A RetroSearch Logo

Home - News ( United States | United Kingdom | Italy | Germany ) - Football scores

Search Query:

Showing content from https://github.com/libocca/occa below:

libocca/occa: Portable and vendor neutral framework for parallel programming on heterogeneous platforms.

 

Performance, Portability, Transparency

OCCA is an open source, portable, and vendor neutral framework for parallel programming on heterogeneous platforms. The OCCA API provides unified models for heterogeneous programming concepts—such as a device, memory, or kernel—while the OCCA Kernel Language (OKL) enables the creation of portable device kernels using a directive-based extension to the C-language.

Mission critical computational science and engineering applications from the public and private sectors rely on OCCA. Notable users include the U.S. Department of Energy and Shell.

Key Features

OCCA uses the CMake build system. Checkout the installation guide for a comprehensive overview of all build settings and instructions for building on Windows or Mac OS.

For convenience, the shell script configure-cmake.sh has been provided to drive the CMake build. Compilers, flags, and other build parameters can be adjusted there. By default, this script uses ./build and ./install for the build and install directories.

The following demonstrates a typical sequence of shell commands to build, test, and install occa:

$ ./configure-cmake.sh
$ cmake --build build --parallel <number-of-threads>
$ ctest --test-dir build --output-on-failure
$ cmake --install build --prefix install

If dependencies are installed in a non-standard location, set the corresponding environment variable to this path.

During installation, the Env Modules file <install-prefix>/modulefiles/occa is generated. When this module is loaded, paths to the installed bin, lib, and include directories are appended to environment variables such as PATH and LD_LIBRARY_PATH.

Building an OCCA application

For convenience, OCCA provides CMake package files which are configured during installation. These package files define an imported target, OCCA::libocca, and look for all required dependencies.

For example, the CMakeLists.txt of downstream projects using OCCA would include

find_package(OCCA REQUIRED)

add_executable(downstream-app ...)
target_link_libraries(downstream-app PRIVATE OCCA::libocca)

add_library(downstream-lib ...)
target_link_libraries(downstream-lib PUBLIC OCCA::libocca)

The OCCA command-line interface can be found in <install-prefix>/bin/occa. This tool can be used to query information about hardware and the configuration of OCCA on a given platform.

For example, calling occa info will available OCCA backends and related hardware specs, while occa env display the values of OCCA related environment variables. To see the list of all available options, call occa --help.

$ occa info
========+======================+=================================
 CPU(s) | Processor Name       | AMD EPYC 7532 32-Core Processor 
        | Memory               | 251.6 GB                        
        | Clock Frequency      | 2.4 MHz                         
        | SIMD Instruction Set | SSE2                            
        | SIMD Width           | 128 bits                        
        | L1d Cache Size       |   1 MB                          
        | L1i Cache Size       |   1 MB                          
        | L2 Cache Size        |  16 MB                          
        | L3 Cache Size        | 256 MB                          
========+======================+=================================
 OpenCL | Platform 0           | NVIDIA CUDA                     
        |----------------------+---------------------------------
        | Device 0             | NVIDIA A100-PCIE-40GB           
        | Device Type          | gpu                             
        | Compute Cores        | 108                             
        | Global Memory        | 39.40 GB                        
========+======================+=================================
 CUDA   | Device Name          | NVIDIA A100-PCIE-40GB           
        | Device ID            | 0                               
        | Memory               | 39.40 GB                        
========+======================+=================================

Need help? Checkout the repository wiki or ask a question in the Q&A discussions category.

To provide feedback, start a conversation in the general or ideas discussion categories.

This work was supported in part by

OCCA is available under a MIT license


RetroSearch is an open source project built by @garambo | Open a GitHub Issue

Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo

HTML: 3.2 | Encoding: UTF-8 | Version: 0.7.4