CUDA 13.2 samples update (#432)

- Added Python samples for CUDA Python 1.0 release
- Renamed top-level `Samples` directory to `cpp` to accommodate Python samples.

cpp/3_CUDA_Features/binaryPartitionCG/.vscode/c_cpp_properties.json

@@ -0,0 +1,18 @@
+{
+    "configurations": [
+        {
+            "name": "Linux",
+            "includePath": [
+                "${workspaceFolder}/**",
+                "${workspaceFolder}/../../../Common"
+            ],
+            "defines": [],
+            "compilerPath": "/usr/local/cuda/bin/nvcc",
+            "cStandard": "gnu17",
+            "cppStandard": "gnu++14",
+            "intelliSenseMode": "linux-gcc-x64",
+            "configurationProvider": "ms-vscode.makefile-tools"
+        }
+    ],
+    "version": 4
+}

cpp/3_CUDA_Features/binaryPartitionCG/.vscode/extensions.json

@@ -0,0 +1,7 @@
+{
+    "recommendations": [
+        "nvidia.nsight-vscode-edition",
+        "ms-vscode.cpptools",
+        "ms-vscode.makefile-tools"
+    ]
+}

cpp/3_CUDA_Features/binaryPartitionCG/CMakeLists.txt

@@ -0,0 +1,34 @@
+cmake_minimum_required(VERSION 3.20)
+
+list(APPEND CMAKE_MODULE_PATH "${CMAKE_CURRENT_SOURCE_DIR}/../../../cmake/Modules")
+
+project(binaryPartitionCG LANGUAGES C CXX CUDA)
+
+find_package(CUDAToolkit REQUIRED)
+
+set(CMAKE_POSITION_INDEPENDENT_CODE ON)
+
+set(CMAKE_CUDA_ARCHITECTURES 75 80 86 87 89 90 100 110 120)
+set(CMAKE_CUDA_FLAGS "${CMAKE_CUDA_FLAGS} -Wno-deprecated-gpu-targets")
+if(ENABLE_CUDA_DEBUG)
+    set(CMAKE_CUDA_FLAGS "${CMAKE_CUDA_FLAGS} -G")        # enable cuda-gdb (may significantly affect performance on some targets)
+else()
+    set(CMAKE_CUDA_FLAGS "${CMAKE_CUDA_FLAGS} -lineinfo") # add line information to all builds for debug tools (exclusive to -G option)
+endif()
+
+# Include directories and libraries
+include_directories(../../../Common)
+
+# Source file
+# Add target for binaryPartitionCG
+add_executable(binaryPartitionCG binaryPartitionCG.cu)
+
+target_compile_options(binaryPartitionCG PRIVATE $<$<COMPILE_LANGUAGE:CUDA>:--extended-lambda>)
+
+target_compile_features(binaryPartitionCG PRIVATE cxx_std_17 cuda_std_17)
+
+set_target_properties(binaryPartitionCG PROPERTIES CUDA_SEPARABLE_COMPILATION ON)
+
+# Include installation configuration
+include(${CMAKE_CURRENT_SOURCE_DIR}/../../../cmake/InstallSamples.cmake)
+setup_samples_install()

cpp/3_CUDA_Features/binaryPartitionCG/README.md

@@ -0,0 +1,32 @@
+# binaryPartitionCG - Binary Partition Cooperative Groups
+
+## Description
+
+This sample is a simple code that illustrates binary partition cooperative groups and reduce within the thread block.
+
+## Key Concepts
+
+Cooperative Groups
+
+## Supported SM Architectures
+
+[SM 5.0 ](https://developer.nvidia.com/cuda-gpus)  [SM 5.2 ](https://developer.nvidia.com/cuda-gpus)  [SM 5.3 ](https://developer.nvidia.com/cuda-gpus)  [SM 6.0 ](https://developer.nvidia.com/cuda-gpus)  [SM 6.1 ](https://developer.nvidia.com/cuda-gpus)  [SM 7.0 ](https://developer.nvidia.com/cuda-gpus)  [SM 7.2 ](https://developer.nvidia.com/cuda-gpus)  [SM 7.5 ](https://developer.nvidia.com/cuda-gpus)  [SM 8.0 ](https://developer.nvidia.com/cuda-gpus)  [SM 8.6 ](https://developer.nvidia.com/cuda-gpus)  [SM 8.7 ](https://developer.nvidia.com/cuda-gpus)  [SM 8.9 ](https://developer.nvidia.com/cuda-gpus)  [SM 9.0 ](https://developer.nvidia.com/cuda-gpus)
+
+## Supported OSes
+
+Linux, Windows
+
+## Supported CPU Architecture
+
+x86_64, armv7l
+
+## CUDA APIs involved
+
+### [CUDA Runtime API](http://docs.nvidia.com/cuda/cuda-runtime-api/index.html)
+cudaStreamCreateWithFlags, cudaFree, cudaMallocHost, cudaFreeHost, cudaStreamSynchronize, cudaMalloc, cudaMemsetAsync, cudaMemcpyAsync, cudaOccupancyMaxPotentialBlockSize
+
+## Prerequisites
+
+Download and install the [CUDA Toolkit](https://developer.nvidia.com/cuda-downloads) for your corresponding platform.
+
+## References (for more details)

cpp/3_CUDA_Features/binaryPartitionCG/binaryPartitionCG.cu

@@ -0,0 +1,158 @@
+/* Copyright (c) 2022, NVIDIA CORPORATION. All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *  * Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ *  * Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ *  * Neither the name of NVIDIA CORPORATION nor the names of its
+ *    contributors may be used to endorse or promote products derived
+ *    from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY
+ * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE COPYRIGHT OWNER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
+ * OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+/*
+ * This sample illustrates basic usage of binary partition cooperative groups
+ * within the thread block tile when divergent path exists.
+ * 1.) Each thread loads a value from random array.
+ * 2.) then checks if it is odd or even.
+ * 3.) create binary partition group based on the above predicate
+ * 4.) we count the number of odd/even in the group based on size of the binary
+       groups
+ * 5.) write it global counter of odd.
+ * 6.) sum the values loaded by individual threads(using reduce) and write it to
+       global even & odd elements sum.
+ *
+ * **NOTE** :
+ *    binary_partition results in splitting warp into divergent thread groups
+ *    this is not good from performance perspective, but in cases where warp
+ *    divergence is inevitable one can use binary_partition group.
+*/
+
+#include <cooperative_groups.h>
+#include <cooperative_groups/reduce.h>
+#include <helper_cuda.h>
+#include <stdio.h>
+
+namespace cg = cooperative_groups;
+
+void initOddEvenArr(int *inputArr, unsigned int size)
+{
+    for (int i = 0; i < size; i++) {
+        inputArr[i] = rand() % 50;
+    }
+}
+
+/**
+ * CUDA kernel device code
+ *
+ * Creates cooperative groups and performs odd/even counting & summation.
+ */
+__global__ void oddEvenCountAndSumCG(int *inputArr, int *numOfOdds, int *sumOfOddAndEvens, unsigned int size)
+{
+    cg::thread_block          cta    = cg::this_thread_block();
+    cg::grid_group            grid   = cg::this_grid();
+    cg::thread_block_tile<32> tile32 = cg::tiled_partition<32>(cta);
+
+    for (int i = grid.thread_rank(); i < size; i += grid.size()) {
+        int  elem    = inputArr[i];
+        auto subTile = cg::binary_partition(tile32, elem & 1);
+        if (elem & 1) // Odd numbers group
+        {
+            int oddGroupSum = cg::reduce(subTile, elem, cg::plus<int>());
+
+            if (subTile.thread_rank() == 0) {
+                // Add number of odds present in this group of Odds.
+                atomicAdd(numOfOdds, subTile.size());
+
+                // Add local reduction of odds present in this group of Odds.
+                atomicAdd(&sumOfOddAndEvens[0], oddGroupSum);
+            }
+        }
+        else // Even numbers group
+        {
+            int evenGroupSum = cg::reduce(subTile, elem, cg::plus<int>());
+
+            if (subTile.thread_rank() == 0) {
+                // Add local reduction of even present in this group of evens.
+                atomicAdd(&sumOfOddAndEvens[1], evenGroupSum);
+            }
+        }
+        // reconverge warp so for next loop iteration we ensure convergence of
+        // above diverged threads to perform coalesced loads of inputArr.
+        cg::sync(tile32);
+    }
+}
+
+/**
+ * Host main routine
+ */
+int main(int argc, const char **argv)
+{
+    int          deviceId = findCudaDevice(argc, argv);
+    int         *h_inputArr, *d_inputArr;
+    int         *h_numOfOdds, *d_numOfOdds;
+    int         *h_sumOfOddEvenElems, *d_sumOfOddEvenElems;
+    unsigned int arrSize = 1024 * 100;
+
+    checkCudaErrors(cudaMallocHost(&h_inputArr, sizeof(int) * arrSize));
+    checkCudaErrors(cudaMallocHost(&h_numOfOdds, sizeof(int)));
+    checkCudaErrors(cudaMallocHost(&h_sumOfOddEvenElems, sizeof(int) * 2));
+    initOddEvenArr(h_inputArr, arrSize);
+
+    cudaStream_t stream;
+    checkCudaErrors(cudaStreamCreateWithFlags(&stream, cudaStreamNonBlocking));
+    checkCudaErrors(cudaMalloc(&d_inputArr, sizeof(int) * arrSize));
+    checkCudaErrors(cudaMalloc(&d_numOfOdds, sizeof(int)));
+    checkCudaErrors(cudaMalloc(&d_sumOfOddEvenElems, sizeof(int) * 2));
+
+    checkCudaErrors(cudaMemcpyAsync(d_inputArr, h_inputArr, sizeof(int) * arrSize, cudaMemcpyHostToDevice, stream));
+    checkCudaErrors(cudaMemsetAsync(d_numOfOdds, 0, sizeof(int), stream));
+    checkCudaErrors(cudaMemsetAsync(d_sumOfOddEvenElems, 0, 2 * sizeof(int), stream));
+
+    // Launch the kernel
+    int threadsPerBlock = 0;
+    int blocksPerGrid   = 0;
+    checkCudaErrors(cudaOccupancyMaxPotentialBlockSize(&blocksPerGrid, &threadsPerBlock, oddEvenCountAndSumCG, 0, 0));
+
+    printf("\nLaunching %d blocks with %d threads...\n\n", blocksPerGrid, threadsPerBlock);
+
+    oddEvenCountAndSumCG<<<blocksPerGrid, threadsPerBlock, 0, stream>>>(
+        d_inputArr, d_numOfOdds, d_sumOfOddEvenElems, arrSize);
+
+    checkCudaErrors(cudaMemcpyAsync(h_numOfOdds, d_numOfOdds, sizeof(int), cudaMemcpyDeviceToHost, stream));
+    checkCudaErrors(
+        cudaMemcpyAsync(h_sumOfOddEvenElems, d_sumOfOddEvenElems, 2 * sizeof(int), cudaMemcpyDeviceToHost, stream));
+    checkCudaErrors(cudaStreamSynchronize(stream));
+
+    printf("Array size = %d Num of Odds = %d Sum of Odds = %d Sum of Evens %d\n",
+           arrSize,
+           h_numOfOdds[0],
+           h_sumOfOddEvenElems[0],
+           h_sumOfOddEvenElems[1]);
+    printf("\n...Done.\n\n");
+
+    checkCudaErrors(cudaFreeHost(h_inputArr));
+    checkCudaErrors(cudaFreeHost(h_numOfOdds));
+    checkCudaErrors(cudaFreeHost(h_sumOfOddEvenElems));
+
+    checkCudaErrors(cudaFree(d_inputArr));
+    checkCudaErrors(cudaFree(d_numOfOdds));
+    checkCudaErrors(cudaFree(d_sumOfOddEvenElems));
+
+    return EXIT_SUCCESS;
+}