPI."
@@ -656,7 +651,7 @@ else (NOT CHAMELEON_SIMULATION)
             " - Activate MPI in Chameleon${ColourReset}")
         endif (MPI_C_FOUND)
 
-        morse_export_imported_target(MPI MPI_C mpi chameleon)
+        morse_export_imported_target(MPI::MPI_C mpi chameleon)
 
     endif (CHAMELEON_USE_MPI)
 
@@ -766,7 +761,7 @@ add_custom_target(chameleon_all_sources ALL DEPENDS ${CHAMELEON_SOURCES_TARGETS}
 
 if(CHAMELEON_ENABLE_DOC)
   add_subdirectory(doc)
-  if (DOXYGEN_EXECUTABLE AND EMACS_COMPILER AND TEX_COMPILER)
+  if (DOXYGEN_EXECUTABLE AND EMACS_COMPILER)
     add_custom_target(doc ALL DEPENDS doc-homepage doxygen-out)
   endif()
 endif()
@@ -791,7 +786,7 @@ endif()
 # Threads::Threads may be a dependency of BLAS/LAPACK and CUDA
 # the target may be required for cmake users linking with CHAMELEONconfig.cmake
 if (TARGET Threads::Threads)
-    morse_export_imported_target(Threads Threads threads chameleon)
+    morse_export_imported_target(Threads::Threads threads chameleon)
 endif()
 
 # see https://cmake.org/cmake/help/latest/module/CMakePackageConfigHelpers.html
@@ -801,10 +796,20 @@ set(BIN_INSTALL_DIR "bin/" CACHE STRING "where to install executables relative t
 set(INC_INSTALL_DIR "include/" CACHE STRING "where to install headers relative to prefix" )
 set(LIB_INSTALL_DIR "lib/" CACHE STRING "where to install libraries relative to prefix" )
 
+set(PREFIX_PATH "")
+if (CMAKE_PREFIX_PATH)
+  list(APPEND PREFIX_PATH ${CMAKE_PREFIX_PATH})
+endif()
+if (DEFINED ENV{CMAKE_PREFIX_PATH})
+  set(ENV_PREFIX_PATH $ENV{CMAKE_PREFIX_PATH})
+  string(REGEX REPLACE ":" ";" ENV_PREFIX_PATH "${ENV_PREFIX_PATH}" )
+  list(APPEND PREFIX_PATH ${ENV_PREFIX_PATH})
+endif()
+
 configure_package_config_file(cmake_modules/CHAMELEONConfig.cmake.in
                               ${CMAKE_CURRENT_BINARY_DIR}/CHAMELEONConfig.cmake
                               INSTALL_DESTINATION ${LIB_INSTALL_DIR}/cmake/chameleon
-                              PATH_VARS BIN_INSTALL_DIR INC_INSTALL_DIR LIB_INSTALL_DIR)
+                              PATH_VARS BIN_INSTALL_DIR INC_INSTALL_DIR LIB_INSTALL_DIR PREFIX_PATH)
 write_basic_package_version_file(CHAMELEONConfigVersion.cmake
                                  VERSION ${CHAMELEON_VERSION}
                                  COMPATIBILITY AnyNewerVersion)
@@ -821,6 +826,8 @@ include(ChameleonPkgConfig)
 # Create .pc file
 # ---------------
 chameleon_generate_pkgconfig_files()
+
+generate_newenv_file(PACKAGE chameleon)
 #------------------------------------------------------------------------------
 
 ###############################################################################

ChangeLog

+chameleon-1.4.0
+------------------------------------------------------------------------
+ - map: Add an option to dynamically change from synchronous to asynchronous when calling map function, to serialize print for example.
+ - Add the laswp driver and testing. Be careful, the interface does not follow the lapack API to propose the op(P) \times A, or A \times op(P) operation with op(P) equal to P or P^{-1}
+ - Add the gesv driver to perform LU factorization and solve with its associated testing.
+ - StarPU: Update the minimum requirement from 1.3 to 1.4
+ - StarPU: When using starpu > 1.4.8, use the new distributed submit interface in the codelets instead of the classical insert task interface.
+ - ci: use -Werror to prevent from adding warning to the code
+ - Fix: all warning reported by the switch to -Werror
+ - testing: Set tiles allocation by default.
+
 chameleon-1.3.0
 ------------------------------------------------------------------------
- - mixed-precision: introduce descripto with precision adapted to local norms
+ - mixed-precision: introduce descriptor with precision adapted to local norms
         - Add CHAMELEON_[dz]gered... functions to reduce the precision of the tiles based on a requested accuracy
         - Add CHAMELEON_[dz]gerst... functions to restore the original numerical precision of the tiles in a descriptor
  - types: add support for half precision arithmetic into the data descriptors
  - cuda: add half precision conversion kernels, and variants of the gemm kernels (hgemm, and gemmex)
+ - cuda: Check error after lauching kernels
  - descriptors: Add the possibility to pass arguments to the rankof
    function. This is used to provide custom distribuitions through a
    given file. *WARNING*: It changes the interface of
-   CHAMELEON_Desc_Create_User that requires aan additional `, NULL`
-   parameters in the general case.
+   CHAMELEON_Desc_Create_User that requires an additional `, NULL`
+        parameters in the general case.
+ - control: Defined the default parameters through environment variables first and make sure the testings use the default value instead of overwritting them.
+ - control: Make the CHAM_context_t structure public, and provide a function to the user to access the pointer in case of the development of its own functionalities using the RUNTIME API.
+ - compute: Refactor the code that compute the kernel dimension to potentially enable variadic tile sizes
+ - compute/map: Rework the map functions family to be able to pass multiple descriptor with parameterized access types.
+ - compute/getrf: Add a basic LU factorization with partial pivoting (WARNING: this functionnality is still under development and does not provide full performance yet)
  - compute/poinv: Add the possibility to use an intermediate distribution for the TRTRI operation
+ - compute/getrf_nopiv: Add lookahead through temporary buffers to better regulate the communication allocations
+ - runtime/starpu: Whenever possible replace the lacpy codelet by a direct memory copy from the input handler to the output one
+ - runtime/starpu: better separation of the public interface from the internal interface for code reusing the RUNTIME API
+ - testings: Display in help message the option possible values when possible
+ - bug: fix issue with undefined vasprintf
+ - bug: Make sure generic algorithms are used when at least one of the data descriptor is not 2D block cyclic and might cause issues.
+ - bug/starpu: Fix the --forcegpu option to integrate HIP devices withing the option and make sure it's applied only when possible
+ - Fix issue 124: RP_CHAMELEON_PRECISION is the set of supported precisions, while CHAMELEON_PRECISION is the set of enabled precisions
+ - Fix the trsm flops issue that was miscalculated.
+ - Fix integer overflow in malloc where size_t was not used
+ - ci/docker: provide a simpler docker image dedicated to the project
 
 chameleon-1.2.0
 ------------------------------------------------------------------------

LICENCE.txt

@@ -4,7 +4,7 @@
 #
 # @copyright 2009-2014 The University of Tennessee and The University of
 #                      Tennessee Research Foundation. All rights reserved.
-# @copyright 2012-2024 Bordeaux INP, CNRS (LaBRI UMR 5800), Inria,
+# @copyright 2012-2025 Bordeaux INP, CNRS (LaBRI UMR 5800), Inria,
 #                      Univ. Bordeaux. All rights reserved.
 # @copyright 2016-2018 KAUST. All rights reserved.
 #

cmake_modules/CHAMELEONConfig.cmake.in

@@ -9,6 +9,9 @@ set_and_check(CHAMELEON_LIB_DIR "@PACKAGE_LIB_INSTALL_DIR@")
 
 check_required_components(CHAMELEON)
 
+# propagate the CMAKE_PREFIX_PATH informations used during chameleon installation to give paths to dependencies
+list(APPEND CMAKE_PREFIX_PATH "@PACKAGE_PREFIX_PATH@")
+
 # dependencies of CHAMELEON
 include("${CMAKE_CURRENT_LIST_DIR}/mTargets.cmake")
 if (EXISTS "${CMAKE_CURRENT_LIST_DIR}/threadsTargets.cmake")
@@ -38,13 +41,13 @@ if (@CHAMELEON_USE_CUDA@ AND NOT @CHAMELEON_SIMULATION@)
   include("${CMAKE_CURRENT_LIST_DIR}/cublasTargets.cmake")
 endif()
 if (@CHAMELEON_USE_HIP@ AND NOT @CHAMELEON_SIMULATION@)
-  include("${CMAKE_CURRENT_LIST_DIR}/hipTargets.cmake")
-  include("${CMAKE_CURRENT_LIST_DIR}/hipblasTargets.cmake")
   if (@CHAMELEON_USE_HIP_ROC@)
-    include("${CMAKE_CURRENT_LIST_DIR}/rochipblasTargets.cmake")
-    include("${CMAKE_CURRENT_LIST_DIR}/hiphostTargets.cmake")
-    include("${CMAKE_CURRENT_LIST_DIR}/hipdeviceTargets.cmake")
-    include("${CMAKE_CURRENT_LIST_DIR}/hipamdhip64Targets.cmake")
+    include(CMakeFindDependencyMacro)
+    find_dependency(hip REQUIRED)
+    find_dependency(hipblas REQUIRED)
+  else ()
+    include("${CMAKE_CURRENT_LIST_DIR}/hipTargets.cmake")
+    include("${CMAKE_CURRENT_LIST_DIR}/hipblasTargets.cmake")
   endif()
 endif()
 

cmake_modules/ChameleonPkgConfig.cmake

@@ -2,17 +2,17 @@
 #
 # @file ChameleonPkgConfig.cmake
 #
-# @copyright 2012-2024 Bordeaux INP, CNRS (LaBRI UMR 5800), Inria,
+# @copyright 2012-2025 Bordeaux INP, CNRS (LaBRI UMR 5800), Inria,
 #                      Univ. Bordeaux. All rights reserved.
 #
 ###
 #
 #  @project Chameleon
-#  @version 1.2.0
+#  @version 1.3.0
 #  @author Mathieu Faverge
 #  @author Florent Pruvost
 #  @author Abel Calluaud
-#  @date 2022-02-22
+#  @date 2024-12-16
 #
 ###
 include(GenPkgConfig)
@@ -93,7 +93,6 @@ macro(chameleon_generate_pkgconfig_files)
 
     if(CHAMELEON_USE_CUDA)
       list(APPEND GPUCUBLAS_PKGCONFIG_LIBS_PRIVATE ${CUDA_CUBLAS_LIBRARIES})
-      list(APPEND GPUCUBLAS_PKGCONFIG_REQUIRED "cuda")
       list(APPEND CHAMELEON_PKGCONFIG_REQUIRED "gpucublas")
     endif()
 
@@ -110,7 +109,28 @@ macro(chameleon_generate_pkgconfig_files)
   )
 
   if(CHAMELEON_USE_MPI)
-    list(APPEND CHAMELEON_PKGCONFIG_REQUIRED "mpi")
+    if("${MPI_C_LIBRARIES}" MATCHES "mpich")
+      set(MPI_NAME "mpich")
+    elseif("${MPI_C_LIBRARIES}" MATCHES "mvapich2")
+      set(MPI_NAME "mvapich2")
+    elseif("${MPI_C_LIBRARIES}" MATCHES "madmpi" OR "${MPI_C_LIBRARIES}" MATCHES "nmad")
+      set(MPI_NAME "nmad")
+    elseif("${MPI_C_LIBRARIES}" MATCHES "openmpi")
+      set(MPI_NAME "ompi")
+    else()
+      set(MPI_NAME "mpi")
+    endif()
+    find_file(MPIPC_PATH "${MPI_NAME}.pc"
+              HINTS ${CMAKE_C_IMPLICIT_LINK_DIRECTORIES}
+              ENV PKG_CONFIG_PATH
+              PATH_SUFFIXES pkgconfig
+              NO_CACHE)
+    if(MPIPC_PATH)
+      list(APPEND CHAMELEON_PKGCONFIG_REQUIRED ${MPI_NAME})
+    else()
+      list(APPEND CHAMELEON_PKGCONFIG_INCS "${MPI_C_INCLUDE_DIRS}")
+      list(APPEND CHAMELEON_PKGCONFIG_LIBS "${MPI_C_LIBRARIES}")
+    endif()
   endif()
 
   generate_pkgconfig_files(

cmake_modules/PrintOpts.cmake

@@ -4,7 +4,7 @@
 #
 # @copyright 2009-2014 The University of Tennessee and The University of
 #                      Tennessee Research Foundation. All rights reserved.
-# @copyright 2012-2024 Bordeaux INP, CNRS (LaBRI UMR 5800), Inria,
+# @copyright 2012-2025 Bordeaux INP, CNRS (LaBRI UMR 5800), Inria,
 #                      Univ. Bordeaux. All rights reserved.
 #
 ###
@@ -17,11 +17,12 @@
 #     Univ. of California Berkeley,
 #     Univ. of Colorado Denver.
 #
-# @version 1.2.0
+# @version 1.3.0
 #  @author Florent Pruvost
 #  @author Mathieu Faverge
 #  @author Loris Lucido
-#  @date 2023-01-30
+#  @author Abel Calluaud
+#  @date 2024-03-27
 #
 ###
 

cmake_modules/gitlab-ci-initial-cache.cmake

@@ -2,7 +2,11 @@ set(BUILD_SHARED_LIBS "ON" CACHE BOOL "")
 set(CMAKE_INSTALL_PREFIX "$ENV{PWD}/install-$ENV{VERSION}" CACHE PATH "")
 set(CMAKE_VERBOSE_MAKEFILE "ON" CACHE BOOL "")
 
-set(CMAKE_C_FLAGS "-Werror")
+if(CMAKE_C_FLAGS)
+  set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -Werror" CACHE STRING "")
+else()
+  set(CMAKE_C_FLAGS "-Werror" CACHE STRING "")
+endif()
 
 option(MORSE_ENABLE_WARNING  "Enable warning messages"        ON)
 option(MORSE_ENABLE_COVERAGE "Enable flags for coverage test" ON)

cmake_modules/local_subs.py

@@ -3,7 +3,7 @@
 
  Python Chameleon specific substitution rules for the Precision Generator script.
 
- @copyright 2019-2024 Bordeaux INP, CNRS (LaBRI UMR 5800), Inria,
+ @copyright 2019-2025 Bordeaux INP, CNRS (LaBRI UMR 5800), Inria,
                       Univ. Bordeaux. All rights reserved.
 
  @version 1.3.0
@@ -11,7 +11,7 @@
  @author Florent Pruvost
  @author Nathalie Furmento
  @author Alycia Lisito
- @date 2023-07-06
+ @date 2024-11-14
 
 """
 _extra_blas = [
@@ -51,6 +51,9 @@ _extra_blas = [
     ('',                     'sprint',               'dprint',               'cprint',               'zprint'              ),
     ('',                     'sgered',               'dgered',               'cgered',               'zgered'              ),
     ('',                     'sgerst',               'dgerst',               'cgerst',               'zgerst'              ),
+    ('',                     'sipiv_allreduce',      'dipiv_allreduce',      'cipiv_allreduce',      'zipiv_allreduce'     ),
+    ('',                     'sperm_allreduce',      'dperm_allreduce',      'cperm_allreduce',      'zperm_allreduce'     ),
+    ('',                     'sperm_reduce',      'dperm_reduce',      'cperm_reduce',      'zperm_reduce'     ),
 ]
 
 _extra_BLAS = [ [ x.upper() for x in row ] for row in _extra_blas ]
@@ -96,7 +99,7 @@ subs = {
         ('',                     'ORGLQ',                'ORGLQ',                'UNGLQ',                'UNGLQ'               ),
         ('',                     'SYEV',                 'SYEV',                 'HEEV',                 'HEEV'                ),
         ('',                     'SYG',                  'SYG',                  'HEG',                  'HEG'                 ),
-        ('',                     'const float ',         'const double ',        'const void \*',        'const void \*'       ),
+        ('',                     'const float ',         'const double ',       r'const void \*',       r'const void \*'       ), # Line that is extremely dangerous.
     ]
     + _extra_blas
     + _extra_BLAS

morse_cmake @ bc34cfd2

-Subproject commit 34e01580a425eb5fc13f5495a9a602fb26a3a589
+Subproject commit bc34cfd2ce19d525e2e6a82b7ec6c8bca45e2c97

compute/CMakeLists.txt

@@ -4,7 +4,7 @@
 #
 # @copyright 2009-2014 The University of Tennessee and The University of
 #                      Tennessee Research Foundation. All rights reserved.
-# @copyright 2012-2024 Bordeaux INP, CNRS (LaBRI UMR 5800), Inria,
+# @copyright 2012-2025 Bordeaux INP, CNRS (LaBRI UMR 5800), Inria,
 #                      Univ. Bordeaux. All rights reserved.
 #
 ###
@@ -28,7 +28,8 @@
 #  @author Loris Lucido
 #  @author Matthieu Kuhn
 #  @author Ana Hourcau
-#  @date 2024-07-17
+#  @author Matteo Marcos
+#  @date 2025-03-24
 #
 ###
 
@@ -40,11 +41,12 @@ set(CHAMELEON_CONTROL
     ../control/context.c
     ../control/control.c
     ../control/descriptor.c
-    ../control/descriptor_rec.c
     ../control/descriptor_helpers.c
     ../control/descriptor_ipiv.c
-    ../control/workspace.c
+    ../control/descriptor_rec.c
+    ../control/lamch.c
     ../control/tile.c
+    ../control/workspace.c
     ../control/chameleon_f77.c
     ../control/chameleon_mf77.c
     map.c
@@ -149,7 +151,7 @@ set(ZSRC
     zgepdf_qr.c
     zgeqrs.c
     zgeqrs_param.c
-    #zgesv.c
+    zgesv.c
     zgesv_incpiv.c
     zgesv_nopiv.c
     #zgetrf.c
@@ -158,6 +160,7 @@ set(ZSRC
     zgetrf.c
     zgetrs_incpiv.c
     zgetrs_nopiv.c
+    zgetrs.c
     zlacpy.c
     zlange.c
     zlanhe.c
@@ -218,8 +221,8 @@ set(ZSRC
     #pzhetrd_hb2ht.c
     pzhetrd_he2hb.c
     #pzlarft_blgtrd.c
-    #pzlaswp.c
-    #pzlaswpc.c
+    pzlaswp.c
+    pzlaswpc.c
     #pztrsmrv.c
     #pzunmqr_blgtrd.c
     #########################
@@ -236,7 +239,7 @@ set(ZSRC
     #zhegv.c
     #zhegvd.c
     zhetrd.c
-    #zlaswp.c
+    zlaswp.c
     #zlaswpc.c
     #ztrsmrv.c
     ##################
@@ -374,7 +377,7 @@ endif()
 if (CHAMELEON_USE_HIP)
   if (NOT CHAMELEON_SIMULATION)
     target_link_libraries(chameleon PUBLIC gpuhipblas)
-    target_link_libraries(chameleon PUBLIC HIP::HIPBLAS)
+    target_link_libraries(chameleon PUBLIC roc::hipblas)
   endif()
 endif()
 target_link_libraries(chameleon PUBLIC MORSE::M)

compute/gepdf.h

@@ -2,16 +2,16 @@
  *
  * @file gepdf.h
  *
- * @copyright 2022-2024 Bordeaux INP, CNRS (LaBRI UMR 5800), Inria,
+ * @copyright 2022-2025 Bordeaux INP, CNRS (LaBRI UMR 5800), Inria,
  *                      Univ. Bordeaux. All rights reserved.
  *
  ***
  *
  * @brief Chameleon GEPDF auxiliary function.
  *
- * @version 1.2.0
+ * @version 1.3.0
  * @author Mathieu Faverge
- * @date 2022-02-22
+ * @date 2024-02-18
  *
  */
 #ifndef _chameleon_gepdf_h_

compute/map.c

@@ -2,7 +2,7 @@
  *
  * @file map.c
  *
- * @copyright 2018-2024 Bordeaux INP, CNRS (LaBRI UMR 5800), Inria,
+ * @copyright 2018-2025 Bordeaux INP, CNRS (LaBRI UMR 5800), Inria,
  *                      Univ. Bordeaux. All rights reserved.
  *
  ***
@@ -11,7 +11,7 @@
  *
  * @version 1.3.0
  * @author Mathieu Faverge
- * @date 2024-03-11
+ * @date 2024-03-14
  *
  */
 #include "control/common.h"
@@ -37,10 +37,11 @@ map_cpu( void *op_args,
 }
 
 static cham_map_operator_t map_op = {
-    .name     = "map",
-    .cpufunc  = map_cpu,
-    .cudafunc = NULL,
-    .hipfunc  = NULL,
+    .name        = "map",
+    .cpufunc     = map_cpu,
+    .cudafunc    = NULL,
+    .hipfunc     = NULL,
+    .synchronous = 0,
 };
 
 /**