With a Carbon Footprint Like HPC’s, It Matters When and Where You Step

By Oliver Peckham

December 9, 2021

From European HPC experts pondering “can fast be green?” to new milestones on the Green500 list, sustainability certainly had a moment at the hybrid SC21 conference. And it’s no wonder: the exascale era is here, and power consumption for HPC is skyrocketing even as efficiency is driven to its extremes. At SC21, another session – “HPC’s Growing Sustainability Challenges and Emerging Approaches” – tackled the topic, bringing together experts across the HPC sector to discuss how the rapidly expanding field could control its energy use and emissions.

“The context is – as I’m sure you’re all aware – that there’s aggressive government goals to decarbonize their economies and reduce negative environmental impacts broadly, and computing is sort of in a special place in this,” explained the session’s host, Andrew Chien, a senior scientist at Argonne National Laboratory. “Computing and HPC together have challenges around sustainability because of their rapidly proliferating use and very, very fast technology lifecycles.”

Chien was joined by Steve Hammond, a senior research advisor at the National Renewable Energy Laboratory; Bill Magro, chief technologist for HPC at Google; Michael McNamara, CEO and founder of Lancium; and Erik Riedel, senior vice president of engineering at ITRenew.

HPC has a big footprint.

“It’s actually particularly timely that we have this panel on sustainability,” Chien continued, “because [the international climate conference] COP26 has been going on, and maybe just completed … but entire economies – including of course HPC and computing – have made the commitment to get to net zero by various times, whether it’s 2050, 2060, or 2070.”

HPC, Chien continued, was not an insignificant slice of the pie: using the Top500 as a proxy, Chien said, researchers had estimated the systems’ aggregate energy load at around 600MW, or 5.2 terawatt-hours per year, with the top ten systems constituting around 100MW of that load. This, he said, would account for around two million metric tons of CO2 per year – equivalent to around 285,000 average households, and that only from 500 publicly-ranked systems.

And HPC’s footprint doesn’t stop at energy use (“scope two” emissions). “Scope three” emissions, which include embedded carbon from along the value chain, are also high for the sector. “We have the challenging practice of a fast-moving technology, which is that we dispose of our systems after a relatively short lifetime,” Chien said, noting that there was a “very, very significant amount of energy” corresponding to production of components for HPC systems (along with e-waste, of course).

Choosing when to step matters.

But, it turns out, with a big footprint, choosing when to step can be crucial. Normally, Chien said, computing loads are constant, making it hard to go carbon-free as the power grids hosting those computing loads fluctuate in terms of carbon intensity throughout the day. To that end, Chien detailed a project he had been working on called zero-carbon cloud (ZCCloud).

“Instead of having all of your datacenters as constant loads in the grid,” he said, “we might be able to have computing equipment modulate its consumption to the availability of excess renewable power.”

“If you take this study to its logical conclusion,” he continued, there are some “amazing opportunities.” The team had looked at workloads at Argonne for about a year, examining expected performance, throughput and energy use between a traditional, reliably-powered system and a system powered by intermittent renewable energy. The latter system, Chien said, “effectively eliminates 50 percent of the carbon emissions footprint for this aggregate computing system.”

“This approach – a holistic approach that looks at TCO from capital equipment on through operational costs and power and so on – could lead you to the conclusion that you could build a system with higher throughput per million dollars of TCO per year and … higher peak performance per million dollars of TCO per year. So this is in addition to the 50 percent emissions reduction.”

“If you look at this in the right way,” he concluded, “there may be opportunities that not only reduce the carbon emissions footprint, but also create new opportunities for new capability and even cost-effectiveness.”

Magro, in his talk, agreed with both the stated problem and the proposed solution. “Even if you want to commit to being carbon-free, the problem is there isn’t carbon-free energy available all day,” he said. “The energy supply is very spiky. … If you just present a stable load, then of course you can’t run carbon-free.”

Google, he explained, had moved away from constant loads in many cases as a way of combating this reality. “In our datacenters, we actually align with the availability of low-carbon energy and we compute more when the wind is blowing and when the sun is shining,” he said.

A simplified visualization of how Google adjusts its workloads based on renewable energy availability. Image courtesy of Bill Magro.

Choosing where to step matters.

But time isn’t the only variable in play – location matters, too. Hammond, who helps to manage the world’s most energy-efficient datacenter at NREL’s campus in Golden, Colorado, detailed the wide range of efficiency choices made to give the datacenter its stunning PUE of 1.036… with one major caveat. “No one approach fits every solution wherever you are,” he said. “What works for us here in Colorado may be different from what works in the Southeastern U.S.”

This, as it turns out, was the entire pitch for another speaker – McNamara – who flipped that idea on its head: what if regionally-specific sustainability considerations played a major role in siting workloads?

McNamara first outlined the characteristics of the modern (and future) power grid. First, he said, solar and wind were getting overbuilt to compensate for their intermittency; second, generation was no longer necessarily colocated to urban centers; and third, as a result of these trends, negative-priced energy – energy that renewable generators needed to offload, but for which there was not demand – was a growing problem.

“Negative-priced energy is now endemic,” he said. “And it’s endemic because the grid of the future has too much wind and solar. … The way to solve this problem is to build, at extremely large scale, many datacenters at critical points on the transmission system in every grid operator that is going decarbonized. Which is all of them.”

Negative pricing frequency, with Lancium’s target region circled. Image courtesy of Michael McNamara.

This, he said, was the pitch for his company, Lancium. “Our vision is: datacenters should act as enormous inverse power plants,” he said. “What power plants do is, when the grid is in times of need, power plants go up. The grid is indifferent if a power plant goes up or a datacenter goes down. … They consume negative-priced energy; the provide the grid inertia that the grid needs because of the retirement of fossil plants; and the net effect, if you run these in a highly flexible manner, is negative eight million tons [of carbon] a year if you build 5000MW of datacenters. And that’s in Texas alone. And that’s what we’re doing.”

McNamara said that this solution – which could, ostensibly, run massive workloads at zero energy cost – was preferable to costly, time-consuming transmission line upgrades and more foolproof than current storage solutions, which he said were good for about four hours. Further, he explained that costly sustainability solutions were not necessary in Lancium’s datacenters, which ran hot and “moved a ton of air.”

Hand-me-downs

But all of that, of course, only addresses the scope two emissions. Scope three emissions were an entirely different beast: “It boggles my mind how much we just have to throw away when we get a new system,” Hammond said. Riedel explained how his company, ITRenew, worked to decommission datacenter equipment for companies like Facebook, Google and Twitter, shredding hard drives and repurposing other equipment to give it a second life.

Chien, earlier in the panel, had pitched this same notion, outlining how datacenter operators could deploy phased-out hardware at a secondary facility with low-cost, on-site renewable energy for another few years. This, he said, would be a win-win-win: low-cost equipment, low-cost energy and lower (annualized) embedded emissions thanks to longer lifetimes for the components.

More sustainability coverage from SC21

At SC21, Experts Ask: Can Fast HPC Be Green?

The Green500’s Crystal Anniversary Sees MN-3 Crystallize Its Winning Streak

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!

Quantum Watchers – Terrific Interview with Caltech’s John Preskill by CERN

July 17, 2024

In case you missed it, there's a fascinating interview with John Preskill, the prominent Caltech physicist and pioneering quantum computing researcher that was recently posted by CERN’s department of experimental physi Read more…

Aurora AI-Driven Atmosphere Model is 5,000x Faster Than Traditional Systems

July 16, 2024

While the onset of human-driven climate change brings with it many horrors, the increase in the frequency and strength of storms poses an enormous threat to communities across the globe. As climate change is warming ocea Read more…

Researchers Say Memory Bandwidth and NVLink Speeds in Hopper Not So Simple

July 15, 2024

Researchers measured the real-world bandwidth of Nvidia's Grace Hopper superchip, with the chip-to-chip interconnect results falling well short of theoretical claims. A paper published on July 10 by researchers in the U. Read more…

Belt-Tightening in Store for Most Federal FY25 Science Budets

July 15, 2024

If it’s summer, it’s federal budgeting time, not to mention an election year as well. There’s an excellent summary of the curent state of FY25 efforts reported in AIP’s policy FYI: Science Policy News. Belt-tight Read more…

Peter Shor Wins IEEE 2025 Shannon Award

July 15, 2024

Peter Shor, the MIT mathematician whose ‘Shor’s algorithm’ sent shivers of fear through the encryption community and helped galvanize ongoing efforts to build quantum computers, has been named the 2025 winner of th Read more…

Weekly Wire Roundup: July 8-July 12, 2024

July 12, 2024

HPC news can get pretty sleepy in June and July, but this week saw a bump in activity midweek as Americans realized they still had work to do after the previous holiday weekend. The world outside the United States also s Read more…

Aurora AI-Driven Atmosphere Model is 5,000x Faster Than Traditional Systems

July 16, 2024

While the onset of human-driven climate change brings with it many horrors, the increase in the frequency and strength of storms poses an enormous threat to com Read more…

Shutterstock 1886124835

Researchers Say Memory Bandwidth and NVLink Speeds in Hopper Not So Simple

July 15, 2024

Researchers measured the real-world bandwidth of Nvidia's Grace Hopper superchip, with the chip-to-chip interconnect results falling well short of theoretical c Read more…

Shutterstock 2203611339

NSF Issues Next Solicitation and More Detail on National Quantum Virtual Laboratory

July 10, 2024

After percolating for roughly a year, NSF has issued the next solicitation for the National Quantum Virtual Lab program — this one focused on design and imple Read more…

NCSA’s SEAS Team Keeps APACE of AlphaFold2

July 9, 2024

High-performance computing (HPC) can often be challenging for researchers to use because it requires expertise in working with large datasets, scaling the softw Read more…

Anders Jensen on Europe’s Plan for AI-optimized Supercomputers, Welcoming the UK, and More

July 8, 2024

The recent ISC24 conference in Hamburg showcased LUMI and other leadership-class supercomputers co-funded by the EuroHPC Joint Undertaking (JU), including three Read more…

Generative AI to Account for 1.5% of World’s Power Consumption by 2029

July 8, 2024

Generative AI will take on a larger chunk of the world's power consumption to keep up with the hefty hardware requirements to run applications. "AI chips repres Read more…

US Senators Propose $32 Billion in Annual AI Spending, but Critics Remain Unconvinced

July 5, 2024

Senate leader, Chuck Schumer, and three colleagues want the US government to spend at least $32 billion annually by 2026 for non-defense related AI systems.  T Read more…

Point and Click HPC: High-Performance Desktops

July 3, 2024

Recently, an interesting paper appeared on Arvix called Use Cases for High-Performance Research Desktops. To be clear, the term desktop in this context does not Read more…

Atos Outlines Plans to Get Acquired, and a Path Forward

May 21, 2024

Atos – via its subsidiary Eviden – is the second major supercomputer maker outside of HPE, while others have largely dropped out. The lack of integrators and Atos' financial turmoil have the HPC market worried. If Atos goes under, HPE will be the only major option for building large-scale systems. Read more…

Everyone Except Nvidia Forms Ultra Accelerator Link (UALink) Consortium

May 30, 2024

Consider the GPU. An island of SIMD greatness that makes light work of matrix math. Originally designed to rapidly paint dots on a computer monitor, it was then Read more…

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_1687123447

Nvidia Economics: Make $5-$7 for Every $1 Spent on GPUs

June 30, 2024

Nvidia is saying that companies could make $5 to $7 for every $1 invested in GPUs over a four-year period. Customers are investing billions in new Nvidia hardwa Read more…

Nvidia Shipped 3.76 Million Data-center GPUs in 2023, According to Study

June 10, 2024

Nvidia had an explosive 2023 in data-center GPU shipments, which totaled roughly 3.76 million units, according to a study conducted by semiconductor analyst fir Read more…

AMD Clears Up Messy GPU Roadmap, Upgrades Chips Annually

June 3, 2024

In the world of AI, there's a desperate search for an alternative to Nvidia's GPUs, and AMD is stepping up to the plate. AMD detailed its updated GPU roadmap, w Read more…

Some Reasons Why Aurora Didn’t Take First Place in the Top500 List

May 15, 2024

The makers of the Aurora supercomputer, which is housed at the Argonne National Laboratory, gave some reasons why the system didn't make the top spot on the Top Read more…

Intel’s Next-gen Falcon Shores Coming Out in Late 2025 

April 30, 2024

It's a long wait for customers hanging on for Intel's next-generation GPU, Falcon Shores, which will be released in late 2025.  "Then we have a rich, a very Read more…

Leading Solution Providers

Contributors

Google Announces Sixth-generation AI Chip, a TPU Called Trillium

May 17, 2024

On Tuesday May 14th, Google announced its sixth-generation TPU (tensor processing unit) called Trillium.  The chip, essentially a TPU v6, is the company's l 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…

IonQ Plots Path to Commercial (Quantum) Advantage

July 2, 2024

IonQ, the trapped ion quantum computing specialist, delivered a progress report last week firming up 2024/25 product goals and reviewing its technology roadmap. 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…

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…

The NASA Black Hole Plunge

May 7, 2024

We have all thought about it. No one has done it, but now, thanks to HPC, we see what it looks like. Hold on to your feet because NASA has released videos of wh 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…

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…

  • arrow
  • Click Here for More Headlines
  • arrow
HPCwire