Intel oneAPI Data Analytics Library
Utilise CPUs and GPUs to implement High-Performance Data Science. By offering highly optimised algorithmic building blocks for all phases of data analytics (preprocessing, transformation, analysis, modelling, validation, and decision making) in batch, online, and distributed processing modes of computation, Big data analysis is accelerated in part by the Intel oneAPI Data Analytics Library (oneDAL).
To boost throughput and scalability, the library optimises algorithmic processing and data input. It has connections to well-known data sources including Spark and Hadoop, as well as C++ and Java APIs. The Intel Distribution for Python Programming Language includes Python wrappers for oneDAL.
Along with its typical features, oneDAL allows GPU utilisation for certain algorithms and offers enhancements to the standard C++ interface using the DPC++ SYCL API.
The library is very helpful when it comes to distributed computing. It offers a comprehensive collection of distributed algorithm building blocks that are not dependent on any communication layer. This enables users to use user-preferred communication methods to create distributed applications that are quick and scalable.
Create Powerful Data Science and Machine Learning Apps
Big data analysis is accelerated at every step of the machine learning pipeline by the Intel oneAPI Data Analytics Library (oneDAL). Implementations of traditional machine learning techniques that are tailored for Intel CPUs and GPUs are available through this open source, cross-platform package.
With this library, you are able to:
- Improve your forecasts more quickly.
- Optimise algorithmic computation and data intake at the same time.
- Create models for distributed, streaming, and offline use.
- To maximise overall application throughput, divide analytical workloads across edge devices and the cloud.
As part of the comprehensive set of Intel AI and machine learning development tools and resources, oneDAL optimises algorithms from well-known machine learning Python libraries like XGBoost and Intel Extension for Scikit-learn.
Features
Portability and Performance
- Each function is fine-tuned to the instruction set, vector width, core count, and memory architecture of each target CPU or GPU in order to guarantee optimal computation speed.
- Observe performance advantages for a variety of applications, including back-end servers and Internet of Things gateways.
- With integrated support for Python, SYCL, and C++, you can work in the language you know best while still getting the best performance out of your application.
Support for Algorithms
For C++, oneDAL contains methods for training and library prediction functions, math functions, and analytic routines, such as:
- Apriori for Mining Association Rules
- Variance-Covariance Matrices and Correlation
- Using Decision Forests for Regression and Classification
- Maximisation of Expectations Gaussian Mixture Model Utilisation (EM-GMM)
- Using Gradient Boosted Trees (GBT) for Regression and Classification
- Least Squares Alternation (ALS) for Cooperative Filtering
- Naïve Bayes Classifier with Multinomial
- Classifying Multiple Classes Using a One-Against-One Approach
- Optimiser for Limited-Memory BFGS (L-BFGS)
- Support for L1 and L2 regularisation in logistic regression
- Regression Linearity
- K-Means Grouping
- KNN, or K-Nearest Neighbour
- Using both linear and radial basis function (RBF) kernels, support vector machines (SVM)
- Analysis of Principal Components (PCA)
- Special Clustering of Applications with Noise based on Density (DBSCAN)
- Forest at Random
Multi-Language Support
With its support for several languages, such as Python, SYCL, and C++, developers may work in the language of their choice and yet get great performance.
Inclusion in Intel oneAPI Base Toolkit
The Intel oneAPI Base Toolkit, which includes oneDAL, is a set of tools and libraries for creating data-centric, high-performance applications.
Optimization for Machine Learning
The Intel Extension for Scikit-learn and XGBoost are two well-known Python machine learning libraries whose algorithms it optimises.
Data Analytics Pipeline Coverage
The preprocessing, transformation, analysis, modelling, validation, and decision-making phases of the data analytics pipeline are all covered by oneDAL. Use the Intel oneAPI Data Analytics Library to Get Started.
FAQ’s
Describe two ways oneDAL assists data analysis.
In the following ways, oneDAL supports data analysis:
Data Ingestion and Algorithm Optimisation OneDAL comprises methods that optimise algorithmic computation and data ingestion alike. The Python machine learning packages XGBoost and Intel Extension for Scikit-learn are among the algorithms that it optimises.
Whole-System Functionality The preprocessing, transformation, analysis, modelling, validation, and decision-making phases of the data analytics pipeline are all covered by the Intel oneAPI Data Analytics Library (oneDAL). Moreover, oneDAL offers a wide range of algorithms for training, library prediction, math, and analytic functions.
What are the languages that Intel OneDAL supports?
Python, SYCL, and C++ are supported by the Intel oneAPI Data Analytics Library (oneDAL). High performance in apps may be achieved while developers utilise the language they are most comfortable with this multi-language capability.
Which compilers work with oneDAL?
Compilers compatible with Intel oneAPI Data Analytics Library (oneDAL) include the following:
The Intel OneAPI Compiler for DPC++/C++
Intel Compiler for C++
Linux’s version of GNU Compiler Collection (GCC)
Compiler for Microsoft Visual C++ on Windows
On macOS, Clang
What toolkits include the oneDAL library?
Included in the Intel oneAPI Base Toolkit is the Intel oneAPI Data Analytics Library (oneDAL). The main tools and libraries in this toolkit are made to help developers create data-driven, high-performance applications for a variety of architectures.
Intel oneAPI Data Analytics Library Specifications
| Category | Components |
|---|---|
| Processors | – Intel Core processors – Intel Xeon processors |
| GPUs | – Intel Processor Graphics Gen9 and above – Iris graphics – Intel Data Center GPUs – Intel Arc A-series graphics |
| Operating Systems | – Linux – Windows – macOS (CPU only) |
| Compilers | – Intel oneAPI DPC++/C++ Compiler – Intel C++ Compiler – GNU Compiler Collection (GCC) on Linux – Microsoft Visual C++ Compiler on Windows – Clang on macOS |
| Languages | – SYCL (Requires Intel oneAPI Base Toolkit) – C++ – Python |
