[gpuCI] Auto-merge branch-0.18 to branch-0.19 [skip ci]

CHANGELOG.md

@@ -1,13 +1,8 @@
-# cuML 0.18.0 (Date TBD)
+# 0.18.0
 
-## New Features
-
-## Improvements
-
-## Bug Fixes
-- PR #3279: Correct pure virtual declaration in manifold_inputs_t
+Please see https://github.com/rapidsai/cuml/releases/tag/branch-0.18-latest for the latest changes to this development branch.
 
-# cuML 0.17.0 (Date TBD)
+# cuML 0.17.0 (10 Dec 2020)
 
 ## New Features
 - PR #3164: Expose silhouette score in Python

README.md

@@ -85,7 +85,7 @@ repo](https://github.com/rapidsai/notebooks-contrib).
 ### Supported Algorithms
 | Category | Algorithm | Notes |
 | --- | --- | --- |
-| **Clustering** |  Density-Based Spatial Clustering of Applications with Noise (DBSCAN) | |
+| **Clustering** |  Density-Based Spatial Clustering of Applications with Noise (DBSCAN) | Multi-node multi-GPU via Dask |
 |  | K-Means | Multi-node multi-GPU via Dask |
 | **Dimensionality Reduction** | Principal Components Analysis (PCA) | Multi-node multi-GPU via Dask|
 | | Incremental PCA | |
@@ -96,6 +96,7 @@ repo](https://github.com/rapidsai/notebooks-contrib).
 | **Linear Models for Regression or Classification** | Linear Regression (OLS) | Multi-node multi-GPU via Dask |
 | | Linear Regression with Lasso or Ridge Regularization | Multi-node multi-GPU via Dask |
 | | ElasticNet Regression | |
+| | LARS Regression | (experimental) |
 | | Logistic Regression | Multi-node multi-GPU via Dask-GLM [demo](https://github.com/daxiongshu/rapids-demos) |
 | | Naive Bayes | Multi-node multi-GPU via Dask |
 | | Stochastic Gradient Descent (SGD), Coordinate Descent (CD), and Quasi-Newton (QN) (including L-BFGS and OWL-QN) solvers for linear models  | |

ci/checks/copyright.py

@@ -38,6 +38,9 @@
     re.compile(r"[.]flake8[.]cython$"),
     re.compile(r"meta[.]yaml$")
 ]
+ExemptFiles = [
+    re.compile(r"_thirdparty")
+]
 
 # this will break starting at year 10000, which is probably OK :)
 CheckSimple = re.compile(
@@ -53,6 +56,9 @@ def checkThisFile(f):
         return False
     if gitutils and gitutils.isFileEmpty(f):
         return False
+    for exempt in ExemptFiles:
+        if exempt.search(f):
+            return False
     for checker in FilesToCheck:
         if checker.search(f):
             return True

conda/recipes/cuml/meta.yaml

@@ -1,4 +1,4 @@
-# Copyright (c) 2021, NVIDIA CORPORATION.
+# Copyright (c) 2018-2021, NVIDIA CORPORATION.
 
 # Usage:
 #   conda build . -c defaults -c conda-forge -c numba -c rapidsai -c pytorch
@@ -40,8 +40,8 @@ requirements:
     - dask-cudf {{ minor_version }}
     - libcuml={{ version }}
     - libcumlprims {{ minor_version }}
+    - cupy>=7.8.0,<9.0.0a0
     - treelite=1.0.0
-    - cupy>7.1.0,<9.0.0a0
     - nccl>=2.5
     - ucx-py {{ minor_version }}
     - dask>=2.12.0

conda/recipes/libcuml/meta.yaml

@@ -1,4 +1,4 @@
-# Copyright (c) 2021, NVIDIA CORPORATION.
+# Copyright (c) 2018-2021, NVIDIA CORPORATION.
 
 # Usage:
 #   conda build . -c defaults -c conda-forge -c nvidia -c rapidsai -c pytorch

cpp/CMakeLists.txt

@@ -1,5 +1,5 @@
 #=============================================================================
-# Copyright (c) 2018-2020, NVIDIA CORPORATION.
+# Copyright (c) 2018-2021, NVIDIA CORPORATION.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -206,6 +206,9 @@ set(CMAKE_MODULE_PATH ${CMAKE_CURRENT_BINARY_DIR}/cmake)
 # https://github.com/rapidsai/cuml/issues/2194
 find_package(CUDA 10.0 REQUIRED)
 
+# Set the -isystem flag to not have an '=' sign to allow parsing of compile_commands.json
+set(CMAKE_INCLUDE_SYSTEM_FLAG_CUDA "-isystem ")
+
 if (NOT DISABLE_OPENMP OR NOT ${DISABLE_OPENMP})
   find_package(OpenMP)
   if(OPENMP_FOUND)
@@ -324,16 +327,18 @@ include(cmake/Dependencies.cmake)
 
 set(CUML_INCLUDE_DIRECTORIES
   ${CUML_INCLUDE_DIR}
-  ${CMAKE_CURRENT_SOURCE_DIR}/src
-  ${CMAKE_CURRENT_SOURCE_DIR}/src_prims
-  ${CMAKE_CURRENT_SOURCE_DIR}/test/prims
   ${CMAKE_CUDA_TOOLKIT_INCLUDE_DIRECTORIES}
   ${CUTLASS_DIR}/src/cutlass
   ${SPDLOG_DIR}/src/spdlog/include
   ${FAISS_INCLUDE_DIRS}
   ${RAFT_DIR}/cpp/include
   ${RMM_INCLUDE_DIRS})
 
+# These directories will be used by dependent targets
+set(CUML_SRC_DIR ${CMAKE_CURRENT_SOURCE_DIR}/src)
+set(CUML_SRC_PRIMS_DIR ${CMAKE_CURRENT_SOURCE_DIR}/src_prims)
+set(CUML_TEST_PRIMS_DIR ${CMAKE_CURRENT_SOURCE_DIR}/test/prims)
+
 if(NOT CUB_IS_PART_OF_CTK)
   list(APPEND CUML_INCLUDE_DIRECTORIES ${CUB_DIR}/src/cub)
 endif(NOT CUB_IS_PART_OF_CTK)
@@ -386,8 +391,6 @@ if(BUILD_CUML_CPP_LIBRARY)
   add_library(${CUML_CPP_TARGET} SHARED
     src/arima/batched_arima.cu
     src/arima/batched_kalman.cu
-    src/common/cumlHandle.cpp
-    src/common/cuml_api.cpp
     src/common/logger.cpp
     src/common/nvtx.cu
     src/datasets/make_arima.cu
@@ -400,6 +403,8 @@ if(BUILD_CUML_CPP_LIBRARY)
     src/fil/fil.cu
     src/fil/infer.cu
     src/glm/glm.cu
+    src/genetic/genetic.cu
+    src/genetic/node.cu
     src/holtwinters/holtwinters.cu
     src/kmeans/kmeans.cu
     src/knn/knn.cu
@@ -442,7 +447,6 @@ if(BUILD_CUML_CPP_LIBRARY)
         src/glm/ridge_mg.cu
         src/kmeans/kmeans_mg.cu
         src/knn/knn_mg.cu
-        src/knn/knn_opg_common.cu
         src/knn/knn_classify_mg.cu
         src/knn/knn_regress_mg.cu
         src/pca/pca_mg.cu
@@ -461,15 +465,24 @@ if(BUILD_CUML_CPP_LIBRARY)
     link_directories(${CUDA_TOOLKIT_ROOT_DIR}/lib64)
   endif(NVTX)
 
+  target_compile_definitions(${CUML_CPP_TARGET}
+    PRIVATE
+      CUML_CPP_API)
+
   target_include_directories(${CUML_CPP_TARGET}
-    PRIVATE ${CUML_INCLUDE_DIRECTORIES})
+    PUBLIC
+      ${CUML_INCLUDE_DIRECTORIES}
+    PRIVATE
+      ${CUML_SRC_DIR}
+      ${CUML_SRC_PRIMS_DIR})
 
   target_link_libraries(${CUML_CPP_TARGET}
     PUBLIC
       ${CUML_PUBLIC_LINK_LIBRARIES}
     PRIVATE
       ${CUML_PRIVATE_LINK_LIBRARIES}
   )
+
   # If we export the libdmlc symbols, they can lead to weird crashes with other
   # libraries that use libdmlc. This just hides the symbols internally.
   target_link_options(${CUML_CPP_TARGET} PRIVATE "-Wl,--exclude-libs,libdmlc.a")
@@ -485,14 +498,23 @@ if(BUILD_CUML_C_LIBRARY)
   set(CUML_C_TARGET "cuml")
 
   add_library(${CUML_C_TARGET} SHARED
+    src/common/cumlHandle.cpp
     src/common/cuml_api.cpp
     src/dbscan/dbscan_api.cpp
     src/glm/glm_api.cpp
     src/holtwinters/holtwinters_api.cpp
-    src/svm/svm_api.cpp)
+    src/knn/knn_api.cpp
+    src/svm/svm_api.cpp
+  )
+
+  target_compile_definitions(${CUML_C_TARGET}
+    PRIVATE
+      CUML_C_API)
 
   target_include_directories(${CUML_C_TARGET}
-    PRIVATE ${CUML_INCLUDE_DIRECTORIES})
+    PRIVATE 
+      ${CUML_SRC_DIR}
+  )
 
   target_link_libraries(${CUML_C_TARGET} PUBLIC ${CUML_CPP_TARGET})
 endif(BUILD_CUML_C_LIBRARY)

cpp/bench/CMakeLists.txt

@@ -1,5 +1,5 @@
 #=============================================================================
-# Copyright (c) 2019-2020, NVIDIA CORPORATION.
+# Copyright (c) 2019-2021, NVIDIA CORPORATION.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -41,8 +41,10 @@ if(BUILD_CUML_BENCH)
     benchmarklib)
 
   target_include_directories(sg_benchmark
-    PRIVATE ${GBENCH_DIR}/include
-    ${CUML_INCLUDE_DIRECTORIES})
+    PRIVATE 
+      ${GBENCH_DIR}/include
+      ${CUML_SRC_PRIMS_DIR}
+  )
 endif(BUILD_CUML_BENCH)
 
 ##############################################################################
@@ -64,17 +66,23 @@ if(BUILD_CUML_PRIMS_BENCH)
     prims/matrix_vector_op.cu
     prims/permute.cu
     prims/reduce.cu
-    prims/rng.cu
-    ../src/common/logger.cpp)  # because prims is header only!
+    prims/rng.cu)
 
   if(NOT CUB_IS_PART_OF_CTK)
     add_dependencies(prims_benchmark cub)
   endif(NOT CUB_IS_PART_OF_CTK)
   add_dependencies(prims_benchmark spdlog)
 
-  target_link_libraries(prims_benchmark ${CUDA_cublas_LIBRARY} benchmarklib)
+  target_link_libraries(prims_benchmark 
+    PUBLIC
+      ${CUDA_cublas_LIBRARY}
+      benchmarklib
+      ${CUML_CPP_TARGET}
+  )
 
   target_include_directories(prims_benchmark
-    PRIVATE ${GBENCH_DIR}/include
-    ${CUML_INCLUDE_DIRECTORIES})
+    PRIVATE 
+      ${GBENCH_DIR}/include
+      ${CUML_SRC_PRIMS_DIR}
+  )
 endif(BUILD_CUML_PRIMS_BENCH)

cpp/bench/sg/dataset.cuh

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2019-2020, NVIDIA CORPORATION.
+ * Copyright (c) 2019-2021, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -18,7 +18,6 @@
 
 #include <raft/cudart_utils.h>
 #include <raft/linalg/transpose.h>
-#include <common/cumlHandle.hpp>
 #include <cuml/cuml.hpp>
 #include <cuml/datasets/make_blobs.hpp>
 #include <fstream>

cpp/bench/sg/dataset_ts.cuh

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2020, NVIDIA CORPORATION.
+ * Copyright (c) 2020-2021, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -16,7 +16,6 @@
 
 #pragma once
 
-#include <common/cumlHandle.hpp>
 #include <cuml/cuml.hpp>
 #include <raft/cuda_utils.cuh>
 

cpp/bench/sg/dbscan.cu

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2019-2020, NVIDIA CORPORATION.
+ * Copyright (c) 2019-2021, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -15,7 +15,7 @@
  */
 
 #include <cuml/cluster/dbscan.hpp>
-#include <cuml/cuml.hpp>
+
 #include <utility>
 #include "benchmark.cuh"
 
@@ -51,10 +51,10 @@ class Dbscan : public BlobsFixture<D, int> {
       state.SkipWithError("Dbscan only supports row-major inputs");
     }
     this->loopOnState(state, [this, &state]() {
-      dbscanFit(*this->handle, this->data.X, this->params.nrows,
-                this->params.ncols, D(dParams.eps), dParams.min_pts,
-                this->data.y, this->core_sample_indices,
-                dParams.max_bytes_per_batch);
+      ML::Dbscan::fit(*this->handle, this->data.X, this->params.nrows,
+                      this->params.ncols, D(dParams.eps), dParams.min_pts,
+                      this->data.y, this->core_sample_indices,
+                      dParams.max_bytes_per_batch);
       state.SetItemsProcessed(this->params.nrows * this->params.ncols);
     });
   }