On September 15, 2021, the oneAPI Developer Summit ended with great success. The virtual conference focused on the oneAPI product suite, Data-Parallel C ++ (DPC ++) and AI for accelerated computing on XPU architectures – CPU, GPU, FPGA and other accelerators.

Organized by Intel Software in association with Analytics India Magazine, the event saw the participation of over 1,000 attendees and over 18 speakers, who presented their work on heterogeneous computing and artificial intelligence, for performances faster applications, increased productivity and increased innovation.

Register for our upcoming Masterclass>>

Some of the key industry leaders and academic speakers who attended the event included Andrew Richards, Henry Gabb, Gustavo Acosta, Himanshu Shekhar, Sameer Shende, Swarnendu Biswas, Soumyadeep Bhattacharya, Sunil Sherlekar, Vinayaka Ram Gururajan, Shrirang Karandikar , Sanjay Wandhekar, Sachin Kelkar, Kavita Aroor, Chris Kachris, Senganal Thirunavukkarasu and Amarpal Kapoor.

With over 18 conferences, the Developer Summit created dialogue and generated content in the form of a keynote, technical talks, blitzes, panel discussions and hands-on workshops. From science simulations to hardcore middleware implementations, the OneAPI DevSummit has addressed a variety of topics across the spectrum.

OneAPI overview

To kick off the developer event, Joe Curley, Senior Director of oneAPI Products, Solutions and Ecosystem, Intel, spoke about some of the challenges of accelerating computer programs and their plans to build a strong oneAPI user community. to make innovations evolve.

“OneAPI is our technology. It’s based on two premises – one you really need, an open standard and a community method for programming accelerators – where we have standards based on language and interfaces. Another is to have common libraries and common library interfaces, where we support non-Intel platforms, ”Curley said.

In addition, he said that Intel is committed to the open source ecosystem where developers can take matters into their own hands and innovate for their own inventions in addition to basic infrastructure. “Our goal here is to create a productive, efficient and reliable way to compete to write accelerated computing programs,” he added.

In a keynote presentation, Henry Gabb, Senior Principal Engineer – SATG, Intel, explored the oneAPI software abstraction for heterogeneous computing. Gabb said that Data-Parallel C ++ provides explicit parallel constructs and offload interfaces to support a wide variety of accelerators. “oneAPI also provides libraries for computation and data intensive areas such as deep learning, scientific computing, video analysis and media processing,” he added.

In addition, he said that he also defines the hardware abstraction or low-level interfaces that allow a language runtime system to effectively use a hardware accelerator for its vendors.

(Source: Henry Gabb, Intel | oneAPI DevSummit 2021)

Check out all the oneAPI source code, use cases and projects on GitHub and the official site.

Initiating the technical discussion session at oneAPI DevSummit 2021, Gustavo Acosta and Himanshu Shekhar from TCS discussed the “TCS IE Dynaport Smartpoint project”. Here, the duo spoke about the domain-specific challenges and how the oneAPI solution has helped them overcome them. On top of that, they also highlighted some of the use cases, design, and performance of the solution, while sharing some of the learnings and experiences they’ve come across while working with the oneAPI platform.

In Keynote 2 – “TAU Performance System & E4S” – Sameer Shende, Associate Research Professor and Director of the University of Oregon Performance Research Lab, described some of the top performing systems, highlighting scale the science software stack (E4S) and how it can be integrated with oneAPI. The session showcased some of TAU’s advancements to support oneAPI. It also explains how a user can use TAU to instrument code without changing the application binary.

In the following session on “Audio from gravitational wave simulations,” Andrew Pastrello, doctoral student at UNSW Sydney and nuclear engineer at the Australian Nuclear Science and Technology Organization, presented a simple interface for synthesizing audio from Gravitational waveforms produced by a binary black hole inspiral-merger-ringdown (IME) simulations, using a modified version of the EOBNRv2HM approximate (effective-one-body-digital-relativity version 2 higher modes) included in LALsuite.

Here are some of the main takeaways from this session:

• The implementation of kernels in DPC ++ is generally straightforward
• The biggest difficulties of this project were understanding and modifying existing code, which is a problem not specific to DPC ++.

In the “Unified Shared Memory (USM) Race Detection” blitz, Dr. Swarnendu Biswas, Assistant Professor in the Department of Computer Science and Engineering at IIT Kanpur, spoke about unintentional concurrency errors, while by discussing program analysis to detect programming errors. in CPUs and GPUs, and even some databases. This program analysis is intended to help improve the debugging toolchain for DPC ++ programs with oneAPI.

During the “lightning bolt” session on “Lattice Boltzmann Solvers in DPC ++”, Soumyadeep Bhattacharya, HPC Platforms and Simulations Architect at SankhyaSutra Labs, spoke about the application side of oneAPI, where they collaborated with many people to leverage their standard C ++ goals towards DPC ++ and target the next high-memory bandwidth architectures, which are becoming crucial for bandwidth-related scientific and engineering simulations.

In another interesting talk, Andrew Richards, CEO and Co-Founder of Codeplay, spoke on the topic “Building an Open, Secure and Accessible AI and Computing Ecosystem”. He said the world of AI and high performance computing (HPC) is dominated by closed, proprietary software models. To achieve high performance today, systems need accelerators with high levels of parallelism. “However, with programming models like CUDA, how do we open that up? How can we make these models safe enough to drive a car? And how do we get the industry to work with industry standards? Richards questioned.

Working on these challenges for years, Richard and his team at Codeplay came up with several solutions to help create a secure solution accessible to the AI ​​& HPC ecosystem with the help of OpenAI. During this session, he discussed some of the progress made, while also sharing the roadmap for the open source ecosystem.

One of the highlights of oneAPI DevSummit was the panel discussion on “Developer retraining for HPC and AI”. The panel discussion was moderated by Sachin Kelkar, Head of Asia Pacific – Developer Relations at Intel, alongside panelists from industry experts such as SankhyaSutra Labs CEO Dr Sunil Sherlekar; Vinayaka Ram Gururajan from TCS; Shrirang Karandikar from AlgoAsylum; and Sanjay Wandhekar of C-DAC.

See also

The panelist discussed some of the challenges and changes over the past few years in terms of developer expectations regarding learning new sets of tools, methodologies and the role of communities. They also brought a unique perspective of HPC research, AI solutions, systems integration areas, etc.

In the second half of the summit, Kavita Aroor, Head of Developer Marketing – APJ at Intel, highlighted some developer ecosystem programs such as Intel Software Innovator and the oneAPI Certified Instructor Program. In addition, her session explored DevCloud, oneAPI learning path modules, and DevMesh project submission. Kavita also played an exciting Jeopardy game where the developers had the chance to show off their oneAPI and DPC ++ knowledge and win exciting prizes.

View the DevMesh Contest Rules here.

At the latest blitz, Chris Kachris, co-founder and CEO of InAccel, provided insightful information and insight into “Accelerated ML on Intel® FPGAs Using oneAPI” and showed how users can run ML applications up to 40 times faster using oneAPI.

Last but not least, Dr Senganal Thirunavukkarasu, Computer Scientist (PhD) at North Carolina State University and Director of Data Science at Tiger Analytics, Singapore, spoke about “using the Intel® oneAPI libraries as oneDNN, oneMKL to optimize the Tiger Demand Forecasting Solution ”, which used the MXNET framework. In addition to that, he also spoke about some of the challenges encountered during this project and how they overcame them to successfully adopt Intel oneAPI libraries.

Finally, to successfully complete the oneAPI DevSummit 2021, during a hands-on session on “Intel® oneAPI Toolkits for HPC and AI applications”, Dr Amarpal Kapoor and Lakshmi Narasimhan from Intel explained in detail the nuances of openAI, while by revealing some of the case studies such as porting an HPC application to Intel GPUs and deploying an AI application on a traditional HPC cluster.

oneAPI in a nutshell

oneAPI is a cross-industry open source programming model that provides a common development experience across accelerator architectures. Additionally, the platform encourages collaboration on the oneAPI specification and compatible oneAPI implementations across the ecosystem.

Its specification extends existing developer programming models to enable a diverse set of hardware through language, a set of library APIs, and a low-level hardware interface to support inter-architecture programming. Its oneAPI specification is based on industry standards and provides an open, cross-platform developer stack to promote compatibility and enable productivity and innovation.

With your priority support active, you can get help through the online service center for your Intel® software development products.

• Submit questions, issues and other technical support issues
• Monitor previously submitted issues

https://software.intel.com/content/www/us/en/develop/support/priority-support.html

Connect with other developers and technical experts through this public forum: https://community.intel.com/t5/Intel-oneAPI-Toolkits/ct-p/oneapi

You can also visit the oneAPI Toolkits page, here.

You can now view the recordings here.

Join our Discord server. Be part of an engaging online community. Join here.

Subscribe to our newsletter

Receive the latest updates and relevant offers by sharing your email.