UC San Diego’s Jupyterhub Platform Aids Students with Data-Intensive Computing Needs

January 24, 2022

Jan. 24, 2022 — In classic UC San Diego fashion, an overheard conversation at a campus coffee cart has turned into an interdisciplinary project that’s making computing-intensive coursework more exciting while saving well over one million dollars so far. The effort gives UC San Diego graduate and undergraduate students – and their professors – better hardware and software ecosystems for exploring real-world, data-intensive and computing-intensive projects and problems in their courses.

Larry Smarr, Distinguished Professor Emeritus, Department of Computer Science and Engineering at the UC San Diego Jacobs School of Engineering.

It all started while UC San Diego computer science and engineering professor Larry Smarr was waiting for coffee in the “Bear” courtyard at the Jacobs School of Engineering a little more than three years ago. In line, Smarr overheard a student say, “I can’t get a job interview if I haven’t run TensorFlow on a GPU on a real problem.”

While this one student’s conundrum may sound extremely technical and highly specific, Smarr heard a general need; and he saw an opportunity. In particular, Smarr realized that innovations coming out of a U.S. National Science Foundation (NSF) funded research project he leads—the Pacific Research Platform (PRP)—could be leveraged to create better computing infrastructure for university courses that rely heavily on machine learning, data visualizations, and other topics that require significant computer resources. This infrastructure would make it easier for professors to offer courses that challenge students to solve real-world data- and computation-intensive problems, including things like what he heard at the coffee cart: running TensorFlow on a GPU on a real problem.

Fast forward to 2022, and Smarr’s spark of an idea has grown into a cross-campus collaboration called the UC San Diego Data Science/Machine Learning Platform or the UC San Diego JupyterHub. Through this platform, the inexpensive, high-performance computational building blocks combining hardware and software that Smarr and his PRP collaborators designed for use in computation-intensive research across the country are now also the backbone of dynamic computing ecosystems for UC San Diego students and professors who use machine learning, data visualization, and other computing- and data- intensive tools in their courses. The Platform has been widely used in every Division on campus, with courses taught in biological sciences, cognitive science, computer science, data science, engineering, health sciences, marine sciences, medicine, music, physical sciences, public health and more. See a list of Jacobs School affiliated faculty and the names of the courses they have taught using the UC San Diego Data Science/Machine Learning Platform.

It’s a unique, collaborative project that leverages Federally funded computing research innovations for classroom use. To make the jump from research to classroom applications, a creative and hardworking interdisciplinary team at UC San Diego came together. UC San Diego’s IT Services / Academic Technology Services stepped up in a big way. Senior architect Adam Tilghman and chief programmer David Andersen led the implementation effort, with leadership and funding support from UC San Diego CIO Vince Kellen and Academic Technology Senior Director Valerie Polichar. The project has already helped the campus avoid well over one million dollars in cloud-computing spend, according to Kellen.

Usage patterns for the UC San Diego Data Science/Machine Learning Platform. The green regions represent available capacity for non-coursework use.
Examples of the software that students are able to run on the UC San Diego Data Science/Machine Learning Platform.

At the same time, the project gives the UC San Diego community tools to encourage the back-and-forth flow of students and ideas between classroom projects and follow-on research projects.

“Our students are getting access to the same level of compte capacity that normally only a researcher using an advanced system like a supercomputer would get. The students are exploring much more complex data problems because they can,” said Smarr, who was also the founding Director of the California Institute for Telecommunications and Information Technology (Calit2), a UC San Diego / UC Irvine partnership.

Personal genomics

One of the many professors from all across campus using the UC San Diego Data Science / Machine Learning Platform for courses is Melissa Gymrek, who is a professor in both the Department of Computer Science and Engineering and the Department of Medicine’s Division of Genetics.

Her students write and run code in a software environment called Jupyter Notebooks that runs on the UC San Diego platform. “They can write code in the notebook and press execute and see the results. They can build figures to visualize data. We focus a lot more now on data visualizations,” said Gymrek.

Xuan Zhang (UC San Diego Chemistry PhD, ’21) is one of the tens of thousands of UC San Diego students and young researchers who has used the UC San Diego Data Science/Machine Learning Platform extensively in courses.

One of the thousands of UC San Diego students who has used the platform extensively is Xuan Zhang. Through the data- and visualization- intensive coursework in CSE 284, Zhang realized that the higher order genetic structures at the center of her chemistry Ph.D. dissertation – R-Loops – could be regulated by the short tandem repeats (STRs) that are at the center of much of the research in Gymrek’s lab. Without the computing-infrastructure for real-world coursework problems, Zhang believes she would not have made the research connection.

After taking Gymrek’s course, Zhang also realized that she could apply to obtain her own independent research profile on the UC San Diego Data Science / Machine Learning Platform in order to retain access to all her coursework and to keep building on it. (When Jupyter Notebooks are hosted on the commercial cloud, students generally lose access to their data-intensive coursework when the class ends, unless they download the data themselves.)

“I thought it was just for the course, but then I realized that Jupyter Notebooks are available for research, without losing access through the UC San Diego Jupyterhub,” said Zhang.

This educational infrastructure has added benefits for professors as well.

“With these Jupyter Notebooks, you can automatically embed the grading system. It saves a lot of work,” said Gymrek. You can designate how many points a student gets if they get the code right, she explained. Before using this system, students sent PDFs of their problem sets which made grading more time intensive. “It was hard to go past a dozen students. Now, you can scale,” said Gymrek. In fact, she has been able to expand access to her personal genomics graduate class to more than 50 students, up from a dozen before she had access to these new tools.

Direct uploading of assignments and grades to the campus learning management system, Canvas, is also now available.

“The platform is truly transforming education. Unlike many learning technology innovations, classes in every division at UC San Diego have used the Data Science/Machine Learning Platform. Many thousands of students use it every year. It’s innovation with real impact, preparing our students in many — sometimes unexpected — fields to be leaders and innovators when they graduate,” said Polichar.

Professors and students from all six departments at the UC San Diego Jacobs School of Engineering are making great use of the UC San Diego Data Science/Machine Learning Platform. The numbers on each stacked bar represent the number of students in that Department using the DSMLP in that quarter.
Courses from all six departments at the UC San Diego Jacobs School of Engineering are run on the UC San Diego Data Science/Machine Learning Platform.
This graph shows courses in all discipines. The numbers in the bars are the number of courses that quarter and the colors show the campus divisions (HDSI is the UCSD Halıcıoğlu Data Science Institute) which used the UCSD DSMLP. This shows how JupyterHub is bringing data science and machine learning computing to a broad set of disciplines.

Commodity hardware for research and education

“If you build your distributed supercomputer, like the PRP, on commodity hardware then you can ride Moore’s Law,” explained Smarr.

UC San Diego ITS senior architect Adam Tilghman poses with some of the innovative computing hardware that has opened the door to more data-intensive and computing-intensive coursework for UC San Diego students. These PCs run a wide range of leading-edge software to help students program the system, record their results in Jupyter notebooks, and execute a variety of data analytic and machine learning algorithms on their problems.

Following this commodity hardware strategy, Smarr and his PRP collaborators developed hardware designs where performance goes up while prices go down over time. The computational building blocks developed by the PRP, that were repurposed by UC San Diego’s ITS, are rack-mounted PCs, containing multi-core CPUs, eight Graphics Processing Units (GPUs), and optimized for data-intensive projects, including accelerating machine learning on the GPUs. These PCs run a wide range of leading-edge software to help students program the system, record their results in Jupyter Notebooks, and to execute a variety of data analytic and machine learning algorithms on their problems.

Building on this commodity hardware approach to high performance computing has allowed UC San Diego to build a dynamic and innovative “on premises” ecosystem for data- and computing- intensive coursework, rather than relying solely on commercial cloud computing services.

“The commercial cloud doesn’t provide an ecosystem that gives students the same platform from course to course, or the same platform they have in their courses as they have in their research,” said Tilghman. “This is especially true in the graduate area where students are starting work in a course context and then they continue that work in their research. It’s that continuity, even starting as a lower division undergraduate, all the way up. I think that’s one of the innovative advantages that we give at UC San Diego.”

UC San Diego professors and students interested in learning more about the Data Science / Machine Learning Platform can find additional details and contact information on their website.

“I’ve been at this for 50 years,” said Smarr. “I don’t know of many examples where I’ve seen such a close linking of research and education all the way around, in a circle.”

This alignment of research and education feeds into UC San Diego’s culture of innovation and relevance.

“It’s essential for the nation that students all across campus learn and work on computing infrastructure that is relevant for their future, whether it’s in industry, academia or the public sector,” said Albert P. Pisano, dean of the UC San Diego Jacobs School of Engineering. “These information technology ecosystems being created and deployed on campus are critical for empowering our students to leverage innovations to serve society.”

Pacific Research Platform

Click here for a video that gives an overview of the Pacific Research Platform (PRP) and includes a sampling of research projects the platform has enabled.

Larry Smarr serves as Principal Investigator on the PRP and allied grants (NSF Awards OAC-1541349, OAC-1826967, CNS-1730158, CNS-2100237) which are administered through the Qualcomm Institute, which is the UC San Diego Division of Calit2.


Source: Daniel Kane, UC San Diego

Subscribe to HPCwire's Weekly Update!

Be the most informed person in the room! Stay ahead of the tech trends with industry updates delivered to you every week!

MLPerf Inference 4.0 Results Showcase GenAI; Nvidia Still Dominates

March 28, 2024

There were no startling surprises in the latest MLPerf Inference benchmark (4.0) results released yesterday. Two new workloads — Llama 2 and Stable Diffusion XL — were added to the benchmark suite as MLPerf continues Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing power it brings to artificial intelligence.  Nvidia's DGX Read more…

Call for Participation in Workshop on Potential NSF CISE Quantum Initiative

March 26, 2024

Editor’s Note: Next month there will be a workshop to discuss what a quantum initiative led by NSF’s Computer, Information Science and Engineering (CISE) directorate could entail. The details are posted below in a Ca Read more…

Waseda U. Researchers Reports New Quantum Algorithm for Speeding Optimization

March 25, 2024

Optimization problems cover a wide range of applications and are often cited as good candidates for quantum computing. However, the execution time for constrained combinatorial optimization applications on quantum device Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at the network layer threatens to make bigger and brawnier pro Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HBM3E memory as well as the the ability to train 1 trillion pa Read more…

MLPerf Inference 4.0 Results Showcase GenAI; Nvidia Still Dominates

March 28, 2024

There were no startling surprises in the latest MLPerf Inference benchmark (4.0) results released yesterday. Two new workloads — Llama 2 and Stable Diffusion Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing po Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HB Read more…

Nvidia Looks to Accelerate GenAI Adoption with NIM

March 19, 2024

Today at the GPU Technology Conference, Nvidia launched a new offering aimed at helping customers quickly deploy their generative AI applications in a secure, s Read more…

The Generative AI Future Is Now, Nvidia’s Huang Says

March 19, 2024

We are in the early days of a transformative shift in how business gets done thanks to the advent of generative AI, according to Nvidia CEO and cofounder Jensen Read more…

Nvidia’s New Blackwell GPU Can Train AI Models with Trillions of Parameters

March 18, 2024

Nvidia's latest and fastest GPU, codenamed Blackwell, is here and will underpin the company's AI plans this year. The chip offers performance improvements from Read more…

Nvidia Showcases Quantum Cloud, Expanding Quantum Portfolio at GTC24

March 18, 2024

Nvidia’s barrage of quantum news at GTC24 this week includes new products, signature collaborations, and a new Nvidia Quantum Cloud for quantum developers. Wh Read more…

Alibaba Shuts Down its Quantum Computing Effort

November 30, 2023

In case you missed it, China’s e-commerce giant Alibaba has shut down its quantum computing research effort. It’s not entirely clear what drove the change. Read more…

Nvidia H100: Are 550,000 GPUs Enough for This Year?

August 17, 2023

The GPU Squeeze continues to place a premium on Nvidia H100 GPUs. In a recent Financial Times article, Nvidia reports that it expects to ship 550,000 of its lat Read more…

Shutterstock 1285747942

AMD’s Horsepower-packed MI300X GPU Beats Nvidia’s Upcoming H200

December 7, 2023

AMD and Nvidia are locked in an AI performance battle – much like the gaming GPU performance clash the companies have waged for decades. AMD has claimed it Read more…

DoD Takes a Long View of Quantum Computing

December 19, 2023

Given the large sums tied to expensive weapon systems – think $100-million-plus per F-35 fighter – it’s easy to forget the U.S. Department of Defense is a Read more…

Synopsys Eats Ansys: Does HPC Get Indigestion?

February 8, 2024

Recently, it was announced that Synopsys is buying HPC tool developer Ansys. Started in Pittsburgh, Pa., in 1970 as Swanson Analysis Systems, Inc. (SASI) by John Swanson (and eventually renamed), Ansys serves the CAE (Computer Aided Engineering)/multiphysics engineering simulation market. Read more…

Choosing the Right GPU for LLM Inference and Training

December 11, 2023

Accelerating the training and inference processes of deep learning models is crucial for unleashing their true potential and NVIDIA GPUs have emerged as a game- Read more…

Intel’s Server and PC Chip Development Will Blur After 2025

January 15, 2024

Intel's dealing with much more than chip rivals breathing down its neck; it is simultaneously integrating a bevy of new technologies such as chiplets, artificia Read more…

Baidu Exits Quantum, Closely Following Alibaba’s Earlier Move

January 5, 2024

Reuters reported this week that Baidu, China’s giant e-commerce and services provider, is exiting the quantum computing development arena. Reuters reported � Read more…

Leading Solution Providers

Contributors

Comparing NVIDIA A100 and NVIDIA L40S: Which GPU is Ideal for AI and Graphics-Intensive Workloads?

October 30, 2023

With long lead times for the NVIDIA H100 and A100 GPUs, many organizations are looking at the new NVIDIA L40S GPU, which it’s a new GPU optimized for AI and g Read more…

Shutterstock 1179408610

Google Addresses the Mysteries of Its Hypercomputer 

December 28, 2023

When Google launched its Hypercomputer earlier this month (December 2023), the first reaction was, "Say what?" It turns out that the Hypercomputer is Google's t Read more…

AMD MI3000A

How AMD May Get Across the CUDA Moat

October 5, 2023

When discussing GenAI, the term "GPU" almost always enters the conversation and the topic often moves toward performance and access. Interestingly, the word "GPU" is assumed to mean "Nvidia" products. (As an aside, the popular Nvidia hardware used in GenAI are not technically... Read more…

Shutterstock 1606064203

Meta’s Zuckerberg Puts Its AI Future in the Hands of 600,000 GPUs

January 25, 2024

In under two minutes, Meta's CEO, Mark Zuckerberg, laid out the company's AI plans, which included a plan to build an artificial intelligence system with the eq Read more…

Google Introduces ‘Hypercomputer’ to Its AI Infrastructure

December 11, 2023

Google ran out of monikers to describe its new AI system released on December 7. Supercomputer perhaps wasn't an apt description, so it settled on Hypercomputer Read more…

China Is All In on a RISC-V Future

January 8, 2024

The state of RISC-V in China was discussed in a recent report released by the Jamestown Foundation, a Washington, D.C.-based think tank. The report, entitled "E Read more…

Intel Won’t Have a Xeon Max Chip with New Emerald Rapids CPU

December 14, 2023

As expected, Intel officially announced its 5th generation Xeon server chips codenamed Emerald Rapids at an event in New York City, where the focus was really o Read more…

IBM Quantum Summit: Two New QPUs, Upgraded Qiskit, 10-year Roadmap and More

December 4, 2023

IBM kicks off its annual Quantum Summit today and will announce a broad range of advances including its much-anticipated 1121-qubit Condor QPU, a smaller 133-qu Read more…

  • arrow
  • Click Here for More Headlines
  • arrow
HPCwire