Ready-to-Use Software Ported from CUDA* to SYCL*.
The organizations, research institutions and universities at the heart of the oneAPI Academic Centers of Excellence do not only help us drive innovation forward. They are leading the adoption of SYCL. SYCL is a royalty-free, cross-platform abstraction layer that enables code for heterogeneous and offload processors to be written using modern standard C++.
In doing so these institutions help to facilitate a world, where software developers are not tied to a different programming paradigm for every hardware configuration or workload host and offload target compute acceleration combination they choose. Open source software projects become truly portable.
Over the next few months, in this new series of blogs we will highlight the projects that have been migrated to support a vendor agnostic SYCL based adaptation of their software codebase.
Every few weeks we will introduce you to the work of another member of this prestigious group.
We will delve into the details of the projects they are working on, highlighting their contributions to the evolution of heterogeneous cross-architecture compute in high performance computing.
Rich Innovation Ecosystem
SYCL is not just a C++ abstraction layer to enable heterogeneous programming. It is evolving into the cornerstone of a vision to free workloads, that want to take advantage of the strength of different types of compute engines, from vendor-lock. The oneAPI specification builds on SYCL and uses it for an initiative that encourages collaboration on cross-architecture implementations across the ecosystem. Its backend architecture is fully open standards based and open source. This allows the integration of existing OpenCL based device drivers, creation of new Level Zero based device drivers or custom device drivers. Targeted compute accelerators can in principle be anything from CPUs, GPUs, FPGAs to ASICs or VPUs.
This open architecture is what makes SYCL and oneAPI so interesting for organizations and research institutions that need their code base to be running at their best, regardless of which HPC platform they will be deployed on next.
Even analytics workloads deployed in the field or on the edge can benefit from this flexible and portable approach.
For that very reason several members of the oneAPI Academic Centers of Excellence have started migrating their data analytics and scientific workloads from CUDA to SYCL, offering SYCL based variants of their projects as binaries or in source code.
This is the beginning of an ever-growing innovation ecosystem for accelerated heterogeneous compute.
Awesome List of oneAPI Projects
Example projects as well as some of these CoE projects are available at the oneAPI Community’s Awesome List of oneAPI Projects.
Projects available cover a range of example projects across many computational categories:
- AI - Computer Vision
- AI - Data Science
- AI - Machine Learning
- AI - Natural Language Processing
- AI - Frameworks and Toolkits
- Autonomous Systems
- Data Visualization and Rendering
- Energy
- Gaming
- Manufacturing
- Mathematics and Science
- Tools & Development
In our blog series however we want to specifically focus on projects inside the oneAPI Academic Centers of Excellence, that are available in SYCL implementation variants.
Some Key Projects Available with SYCL Support
Below are some of projects inside the Awesome List of oneAPI Projects or worked on within oneAPI Academic Centers of Excellence. We will dive deeper into each one of them in upcoming write-ups:
- Amber*: A high-performance molecular dynamics (MD) collection of programs used by thousands of scientists in academia, national labs, and industry for molecular dynamics simulations, particularly on biomolecules for computational drug discovery and related research. Andreas Goetz and his team with the San Diego Supercomputer Center (SDSC) at the University of California San Diego (UCSD) are leading the effort of enabling Amber with SYCL.
- Autodock*: A molecular modeling simulation software used in medical and pharmaceutical research, available under the GNU General Public License. It is especially effective for protein-ligand docking. Andreas Koch and Leonardo Solis-Vasquez and their teams at the Technical University of Darmstadt made it available powered by and optimized with oneAPI and SYCL.
- GinkgoOneAPI*: A high-performance linear algebra library for manycore systems, with a focus on solution of sparse linear systems. Professor Hartwig Anzt and his team at the University of Tennessee at Knoxville are implementing the port of the open source Gingko software package to the oneAPI backend.
- GROMACS*: A free open source software suite for high-performance molecular dynamics and output analysis. mainly designed for simulations of proteins, lipids, and nucleic acids. It was originally developed in the Biophysical Chemistry Department of University of Groningen and is now maintained by contributors in universities and research centers worldwide. Erik Lindahl and his team at the Stockholm University KTH Royal Institute of Technology have ported GROMACS to use SYCL and oneAPI.
- NAMD*: Nanoscale Molecular Dynamics (NAMD) is a software solution for molecular dynamics simulation. It has been developed by the collaboration of the Theoretical and Computational Biophysics Group (TCB) and the Parallel Programming Laboratory (PPL) at the University of Illinois at Urbana–Champaign. Emad Tajkhorshid and David Hardy and their teams are making it available with SYCL and oneAPI support.
Join Us
Become part of the effort to make high-performance cross-architecture compute transparent, portable, and flexible. Include SYCL as the accelerator and GPU offload solution in your code path. Adopt oneAPI as the means to implementations free from vendor-lock. Find out more about Intel's contribution to oneAPI.
Make it part of your science data analytics and simulation solution.
Please stay tuned as we get into more detail about the work that these researchers and developers are doing and how our joint efforts help to offer more choice for compute intensive workloads.
Get the Software
Additional Resources
- Academic oneAPI Centers of Excellence
- Essentials of SYCL*
- CUDA* to SYCL* Catalog
- Migrate from CUDA* to C++ with SYCL*
- Workflow for a CUDA* to SYCL* Migration
- An Awesome List of oneAPI projects
- SYCLomatic
Notices and Disclaimers
Performance varies by use, configuration and other factors. Learn more at www.Intel.com/PerformanceIndex. Results may vary.
Performance results are based on testing as of dates shown in configurations and may not reflect all publicly available updates. No product or component can be absolutely secure.
Your costs and results may vary.
Intel technologies may require enabled hardware, software or service activation.
Intel does not control or audit third-party data. You should consult other sources to evaluate accuracy.
© Intel Corporation. Intel, the Intel logo, and other Intel marks are trademarks of Intel Corporation or its subsidiaries.
*Other names and brands may be claimed as the property of others.
Rob enables developers to streamline programming efforts across multiarchitecture compute devices for high performance applications taking advantage of Intel's family of development tools. He has extensive 20+ years of experience in technical consulting, software architecture and platform engineering working in IoT, edge, embedded software and hardware developer enabling.