Visible to Intel only — GUID: GUID-DEE81E0E-C7A3-4BB6-BF4C-3C986BD68306
basic_statistics_dense_batch.cpp
basic_statistics_dense_online.cpp
column_accessor_homogen.cpp
cor_dense_batch.cpp
cor_dense_online.cpp
cov_dense_batch.cpp
cov_dense_online.cpp
csr_accessor.cpp
dbscan_brute_force_batch.cpp
df_cls_hist_batch.cpp
df_cls_hist_batch_random.cpp
df_cls_traverse_model.cpp
df_reg_hist_batch.cpp
df_reg_hist_batch_random.cpp
df_reg_traverse_model.cpp
kmeans_init_dense.cpp
kmeans_lloyd_dense_batch.cpp
knn_cls_brute_force_dense_batch.cpp
knn_reg_brute_force_dense_batch.cpp
knn_search_brute_force_dense_batch.cpp
linear_kernel_dense_batch.cpp
linear_regression_dense_batch.cpp
linear_regression_dense_online.cpp
logistic_regression_dense_batch.cpp
pca_cor_dense_batch.cpp
pca_cor_dense_online.cpp
pca_precomputed_cor_dense_batch.cpp
pca_precomputed_cov_dense_batch.cpp
rbf_kernel_dense_batch.cpp
svm_two_class_thunder_dense_batch.cpp
basic_statistics_dense_batch.cpp
basic_statistics_dense_online.cpp
column_accessor_homogen.cpp
connected_components_batch.cpp
cor_dense_batch.cpp
cor_dense_online.cpp
cov_dense_batch.cpp
cov_dense_online.cpp
csr_accessor.cpp
dbscan_brute_force_batch.cpp
df_cls_dense_batch.cpp
df_reg_dense_batch.cpp
directed_graph.cpp
graph_service_functions.cpp
jaccard_batch.cpp
jaccard_batch_app.cpp
kmeans_init_dense.cpp
kmeans_lloyd_dense_batch.cpp
knn_cls_brute_force_dense_batch.cpp
knn_cls_kd_tree_dense_batch.cpp
knn_search_brute_force_dense_batch.cpp
linear_kernel_dense_batch.cpp
linear_regression_dense_batch.cpp
linear_regression_dense_online.cpp
logloss_dense_batch.cpp
louvain_batch.cpp
pca_dense_batch.cpp
pca_dense_online.cpp
pca_precomputed_dense_batch.cpp
polynomial_kernel_dense_batch.cpp
rbf_kernel_dense_batch.cpp
shortest_paths_batch.cpp
sigmoid_kernel_dense_batch.cpp
subgraph_isomorphism_batch.cpp
svm_multi_class_thunder_csr_batch.cpp
svm_multi_class_thunder_dense_batch.cpp
svm_nu_cls_thunder_csr_batch.cpp
svm_nu_cls_thunder_dense_batch.cpp
svm_nu_reg_thunder_csr_batch.cpp
svm_nu_reg_thunder_dense_batch.cpp
svm_reg_thunder_csr_batch.cpp
svm_reg_thunder_dense_batch.cpp
svm_two_class_smo_csr_batch.cpp
svm_two_class_smo_dense_batch.cpp
svm_two_class_thunder_csr_batch.cpp
svm_two_class_thunder_dense_batch.cpp
triangle_counting_batch.cpp
K-Means Clustering
Glossary
Correlation and Variance-Covariance Matrices
Principal Component Analysis
Principal Components Analysis Transform
Singular Value Decomposition
Association Rules
Kernel Functions
Expectation-Maximization
Cholesky Decomposition
QR Decomposition
Outlier Detection
Distance Matrix
Distributions
Engines
Moments of Low Order
Quantile
Quality Metrics
Sorting
Normalization
Optimization Solvers
Decision Forest
Decision Trees
Gradient Boosted Trees
Stump
Linear and Ridge Regressions
LASSO and Elastic Net Regressions
k-Nearest Neighbors (kNN) Classifier
Implicit Alternating Least Squares
Logistic Regression
Naïve Bayes Classifier
Support Vector Machine Classifier
Multi-class Classifier
Boosting
Visible to Intel only — GUID: GUID-DEE81E0E-C7A3-4BB6-BF4C-3C986BD68306
Array
The array is a simple concept over the data in oneDAL. It represents a storage that:
Holds the data allocated inside it or references to the external data. The data are organized as one homogeneous and contiguous memory block.
Contains information about the memory block’s size.
Supports both immutable and mutable data.
Provides an ability to change the data state from immutable to mutable one.
Holds ownership information on the data (see the data ownership requirements section).
Ownership information on the data can be shared between several arrays. It is possible to create a new array from another one without any data copies.
Usage Example
The following listing provides a brief introduction to the array API and an example of basic usage scenario:
#include <sycl/sycl.hpp>
#include <iostream>
#include <string>
#include "oneapi/dal/array.hpp"
using namespace oneapi;
void print_property(const std::string& description, const auto& property) {
std::cout << description << ": " << property << std::endl;
}
int main() {
sycl::queue queue { sycl::default_selector() };
constexpr std::int64_t data_count = 4;
const float data[] = { 1.0f, 2.0f, 3.0f, 4.0f };
// Creating an array from immutable user-defined memory
auto arr_data = dal::array<float>::wrap(data, data_count);
// Creating an array from internally allocated memory filled by ones
auto arr_ones = dal::array<float>::full(queue, data_count, 1.0f);
print_property("Is arr_data mutable", arr_data.has_mutable_data()); // false
print_property("Is arr_ones mutable", arr_ones.has_mutable_data()); // true
// Creating new array from arr_data without data copy - they share ownership information.
dal::array<float> arr_mdata = arr_data;
print_property("arr_mdata elements count", arr_mdata.get_count()); // equal to data_count
print_property("Is arr_mdata mutable", arr_mdata.has_mutable_data()); // false
/// Copying data inside arr_mdata to new mutable memory block.
/// arr_data still refers to the original data pointer.
arr_mdata.need_mutable_data(queue);
print_property("Is arr_data mutable", arr_data.has_mutable_data()); // false
print_property("Is arr_mdata mutable", arr_mdata.has_mutable_data()); // true
queue.submit([&](sycl::handler& cgh){
auto mdata = arr_mdata.get_mutable_data();
auto cones = arr_ones.get_data();
cgh.parallel_for<class array_addition>(sycl::range<1>(data_count), [=](sycl::id<1> idx) {
mdata[idx[0]] += cones[idx[0]];
});
}).wait();
std::cout << "arr_mdata values: ";
for(std::int64_t i = 0; i < arr_mdata.get_count(); i++) {
std::cout << arr_mdata[i] << ", ";
}
std::cout << std::endl;
return 0;
}
Data ownership requirements
The array supports the following requirements on the internal data management:
An array owns two properties representing raw pointers to the data:
data for a pointer to immutable data block
mutable_data for a pointer to mutable data block (see the api array)
If an array owns mutable data, both properties point to the same memory block.
If an array owns immutable data, mutable_data is nullptr.
An array stores the number of elements in the block it owns and updates the count property when a new memory block is assigned to the array.
An array stores a pointer to the ownership structure of the data:
The reference count indicating how many array objects refer to the same memory block.
The deleter object used to free the memory block when reference count is zero.
An array creates the ownership structure for a new memory block not associated with such structure.
An array decrements the number of references to the memory block when the array goes out of the scope. If the number of references is zero, the array calls the deleter on this memory block and free the ownership structure.
An array stores the pointer to the ownership structure created by another array when they share the data. An array increments the reference count for it to be equal to the number of array objects sharing the same data.
Programming interface
Refer to API Reference: Array.