EU Projects Unite on Heterogeneous ARM-based Exascale Prototype

By Tiffany Trader

February 24, 2016

A trio of partner projects based in Europe – Exanest, Exanode and Ecoscale – are working in close collaboration to develop the building blocks for an exascale architecture prototype that will, as they describe, put the power of ten million computers into a single supercomputer. The effort is unique in seeking to advance the ARM64 + FPGA architecture as a foundational “general-purpose” exascale platform.

Funded for three years as part of Europe’s Horizon2020 program, the partners are coordinating their efforts with the goal of building an early “straw man” prototype late this year that will consist of more than one-thousand energy-efficient ARM cores, reconfigurable logic, plus advanced storage, memory, cooling and packaging technologies.

Exanest is the project partner that is focused on the system level, including interconnection, storage, packaging and cooling. And as the name implies, Exanode is responsible for the compute node and the memory of that compute node. Ecoscale focuses on employing and managing reconfigurable logic as accelerators within the system.

Exanest

Manolis Katevenis, the project coordinator for Exanest and head of computer architecture at FORTH-ICS in Greece, explains that Exanest has set an early target of 2016 to build this “relatively-large” first prototype, comprised of at least one-thousand ARM cores.

He says, “We are starting early with a prototype based on existing technology because we want system software to be developed and applications to start being ported and tuned. For the remainder of the two years, there will be ongoing software development, plus research on interconnects, storage and cooling technologies. We also believe that there will be new interesting compute nodes coming out from our partner projects and we will use such nodes.”

In discussing target workloads, Katevenis emphasizes flexibility and breadth, echoing the sentiments we are hearing from across the HPC community. The goal for this platform is to be able to support a range of applications, both on the traditional compute and physics side and the data-intensive side. A look at the Exanest partner list hints at the kind of high-performance applications that will be supported: astrophysics, nuclear physics, simulation-based engineering, and even in-memory databases with partner MonetDB Solutions. Allinea will be providing the ARMv8 profiling and debugging tools.

Although the projects are still in the specification phase, they will be making selections with the aim of overcoming the specific challenges related to exascale. Areas of focus include compact packaging, permanent storage, interconnection, resilience and application behavior. Some of the design decisions were revealed in this poster from Exanest that shows a diagram of the daughterboard and blade design. Note that Xilinx is a key partner.

Exanest daughterboard and blade design

To achieve a complete prototype capable of running real-world benchmarks and applications by 2018, the primary partners are collaborating with a number of other academic groups and industry partners using co-design principles to develop the hardware and software elements. This is a classic public-private arrangement where academic and industrial partners join forces and industrial partners benefit by being able to reuse the technology that is developed.

On the technology side, packaging and cooling is a key focus for Exanest, which will rely on Iceotope, the immersive cooling vendor, to design an innovative cooling environment. The first prototype will employ Iceotope technology and there is the expectation that technology with even higher power density will be developed as the project progresses.

One of the primary criteria for the project partners is low-energy consumption for the main processor. They have chosen 64-bit ARM processors as their main compute engine. Katevenis affirms that having a processor that consumes dramatically less power allows many more cores to be packaged in the same physical volume and within the same total power consumption budget. “One way we will achieve scale is this low-power consumption,” says the project lead, “but another is by having accelerators to provide floating point performance boost to appropriate applications.”

As for topology, the Exanest team is discussing the family of networks that includes fat trees and Dragonfly topology. They will be linking blades through optical fibers that they can plug and unplug allowing them to experiment with more than one topology. Exanest will also be using FPGAs for building the interconnection network so they can experiment with novel protocols.

Exanode

Denis Dutoit, the project coordinator for Exanode, tells HPCwire the goal of that project is to build a node-level prototype with technologies that exhibit exascale potential. The three building blocks are heterogeneous compute elements (ARM-v8 low-power processors plus various accelerators, namely FPGAs although ASICs and GPGPUs may also be explored); 3D interposer integration for compute density; and, continuing the efforts of the EUROSERVER project, an advanced memory scheme for low-latency, high-bandwidth memory access, scalable to exabyte levels.

ExaNoDe_Figures_04-1024x768

Dutoit, who is the strategic marketing manager, architecture, IC design and embedded software division at CEA-Leti, notes that this is a technology driven project at the start, but on top of this prototype, there will be a complete software stack for HPC capability. Evaluation will be done first will be done on the node level, explains Dutoit. They will utilize emulated hardware first and representative HPC applications to evaluate at the level nodes, but after that, Exanest will reuse these compute nodes and integrate them into their complete machine to do the full testing and evaluation with real applications.

There will be a formal effort to productize the resulting technology through a partnership with Kaleao, a UK company that focuses on energy-efficient, compact hyperconverged platforms.

Ecoscale

Iakovos Mavroidis, project coordinator for Ecoscale, says that while there are three main projects, he sees it as one big project with Ecoscale dedicated to reconfigurable computing.

A member of Computer Architecture and VLSI Systems (CARV) Laboratory of FORTH-ICS and a member of Telecommunication Systems Institute, Mavroidis notes that the main problem being addressed is how to improve today’s HPC servers. Simple scaling without improving technologies is unfeasible due to utility costs and power consumption limitations. Ecoscale is tackling these challenges by proposing a scale-out hybrid MPI+OpenCL programming environment and a runtime system, along with a hardware architecture which is tailored to the needs of HPC applications. The programming model and runtime system follows a hierarchical approach where the system is partitioned into multiple autonomous workers (i.e. compute nodes).

ecoscale_framework

“The main focus of Ecoscale is to support shared partitioned reconfigurable resources, accessed by these compute nodes,” says Mavroidis. “The intention is to have a global notion of the reconfigurable resources so that each compute node can access remote reconfigurable resources not only its own local resources. The logic can also be shared by several compute nodes working in parallel.” To accomplish this, workers are interconnected in a tree-like structure in order to form larger Partitioned Global Address Space (PGAS) partitions, which are further hierarchically interconnected via an MPI protocol.

“The virtualization will happen automatically in hardware and it has to be done because reconfigurable resources are very limited unless remote access is enabled,” states Mavroidis. “The aim is to provide a user-friendly way for the programmer to use all the reconfigurable logic in the system. This requires a very high-speed low-latency interconnection topology and this is what Exanest will provide.”

Mavroidis explains there must be means for the programmer to access the system and at a higher-level the run-time system has to be redefined to understand the needs of the application so it can reconfigure the machine. He believes that in order to fully implement this, there will need to be innovation in all the layers of the stack, and also the programming model itself will also need to be redefined. The partners are aiming to support most of the existing and common HPC libraries in order to make this architecture available to most of the existing applications.

The main focus of Ecoscale is to automate out the complexity of FPGA programming. Anyone who has watched FPGAs struggle to get a foothold in HPC knows this is not an easy task, but the need for low-power performance is driving interest and innovation. “The programmer should not have to be aware that the machine uses reconfigurable computing, but rather be able to write the program using high-level programming model such as MPI or Standard C,” states Mavroidis.

On a related note, Exanest project partner BeeGFS has just announced that the BeeGFS parallel file system is now available as open source from www.beegfs.com. “Although BeeGFS can already run out of the box on ARM systems today, this project [Exanest] will give us the opportunity to make sure that we can deliver the maximum performance on this architecture as well,” shares Bernd Lietzow, BeeGFS head for Exanest.

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!

Can Cerabyte Crack the $1-Per-Petabyte Barrier with Ceramic Storage?

July 20, 2024

A German startup named Cerabyte is hoping to solve the burgeoning market for secondary and archival data storage with a novel approach that uses lasers to etch bits onto glass with a ceramic coating. The “grey ceramic� Read more…

Weekly Wire Roundup: July 15-July 19, 2024

July 19, 2024

It's summertime (for most of us), and the HPC-related headlines aren't as plentiful as they once were. But not everything has to happen at high tide-- this week still had some waves! Idaho National Laboratory's Bitter Read more…

ARM, Fujitsu Targeting Open-source Software for Power Efficiency in 2-nm Chip

July 19, 2024

Fujitsu and ARM are relying on open-source software to bring power efficiency to an air-cooled supercomputing chip that will ship in 2027. Monaka chip, which will be made using the 2-nanometer process, is based on the Read more…

SCALEing the CUDA Castle

July 18, 2024

In a previous article, HPCwire has reported on a way in which AMD can get across the CUDA moat that protects the Nvidia CUDA castle (at least for PyTorch AI projects.). Other tools have joined the CUDA castle siege. AMD Read more…

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…

Can Cerabyte Crack the $1-Per-Petabyte Barrier with Ceramic Storage?

July 20, 2024

A German startup named Cerabyte is hoping to solve the burgeoning market for secondary and archival data storage with a novel approach that uses lasers to etch Read more…

SCALEing the CUDA Castle

July 18, 2024

In a previous article, HPCwire has reported on a way in which AMD can get across the CUDA moat that protects the Nvidia CUDA castle (at least for PyTorch AI pro 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…

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…

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…

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…

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…

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…

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…

  • arrow
  • Click Here for More Headlines
  • arrow
HPCwire