Code Together: Recent Episodes
Intel Corporation
Code Together is an interview series that explores challenges at the forefront of cross-architecture development. We’ll talk with those at Intel and across the industry who are forging a path on this often treacherous journey through an increasingly diverse, data-centric world. Join the conversation at oneapi.com.
Transformer models are the powerful neural networks that have become the standard for delivering advanced performance behind these innovations. But there is a challenge: Training these deep learning models at scale and doing inference on them requires a large amount of computing power. This can make the process time-consuming, complex, and costly.
Today we will talk about all kinds of issues around accessible, production level AI solutions. We also talk about ethical questions around AI usage and why open, democratized AI solutions are important.
Learn more:
Hugging Face
https://huggingface.co
Hugging Face Hub
https://huggingface.co/models
Fast Inference on Large Language Models: BLOOMZ on Habana Gaudi2 Accelerator
https://huggingface.co/blog/habana-gaudi-2-bloom
Accelerating Stable Diffusion Inference on Intel CPUs
https://huggingface.co/blog/stable-diffusion-inference-intel
Transformer Performance with Intel & Hugging Face Webinar
https://www.intel.com/content/www/us/en/developer/videos/optimize-end-to-end-transformer-model-performance.html#gs.pomt5k
Intel Explainable AI Tools
https://github.com/IntelAI/intel-xai-tools
Intel Distribution of OpenVINO Toolkit
https://www.intel.com/content/www/us/en/developer/tools/openvino-toolkit/overview.html
Intel AI Analytics Toolkit (AI Kit)
https://www.intel.com/content/www/us/en/developer/tools/oneapi/ai-analytics-toolkit.html
Guests:
Julien Simon – Chief Evangelist @ Hugging Face
Ke Ding – Principal Engineer @ Intel
As the world moves towards more specialized hardware like GPUs, Google TPUs, Habana Gaudi (HPU), FPGAs, etc, enabling developers to run their code on the most efficient hardware can be challenging. Intel oneAPI and xPUs are at the vanguard of this shift and Digital Cortex is leading the way forward with its comprehensive set of products built on these emerging technologies. By providing unprecedented access to compute across varying architectures and platforms, Digital Cortex is helping developers unlock new levels of performance while simplifying development processes.
We talk to Digital Cortex CEO Charlie Wardell to learn how Digital Cortex is building on top of the oneAPI vision to bring a marketplace of compute kernels to their platform that can run anywhere. Charlie also shares his thoughts on entrepreneurship and what made him make the leap from working in a company to founding a company.
Learn more:
Digital Cortex
digital-cortex.io
oneAPI
oneapi.io
Intel oneAPI toolkits
intel.com/content/www/us/en/developer/tools/oneapi/overview.html#gs.rahkwl
Guest:
Charlie Wardell – CEO of Digital Cortex
Over the last couple of decades, the world of computer graphics has been growing at an amazing rate. From animated movies to video games, consumers have reaped the benefits of these technological advances. One of the latest algorithmic advances in rendering is path guiding. Learn about path guiding and how you can take advantage of this cool technology using Chaos V-Ray and the Intel Path Guiding Library (OpenPGL)
Learn more:
Chaos V-Ray:
chaos.com
Intel Open Path Guiding Library (OpenPGL) Github
github.com/OpenPathGuidingLibrary/openpgl
Paper: A 5D Tree to Reduce the Variance of Monte Carlo Ray Tracing
Author: Eric Lafortune
researchgate.net/publication/2574746_A_5D_Tree_to_Reduce_the_Variance_of_Monte_Carlo_Ray_Tracing
Paper: Importance driven path tracing using the photon map.
Author: Henrik Jensen
graphics.stanford.edu/~henrik/papers/ewr6.html
Paper: Robust Fitting of Parallax-Aware Mixtures for Path Guiding
Author: Lukas Ruppert
dl.acm.org/doi/10.1145/3386569.3392421
Paper: Practical Path Guiding for Efficient Light-Transport Simulation (a.k.a. Practical Path Guiding)
Author: Thomas Müller
dl.acm.org/doi/10.1111/cgf.13227
Paper: Path Guiding in Production
Author: Jikra Vorba
cgg.mff.cuni.cz/~jirka/path-guiding-in-production/2019/index.htm
See also:
oneapi.com
Guests:
Dian Nikolov – Software Engineer @ Chaos
Sebastian Herholz – Software Engineer @ Intel
Quantum computing technology is becoming more viable, which means programming models are going to have to be created to harness this potentially world changing technology. To many developers, understanding the possibilities of quantum computing and moving from a binary world to a world of quantum states is challenging.
Tony talks to Anne Matsuura, a physicist and Director of Quantum Applications and Architecture in Intel Labs about what quantum computing can do for society and how Intel is working on building hardware and software solutions to make that future a reality.
We also talk about what a quantum computer actually looks like today (hint: it’s not a tiny flat chip).
Learn more:
Intel Quantum Computing Homepage
intel.com/content/www/us/en/research/quantum-computing.html
oneAPI
oneapi.io
intel.com/content/www/us/en/developer/tools/oneapi/overview.html
Guests:
Anne Matsuura – Director of Quantum Applications and Architecture in Intel Labs
AI is the new hot thing in business. Everyone wants to talk about how they are harnessing AI to make their business better. However this is not a trivial process, and maybe companies are left wondering what went wrong. Aible is a company devoted to providing business impact in less than 30 days.
We talk to Arijit Sengupta, the CEO of Aible, about how Aible is able to make an impact quickly. He talks about how Aible can transform an AI solution into a business solution, using high schoolers and Aible versus data scientists to prove the point. We also discuss bias in AI, the future of AI, and how AI has the power to transform how we work and live.
Learn more:
Aible
aible.com
Aible/Intel blog
intel.com/content/www/us/en/developer/articles/technical/30-days-to-ai-value-intel-aible-solution.html#gs.krmfw7
Intel/Aible Performance Benchmark and Case Studies Report
enaible.aible.com/hubfs/Intel_Aible_Benchmark.pdf
Case Study Videos: How to guarantee impact from AI in 30 days
aible.wistia.com/medias/2dddd6pr8i
AI is a Waste of Money
aible.com/ai-book
Intel AI Analytics Toolkit
intel.com/content/www/us/en/developer/tools/oneapi/ai-analytics-toolkit.html
Intel Distribution of OpenVINO™ Toolkit
intel.com/content/www/us/en/developer/tools/openvino-toolkit/overview.html
oneAPI
oneapi.io
intel.com/content/www/us/en/developer/tools/oneapi/overview.html
Guests:
Arijit Sengupta – Aible CEO
As AI becomes more prevelent in society, one of the biggest questions is how can we take AI and harness it for good? One area that is extremely active is healthcare. AI is being explored across a variety of disciplines in the medical field.
One way to use AI in healthcare is to make doctors more efficient, and that is precisely what Chest-rAi aims to do. Learn a little bit about how AI can be harnessed to make it easier for doctors to be more efficient and help more patients through use of Chest-rAi.
Learn more:
LTTS Chest-rAi and Intel AI Analytics and OpenVINO Toolkits blog
intel.com/content/www/us/en/developer/articles/technical/ltts-chest-rai-and-intel-tools-redefine-radiology.html
Chest-rAi
ltts.com/solutions/healthcare/AI-radiography/chest-rAi
Intel AI Analytics Toolkit
intel.com/content/www/us/en/developer/tools/oneapi/ai-analytics-toolkit.html
Intel Distribution of OpenVINO™ Toolkit
intel.com/content/www/us/en/developer/tools/openvino-toolkit/overview.html
oneAPI
oneapi.io
intel.com/content/www/us/en/developer/tools/oneapi/overview.html
The accelerator landscape has been rapidly changing over the past decade, and as super computers come online, they face the challenge of supporting the latest and greatest accelerators. Through use of C++. Kokkos and SYCL, the engineers working on the Aurora project at Argonne National Laboratory are demonstrating how to enable workloads that are performant now, while continuing to keep an eye on the future.
Guests:
Nevin Liber – Computer Scientist @ Argonne Leadership Computing Facility
alcf.anl.gov/about/people/nevin-liber
James Reinders – Technical Evangelist @ Intel
linkedin.com/in/jamesreinders
Learn more:
See what next-generation scientific problems are being slated to run on Aurora
intel.com/content/www/us/en/customer-spotlight/stories/argonne-aurora-customer-story.html
Argonne Leadership Computing Facility and Aurora
alcf.anl.gov
alcf.anl.gov/aurora
SYCL Tech
sycl.tech
oneAPI
oneapi.io
intel.com/content/www/us/en/developer/tools/oneapi/overview.html
Kokkos
github.com/kokkos/kokkos
Today, Codeplay Software announces expanding oneAPI for Nvidia and AMD GPUs, making multiarchitecture, multivendor programming much easier. This includes new plugins: oneAPI for Nvidia GPUs and oneAPI for AMD GPUs (beta).
Software developers are increasingly looking to move away from proprietary interfaces and use open standards that are defined by the industry. This can allow them to target a broad set of platforms and architectures. oneAPI aims to bring this to reality by supporting the widest set of processors and accelerators. As a cross-industry initiative, oneAPI offers an open, standards-based, multiarchitecture programming model and specification.
Having official plugins for Nvidia and AMD platforms is a major milestone for the oneAPI open community. The oneAPI Nvidia and AMD GPU plugins seamlessly work with the Intel oneAPI DPC++/C++ Compiler 2023 and many popular libraries, bringing open, standards-based programming to a range of processors from multiple hardware vendors. The oneAPI for Nvidia GPUs and oneAPI for AMD GPUs packages are available to download for free on the Codeplay website.
Learn more:
Intel oneAPI toolkits
intel.com/content/www/us/en/developer/tools/oneapi/toolkits.html
Codeplay
developer.codeplay.com/home/
oneAPI for NVIDIA GPUs
developer.codeplay.com/products/oneapi/nvidia/home/
oneAPI for AMD GPUs (beta)
developer.codeplay.com/products/oneapi/amd/home/
James’ blog on the Intel oneAPI 2023 Toolkit release
intel.com/content/www/us/en/developer/articles/news/oneapi-2023.html
Codeplay blog on the oneAPI for NVIDIA and AMD plugins
codeplay.com/portal/blogs/2022/12/16/bringing-nvidia-and-amd-support-to-oneapi.html
Guests:
Gordon Brown – Codeplay oneAPI Product Manager
James Reinders – Intel Evangelist
ChatGPT, DALL-E, Stable Diffusion and other generative AI have rapidly captivated the public. They provide new, easy ways, to create new content based on AI models that have been trained on large, public data sets. How does this technology work? What does it mean for society? In this podcast we discuss the technology behind generative AI, some of the potential legal and societal issues, and what the pros and cons of generative AI are.
For more information see:
Attention is All You Need
arxiv.org/abs/1706.03762
Copilot
github.com/features/copilot
ChatGPT
openai.com/blog/chatgpt
Stable Diffusion
stability.ai/blog/stable-diffusion-public-release
Midjourney
midjourney.com
openVINO
intel.com/content/www/us/en/developer/tools/openvino-toolkit/overview.html
Intel Extension for PyTorch
intel.com/content/www/us/en/developer/tools/oneapi/extension-for-pytorch.html
oneAPI
intel.com/content/www/us/en/developer/tools/oneapi/overview.html
Tim Davis @ Texas A&M Github Copilot Tweet
twitter.com/DocSparse/status/1581461734665367554
ChatGPT Tech Radar Article
techradar.com/news/i-asked-chatgpt-to-write-about-itself-in-my-style-so-i-didnt-have-to
Guests
Ezequiel Lanza – Open Source, AI Evangelist @ Intel
Benjamin Consolvo – AI Community Manager @ Intel
Parallel programming models are often used on large scale systems, with huge datasets and a large amount of obvious parallelism....[…]
One of the greatest challenges to using AI/ML in industry is reducing your time to solution so that your data...[…]
The Argonne Leadership Computing Facility (ALCF) and Intel are working together on Aurora, an Exascale supercomputer. As the ALCF and...[…]
The Argonne Leadership Computing Facility (ALCF) and Intel are working together on Aurora, an Exascale supercomputer. As the ALCF and...[…]
As AI continues to play an ever-increasing role in applications and decision-making, the challenges facing developers of these AI workloads...[…]
Intel, HPE, and Argonne National Lab are collaborating closely to deliver Aurora. Built on Intel's architecture, Aurora will be Argonne's...[…]
The exponential growth of data generation and the need for it’s storage is driving the next phase of innovation in...[…]
Ever since its introduction in 2014, SYCL has grown in popularity and adoption. It is a royalty-free, cross-platform abstraction layer that enables code for heterogeneous processors, written in a “single-source” style using C++ standards. The flexibility to deploy across multiple platforms, reuse code helps enable advanced hardware features that can be used by developers. This […]
The rapid advancement in machine learning and data science fields have aided data scientists in arriving at meaningful insights. However, it’s not been an easy task to optimize machine learning infrastructures to allow data scientists to focus on their core expertise. Today, we will discuss how certain tools and hardware optimizations are not only saving […]
In the gaming world, artificial intelligence (AI) is proving to be a game-changer. It has played a pivotal role in enhancing game-player’s experiences. Amongst other capabilities, Artificial intelligence is used to generate responsive, adaptive, or intelligent behaviors primarily in non-player characters (NPCs) similar to human-like intelligence. With the right tools and access to advanced technologies, developers […]
Scientific visualization (SciVis) is a process of representing raw scientific data as images, which helps scientists improve their interpretations of large data sets. More and more, advanced visualization tools are becoming an integral part of a researcher’s analysis toolkit; and when in the hands of scientists and researchers, they can visualize, interact with, and get […]
Adoption of Python has been enormous over the last decade. Why? It’s easy to use, accessible, versatile and can be used for AI, machine learning, data analytics, data visualization, and all types of science. Tune into how 3 experts are involved in building common standards for Python data APIs that can help users use a […]
Ginkgo is a production-ready, sparse linear algebra library used for HPC on GPU architectures. It’s now using oneAPI cross-architecture programming to support its foundational design with a high level of performance portability, and focus on software sustainability. ExpertsHartwig Anzt at Karlsruhe Institute of Technology (KIT) and Univ. of Tennessee, and Terry Cojean of KIT provide […]
Scientific visualization (SciVis) is providing new insights in health care and many other research areas in ways that we couldn’t imagine a decade ago. It’s an evolving field where extracting visualizations from very large data sets can improve analysis, diagnosis and patient outcomes while reducing costs – or even uncover hidden discrimination. Hear from Teodora […]
Heterogeneous programming is challenging for developers because it’s been difficult to code or forces compromises in performance. oneAPI is working to break that mold. Many HPC research centers, enterprises and developers are now adopting oneAPI. Hear from experts at Zuse Institute Berlin (ZIB) on their oneAPI story: “We want to have most code portability across […]
Just as a carpenter needs tools to build a house, a developer needs tools to write code and programs. This often means relying on other coders to develop tools and building on that foundation. Xiaozhu Meng, a research scientist at Rice University shares the progress of the HPC Toolkit Project. It’s focused on performance tuning […]
Bioinformatics is advancing health-care research by using computation to understand biological data. It’s useful for large complex data sets used in determining gene and protein functions, establishing evolutionary relationships, and predicting 3D shapes of proteins. Two experts, Sergio Santander-Jimenez at University of Extremadura, and Ricardo Nobre of INESC-ID, are combining the power of modern hardware […]
The next generation of scientists and coders need more than a love of science and technology. The use of supercomputers in digital design to simulation pose opportunities solving the world’s biggest questions. Katherine Riley, director of science for Argonne’s Leadership Computing Facility, and Joe Curley, Intel senior director of oneAPI products, solutions and ecosystem, share […]
INESC-ID researchers Aleksandar Ilic and Diogo Augusto Pereira Marques reveal their journey in extending Roofline Modeling for use in application optimization, known as the Cache-aware Roofline Model, or CARM, which has been incorporated into the Intel® Advisor tool and recognized with an award from the HiPEAC community in Europe. They are now taking this model […]
Despite its tough challenges, our global pandemic has opened the door to big opportunities, ushering in incredible innovation in visual effects and the discovery of amazing talent regardless of location. Real-time technology and tools, machine learning and artificial intelligence, and the rise of volume rendering, in conjunction with geometric rendering, are transforming storytelling. Collaboration on […]
Dr. Ruymán Reyes Castro, CTO at Codeplay Software, and Kevin Harms, Senior Software Developer at Argonne Leadership Computing Facility, talk about their collaboration with the National Energy for Research Scientific Computing Center (NERSC) at Lawrence Berkeley Lab to enhance the LLVM-based DPC++ open source compiler, based on the SYCL standard, to support Nvidia GPUs. They […]
Epistasis detection holds great promise for the prevention and treatment of diseases such as Alzheimer’s, diabetes and breast cancer. INESC-ID researchers Aleksander Ilic and Rafael Torres Campos are collaborating with other researchers and developers across Portugal through the HiPErBio Project to advance the precision and efficiency of epistasis detection, using oneAPI DevCloud, Intel CPUs and […]
“It’s really exciting to see all the sorts of things that are becoming smart in our daily lives,” remarks John Melonakos, CEO & co-founder of ArrayFire. In this episode, John talks about the challenges he sees as heavy computational workloads move closer to the edge, and how important data parallelism and open standards are in […]
We explore the critical role of visualization and visual analytics in disaster management and medical research with Paul Navrátil, Director of Visualization at the Texas Advanced Computing Center (TACC), and Jim Jeffers, Sr. Principal Engineer and Sr. Director of Advanced Rendering and Visualization at Intel. Paul reveals TACC’s approach to recent weather crises, as well […]
In this second episode about GROMACS, one of the world’s most widely used open source molecular dynamics (MD) applications, we explore the quest to simplify portable, performant programming with Erik Lindahl, Biophysics Professor at Stockholm University & KTH Royal Institute of Technology, and Roland Schulz, Parallel Software Engineer at Intel. Andrey Alekseenko, Postdoctoral Fellow at […]
Open standards underpin the future of heterogeneous programming, and as Intel Principal Engineer Jeff Hammond points out, SYCL has proven in many ways that “standards don’t cost you a thing and you get portability.” A modern C++ language, SYCL has demonstrated amazing growth and now boasts four implementations—ComputeCpp, triSYCL, hipSYCL and DPC++. hipSYCL Creator Aksel […]
Increased processing power, massive amounts of data, and the development of more advanced algorithms have brought deep learning to the forefront, and TensorFlow has emerged as one of the world’s most popular machine learning frameworks. Penporn Koanantakool, Google Senior Software Engineer, and Ramesh, Intel Principal Engineer, share how their teams are collaborating to optimize TensorFlow […]
Andrew Lumsdaine, Chief Scientist at Northwest Institute for Advanced Computing, shares his journey in parallelism, from the HPC community’s early skepticism of and eventual move to C++ to build large systems, to his pursuit to develop a series of C++ libraries to strike at the contradiction between abstraction and performance, to today’s reality that parallelism […]
Who better to have a spicy discussion with about #OpenMP than Tim Mattson and Bronis de Supinski? These two have truly lived at the forefront of the amazing, decades-long OpenMP journey, from its inception to its preeminence as a foundational tool for HPC application programmers. Listen to what’s coming in 5.1 and beyond, how the […]
Denisa Constantinescu, a PhD student in Mechatronics, and a researcher in the Computer Architecture Department at the University of Malaga, and Maura Tokay, a lead software programmer at Science Systems and Applications, Inc., and a computer scientist within the Department of Agriculture, share how their work is helping advance the fields of robotics, economics, manufacturing, […]
Dr. Tom Deakin, senior research associate and lecturer in the High-Performance Computing Research Group at the University of Bristol, and Dr. James Brodman, software engineer at Intel, unpack the tricky topic of performance portability to reveal what this concept truly means and ways to achieve it. As contributors to the Khronos SYCL Working Group—from the […]
As high performance computing (HPC) moves far beyond its traditional users to the domains of AI, machine learning, enterprises and the cloud, and heterogeneous programming becomes the norm, Nico Mittenzwey, senior HPC engineer at Megware Computer, and Mike Lee, technical, enterprise and cloud segment marketing manager at Intel, discuss the considerations in optimizing HPC clusters […]
Data scientists spend 60% of their time cleaning and preprocessing data, transforming this dirty data into crystallized insights. Dataframes, such as Pandas, provide exceptional tooling to address data wrangling tasks, yet Pandas themselves increasingly lack ease and speed as they scale. Alex Baden, Technical Director at OmniSci, and Devin Petersohn, Machine Learning Engineer at Intel, […]
To deliver increasingly captivating stories in the world of animated movies takes passion and determination. The passion to continually push the boundaries of what’s possible. The determination to continually seek ways to make technology easily approachable by artists, so that the artists can focus on what they’re best at—telling stories that capture our imaginations. Max […]
In Part I of two episodes about GROMACS, one of the world’s most widely used open source molecular dynamics (MD) applications, Erik Lindahl, Biophysics Professor at Stockholm University & KTH Royal Institute of Technology, and Roland Schulz, Parallel Software Engineer at Intel, unpack the mysteries of MD. As Erik explains, MD helps us understand “everything […]
Krystian Ligenza, Autodesk Maya Software Architect, and Mike Voss, Intel Principal Engineer, talk about how the need for immersive experiences in VFX and 3D animated movies and games has driven the need for increasingly more capable (and complex) software and hardware. They discuss how Autodesk and Intel are addressing these challenges, and explain how the […]
Is oneAPI living up to its promise? How do oneAPI and DPC++ relate to the Khronos Group SYCL and ISO C++ standards? Andrew Richards, Codeplay Software CEO, and Hal Finkel, Lead for Compiler Technology and Programming Languages at Argonne National Lab, respond to these provocative questions and more from Sanjiv Shah, Intel GM and VP […]
Ronan Keryell, principal software engineer at Xilinx, and Jeff Hammond, principal engineer at Intel, explain why open collaboration — modeled through open source and open standards — is key to solving some of today’s biggest challenges in research and industry, revealing some of the misconceptions, or least understood aspects, along the way. Then they explore […]
Peter Wang, CEO and Co-Founder of Anaconda, explains the series of happy accidents that have led to organic adoption of Python as the number one language among developers in data science and machine learning in this episode of Code Together. He and David Liu, an AI Solutions Engineer at Intel, talk about the latest advancements […]
Existing math kernel libraries have lacked portability across heterogeneous platforms—until now. A unifying programming model—and availability of standard library interfaces—enables development of performance-portable libraries among diverse hardware architectures. In this episode, Julia Sukharina, Senior Engineering Manager at Intel, and Mehdi Goli, Principal Software Engineer at Codeplay Software, talk about a collaborative project to make this […]
In this episode of Code Together, Geoff Lowney, Senior Fellow at Intel, and Andrew Richards, CEO at Codeplay Software, discuss the value of standards in enabling hardware and software developers to collaborate and innovate. From the early days of video games to modern AI, Andrew weaves an unexpected yet common thread between these, and the […]
In this episode of Code Together, we talk to Alice Chan, Vice President and General Manager of Compiler Engineering at Intel, and Hal Finkel, Lead for Compiler Technology and Programming Languages at Argonne National Lab, about how the industry is uniting to address the need for programming portability and performance across diverse architectures, particularly important […]