remove cupla in comments and documentation

- update clang-format and remove cupla include rule - update CI checks for cupla includes - update documantion (dependency image) - update code comments
ComputationalRadiationPhysics · Mar 4, 2024 · 208af4d · 208af4d
1 parent 1c425a2
commit 208af4d
Show file tree

Hide file tree

Showing 32 changed files with 92 additions and 136 deletions.
diff --git a/.clang-format b/.clang-format
@@ -113,22 +113,19 @@ IncludeCategories:
     # <pmacc/foo.hpp>
   - Regex: '<pmacc/([A-Za-z0-9.\/-_])+>'
     Priority: 4
-    # <cupla/foo.hpp>
-  - Regex: '<cupla/([A-Za-z0-9.\/-_])+>'
-    Priority: 5
   # <alpaka/foo.hpp>
   - Regex: '<alpaka/([A-Za-z0-9.\/-_])+>'
-    Priority: 6
+    Priority: 5
   # <boost/foo.hpp>
   - Regex: '<boost/([A-Za-z0-9.\/-_])+>'
-    Priority: 7
+    Priority: 6
   # C++ standard library headers are the last group to be included
   - Regex: '<([A-Za-z0-9\/-_])+>'
-    Priority: 8
+    Priority: 7
   # Includes that made it this far are third-party headers and will be placed
   # below picongpu/pmacc includes
   - Regex: '<([A-Za-z0-9.\/-_])+>'
-    Priority: 9
+    Priority: 8
 
 # Future options - not supported in clang-format 12
 #

diff --git a/INSTALL.rst b/INSTALL.rst
@@ -107,10 +107,6 @@ alpaka 1.1.X
 """"""""""""""""
 - `alpaka <https://github.com/alpaka-group/alpaka>`_ is included in the PIConGPU source code
 
-cupla 0.5.X-dev
-"""""""""""""""
-- `cupla <https://github.com/alpaka-group/cupla>`_ is included in the PIConGPU source code
-
 mallocMC 2.6.0crp-dev
 """""""""""""""""""""
 - only required for CUDA and HIP backends

diff --git a/docs/source/dev/debugging.rst b/docs/source/dev/debugging.rst
@@ -16,7 +16,7 @@ The following build options can assist the investigation:
 * ``PIC_VERBOSE=<N>`` sets log detail level for PIConGPU, highest level is 127.
 * ``PMACC_VERBOSE=<N>`` sets log detail level for pmacc, highest level is 127.
 * ``PMACC_BLOCKING_KERNEL=ON`` makes each kernel invocation blocking, which helps to narrow a crash down to a particular kernel.
-* ``CUPLA_STREAM_ASYNC_ENABLE=OFF`` disables asynchronous streams, also helps to narrow a crash down to a particular place.
+* ``PMACC_ASYNC_QUEUES=OFF`` disables asynchronous alpaka queues, also helps to narrow a crash down to a particular place.
 * ``CMAKE_BUILD_TYPE=Debug`` compile in debug mode where all assertions are activated. Compiling in debug mode will slowdown your kernels!
 
 These options can be passed when building, or manually modified via cmake.

diff --git a/docs/source/dev/extending.rst b/docs/source/dev/extending.rst
@@ -83,9 +83,9 @@ Otherwise, one often has to implement an own kernel.
 Writing a Kernel
 ----------------
 
-Computational kernels are written using library `cupla <https://github.com/alpaka-group/cupla>`_ which is a layer on top of library `alpaka <https://github.com/alpaka-group/alpaka>`_.
+Computational kernels are written using library `alpaka <https://github.com/alpaka-group/alpaka>`_.
 Most kernel functors are templates parametrized with the number of threads per block, often called ``numWorkers``.
-Kernel invocations are wrapped into a helper macro ``PMACC_KERNEL``.
+Kernel invocations are wrapped into a helper macro ``PMACC_KERNEL`` or ``PMACC_LOCKSTEP_KERNEL``.
 
 A vast majority of PIConGPU kernels operate on two levels of parallelism: between supercells and inside each supercell.
 This parallel pattern is covered by the mapper concept.

diff --git a/docs/source/install/libraryDependencies.dot b/docs/source/install/libraryDependencies.dot
@@ -24,7 +24,6 @@ digraph PIConGPU {
 
     compiler -> alpaka;
     boost -> alpaka;
-    alpaka -> cupla;
 
     mallocmc [label="mallocMC"]
     alpaka -> mallocmc;
@@ -53,7 +52,6 @@ digraph PIConGPU {
 
     picongpu [label="PIConGPU"];
     alpaka -> picongpu;
-    cupla -> picongpu;
     mpi -> picongpu;
     boost -> picongpu;
     mallocmc -> picongpu;

diff --git a/include/picongpu/fields/currentDeposition/Strategy.def b/include/picongpu/fields/currentDeposition/Strategy.def
@@ -189,7 +189,7 @@ namespace picongpu
 
             /** Default strategy for the current deposition
              *
-             * Default will be selected based on the cupla accelerator.
+             * Default will be selected based on the alpaka accelerator.
              *
              * @tparam T_Acc the accelerator type
              */

diff --git a/include/picongpu/particles/Particles.hpp b/include/picongpu/particles/Particles.hpp
@@ -51,7 +51,7 @@ namespace picongpu
     using namespace pmacc;
 
 #if(!BOOST_LANG_CUDA && !BOOST_COMP_HIP)
-    /* dummy because we are not using mallocMC with cupla
+    /* dummy because we are not using mallocMC with CPU backends
      * DeviceHeap is defined in `mallocMC.param`
      */
     struct DeviceHeap

diff --git a/include/picongpu/particles/atomicPhysics/electronDistribution/histogram/Histogram.hpp b/include/picongpu/particles/atomicPhysics/electronDistribution/histogram/Histogram.hpp
@@ -154,9 +154,6 @@ namespace picongpu
                             else
                             {
                                 // Otherwise we add it to a collection of new bins
-                                // Note: in current dev the namespace is different in cupla
-                                // get Index where to deposit it by atomic add to numNewBins
-                                // this assures that the same index is not used twice
                                 auto newBinIdx = alpaka::atomicAdd(
                                     worker.getAcc(),
                                     &numNewBins,

diff --git a/include/picongpu/plugins/BinEnergyParticles.hpp b/include/picongpu/plugins/BinEnergyParticles.hpp
@@ -71,7 +71,7 @@ namespace picongpu
          *
          * @tparam T_ParBox pmacc::ParticlesBox, particle box type
          * @tparam T_BinBox pmacc::DataBox, box type for the histogram in global memory
-         * @tparam T_Mapping type of the mapper to map a cupla block to a supercell index
+         * @tparam T_Mapping type of the mapper to map an alpaka block to a supercell index
          * @tparam T_Worker lockstep worker type
          *
          * @param acc alpaka accelerator
@@ -80,7 +80,7 @@ namespace picongpu
          * @param numBins number of bins in the histogram (must be fit into the shared memory)
          * @param minEnergy particle energy for the first bin
          * @param maxEnergy particle energy for the last bin
-         * @param mapper functor to map a cupla block to a supercells index
+         * @param mapper functor to map an alpaka block to a supercells index
          */
         template<typename T_ParBox, typename T_BinBox, typename T_Mapping, typename T_Filter, typename T_Worker>
         DINLINE void operator()(

diff --git a/include/picongpu/plugins/kernel/CopySpecies.kernel b/include/picongpu/plugins/kernel/CopySpecies.kernel
@@ -40,7 +40,7 @@ namespace picongpu
          * @tparam T_Filter type of filer with particle selection rules
          * @tparam T_Space type of coordinate description
          * @tparam T_Identifier type of identifier for the particle cellIdx
-         * @tparam T_Mapping type of the mapper to map cupla idx to supercells
+         * @tparam T_Mapping type of the mapper to map alpaka idx to supercells
          * @tparam T_Worker lockstep worker type
          *
          * @param worker lockstep worker
@@ -54,7 +54,7 @@ namespace picongpu
          * @param domainCellIdxIdentifier the identifier for the particle cellIdx
          *                                that is calculated with respect to
          *                                domainOffset
-         * @param mapper map cupla idx to supercells
+         * @param mapper map alpaka idx to supercells
          */
         template<
             typename T_DestFrame,

diff --git a/include/picongpu/plugins/openPMD/Json.cpp b/include/picongpu/plugins/openPMD/Json.cpp
@@ -33,7 +33,7 @@
 /*
  * Note:
  * This is a hostonly .cpp file because CMake will not use -isystem for system
- * include paths on NVCC targets created with cupla_add_executable.
+ * include paths on NVCC targets created with alpaka_add_executable.
  * Since <nlohmann/json.hpp> throws a number of warnings, this .cpp file
  * ensures that NVCC never sees that library.
  */

diff --git a/include/picongpu/plugins/openPMD/Json_private.hpp b/include/picongpu/plugins/openPMD/Json_private.hpp
@@ -32,7 +32,7 @@
  * Note:
  * This header is included only into hostonly .cpp files because CMake
  * will not use -isystem for system include paths on NVCC targets created
- * with cupla_add_executable.
+ * with alpaka_add_executable.
  * Since <nlohmann/json.hpp> throws a number of warnings, this design
  * ensures that NVCC never sees that library.
  */

diff --git a/include/pmacc/Environment.tpp b/include/pmacc/Environment.tpp
@@ -259,10 +259,10 @@ namespace pmacc
             int numAvailableDevices = manager::Device<ComputeDevice>::get().count();
 
 #if(BOOST_LANG_CUDA || BOOST_COMP_HIP)
-            // ##ERROR handling
-            if(numAvailableDevices < 1) // check if cupla device is found
+            // check if device is found
+            if(numAvailableDevices < 1)
             {
-                throw std::runtime_error("no CUDA capable devices detected");
+                throw std::runtime_error("no capable alpaka compute devices detected");
             }
 #endif
 

diff --git a/include/pmacc/PMaccConfig.cmake b/include/pmacc/PMaccConfig.cmake
@@ -90,8 +90,8 @@ endif()
 # setup alpaka
 ################################################################################
 
-# the min and max. supported alpaka version is also copied to the cuplaConfig.cmake
-set(_PMACC_MIN_ALPAKA_VERSION 1.0.0)
+# the min and max. supported alpaka version
+set(_PMACC_MIN_ALPAKA_VERSION 1.1.0)
 set(_PMACC_MAX_ALPAKA_VERSION 1.1.0)
 
 # do not search for alpaka if it already exists

diff --git a/include/pmacc/algorithms/GlobalReduce.hpp b/include/pmacc/algorithms/GlobalReduce.hpp
@@ -36,8 +36,6 @@ namespace pmacc
         {
         public:
             /** Constructor
-             *
-             * @attetion Don't create a instance before you have set you cupla device!
              *
              * @param byte how many bytes in global gpu memory can be reserved for the reduction algorithm
              * @param sharedMemByte limit the usage of shared memory per block on gpu

diff --git a/include/pmacc/alpakaHelper/ValidateCall.hpp b/include/pmacc/alpakaHelper/ValidateCall.hpp
@@ -29,17 +29,15 @@
 #include <stdexcept>
 
 
-/**
- * Print a cupla error message including file/line info to stderr
+/** Print a error message including file/line info to stderr
  */
 #define PMACC_PRINT_ALPAKA_ERROR(msg)                                                                                 \
     do                                                                                                                \
     {                                                                                                                 \
         std::cerr << "[alpaka] Error: <" << __FILE__ << ">:" << __LINE__ << " " << msg << std::endl;                  \
     } while(false)
 
-/**
- * Print a cupla error message including file/line info to stderr and raises an exception
+/** Print a error message including file/line info to stderr and raises an exception
  */
 #define PMACC_PRINT_ALPAKA_ERROR_AND_THROW(msg)                                                                       \
     do                                                                                                                \
@@ -48,10 +46,9 @@
         throw std::runtime_error(std::string("[alpaka] Error: ") + msg);                                              \
     } while(false)
 
-/**
- * Captures CUDA errors and prints messages to stdout, including line number and file.
+/** Captures an expression in a catch throw block and prints messages to stdout, including line number and file.
  *
- * @param cmd command with cuplaError_t return value to check
+ * @param ... expression to capture in a catch throw block
  */
 #define PMACC_CHECK_ALPAKA_CALL(...)                                                                                  \
     do                                                                                                                \

diff --git a/include/pmacc/device/reduce/Kernel.hpp b/include/pmacc/device/reduce/Kernel.hpp
@@ -109,7 +109,7 @@ namespace pmacc
                  *
                  * This method can be used to reduce a chunk of an array.
                  * This method is a **collective** method and needs to be called by all
-                 * threads within a cupla block.
+                 * threads within a alpaka block.
                  *
                  * @tparam T_SrcBuffer type of the buffer
                  * @tparam T_Functor type of the binary functor to reduce two elements
@@ -125,7 +125,7 @@ namespace pmacc
                  *        first argument is the source and get the new reduced value.
                  * @param sharedMem shared memory buffer with storage for `linearThreadIdxInBlock` elements,
                  *        buffer must implement `operator[](size_t)` (one dimensional access)
-                 * @param blockIndex index of the cupla block,
+                 * @param blockIndex index of the alpaka block,
                  *                   for a global reduce: `device::getBlockIdx(worker.getAcc()).x()`,
                  *                   for a reduce within a block: `0`
                  *

diff --git a/include/pmacc/device/threadInfo.hpp b/include/pmacc/device/threadInfo.hpp
@@ -1,21 +1,22 @@
-/* Copyright 2024 Rene Widera
+/* Copyright 2014 Rene Widera
  *
- * This file is part of cupla.
+ * This file is part of PMacc.
  *
- * cupla is free software: you can redistribute it and/or modify
- * it under the terms of the GNU Lesser General Public License as published by
+ * PMacc is free software: you can redistribute it and/or modify
+ * it under the terms of either the GNU General Public License or
+ * the GNU Lesser General Public License as published by
  * the Free Software Foundation, either version 3 of the License, or
  * (at your option) any later version.
  *
- * cupla is distributed in the hope that it will be useful,
+ * PMacc is distributed in the hope that it will be useful,
  * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- * GNU Lesser General Public License for more details.
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License and the GNU Lesser General Public License
+ * for more details.
  *
- * You should have received a copy of the GNU Lesser General Public License
- * along with cupla.
+ * You should have received a copy of the GNU General Public License
+ * and the GNU Lesser General Public License along with PMacc.
  * If not, see <http://www.gnu.org/licenses/>.
- *
  */
 
 #pragma once

diff --git a/include/pmacc/eventSystem/events/CudaEvent.hpp b/include/pmacc/eventSystem/events/CudaEvent.hpp
@@ -35,16 +35,16 @@ namespace pmacc
     class CudaEvent
     {
     private:
-        /** native cupla event */
+        /** native alpaka event */
         AlpakaEventType event;
-        /** native cupla stream where the event is recorded
+        /** native alpaka queue where the event is recorded
          *
          *  only valid if isRecorded is true
          */
         std::optional<AccStream> stream;
         /** state if a recorded event is finished
          *
-         * avoid cupla driver calls after `isFinished()` returns the first time true
+         * avoids that alpaka calls backend API methods after `isFinished()` returns the first time true
          */
         bool finished{true};
 
@@ -55,7 +55,7 @@ namespace pmacc
     public:
         /** Constructor
          *
-         * if called before the cupla device is initialized the behavior is undefined
+         * if called before the alpaka device is initialized the behavior is undefined
          */
         CudaEvent();
 
@@ -70,7 +70,7 @@ namespace pmacc
 
         /** get native AlpakaEventType object
          *
-         * @return native cupla event
+         * @return native alpaka event
          */
         AlpakaEventType operator*() const
         {
@@ -79,7 +79,7 @@ namespace pmacc
 
         /** get stream in which this event is recorded
          *
-         * @return native cupla stream
+         * @return native alpaka queue
          */
         AccStream getStream() const
         {
@@ -95,7 +95,7 @@ namespace pmacc
 
         /** record event in a device stream
          *
-         * @param stream native cupla stream
+         * @param stream native alpaka queue
          */
         void recordEvent(AccStream const& stream);
     };

diff --git a/include/pmacc/eventSystem/events/CudaEventHandle.hpp b/include/pmacc/eventSystem/events/CudaEventHandle.hpp
@@ -60,9 +60,9 @@ namespace pmacc
         ~CudaEventHandle();
 
         /**
-         * get native cupla event
+         * get native alpaka event
          *
-         * @return native cupla event
+         * @return native alpaka event
          */
         AlpakaEventType operator*() const;
 
@@ -75,13 +75,13 @@ namespace pmacc
 
         /** get stream in which this event is recorded
          *
-         * @return native cupla stream
+         * @return native alpaka queue
          */
         AccStream getStream() const;
 
-        /** record event in a device stream
+        /** record event in a device queue
          *
-         * @param stream native cupla stream
+         * @param stream native alpaka queue
          */
         void recordEvent(AccStream const& stream);
     };