Cray XT5m Midrange Supercomputer Builds Market Momentum

By Nicole Hemsoth

September 28, 2009

In March 2009, Cray announced the Cray XT5m system, a compatible midrange extension of the high-end Cray XT5 product line that was first to break the sustained petaflop performance barrier on real-world applications. HPCwire asked Barry Bolding, Cray’s vice president of scalable systems, for an update on the midrange product.

Cray XT5HPCwire: How is the Cray XT5m midrange supercomputer doing in the market place?

Bolding: We disclosed one customer name with the March announcement, the High Performance Computing Center Stuttgart, where it’s being used for automotive, academic and public sector work. We have also publically announced a second customer, the Finnish Meteorological Institute (FMI) in the production weather forecasting segment. The Cray XT5m has been meeting our sales goals since then. Granted, our expectations for the first six months were fairly modest because the midrange is a new space for us and we viewed this as a learning period. The good news is that we have validated that the XT5m can be a leader in the mid-range. We have won customers in several of our key target segments and our pricing is very competitive in this price sensitive space so we expect higher volumes going forward. Today we have a large number of bids outstanding in academia, automotive, aerospace, pharma and weather forecasting, as well as in government and R&D labs.

HPCwire: Why did Cray decide to develop this product?

Bolding: There were several reasons. We were consistently running into procurements that were in the midrange and needed a more price-competitive system for that space, which we view as a growth area for us. According to IDC, Cray is the current global market share leader for HPC systems priced at $3 million and up, but we have a very small presence in the midrange segment. Many of the characteristics of our largest machines are applicable in the midrange, but to succeed in this space we needed to develop a price-competitive product, turn around procurements faster, and provide a wider range of ISV applications. ISV applications are also important at the high end, but industry typically buys midrange systems and relies more heavily on third-party apps than government and academia do. The Cray XT5m is the initial product on our midrange roadmap.

HPCwire: Why would someone buy a Cray XT5m system instead of a same-size cluster? What are the key differences?

Bolding: Primarily they get Cray’s more capable SeaStar interconnect and interconnect roadmap. The Cray SeaStar has proven itself over the past several years as the industry-leading interconnect for MPI scalability.  With the Cray SeaStar interconnect, the Cray XT5m handles complex messaging traffic very efficiently.  You also get the entire Cray software stack that has been scaled and validated up to the petascale performance level, plus the network roadmap driving toward global addressability and high-performance UPC and Co-Array Fortran, along with MPI of course. So, Cray XT5m customers are buying into a broader portfolio than just a typical midrange system.  They’re buying into the demonstrated petascale scalability of the XT5 architecture. To achieve true scaling today, even at midrange size, one needs a full portfolio of network, software and infrastructure support, which Cray provides in the XT5m. And with multicore processors becoming prevalent, users will soon be facing the need to scale substantially higher even with midrange systems, especially starting in 2010.  Our midrange systems are designed to benefit from Cray’s high-end system development and this will continue. For example, the Cray XT5m line provides the same TCO benefits to our midrange customers as our largest supercomputer customers enjoy with the ECOphlex cooling technology we developed for petascale systems.

HPCwire: Assuming the Cray XT5m has a more capable network and memory subsystem than a standard cluster, with better bandwidth and latency characteristics, wouldn’t it be able to tackle a broader range of applications efficiently?

Bolding: The Cray XT5m today is a midrange industry leader from a network bandwidth and latency perspective. This gives users an important advantage over commodity InfiniBand networks and allows the Cray XT5m to handle a broader range of applications efficiently. The system is aimed at codes scaling to 256 cores and beyond, and in this range users typically see significant benefits from the overall system and software design.

HPCwire: Are customers using their Cray XT5m systems as their main HPC systems or for specific portions of their workloads?

Bolding: It varies. In production weather forecasting, we have a customer using it as their primary production system. At academic sites it varies from being primary system to one among multiple HPC systems. Some users are experimenting with the new functionalities, including the network features and the Cray software and compilers. It’s also allowing users to experiment with scalability in ways they haven’t been able to do before. So, the Cray XT5m is being used both as a development platform and a production platform.

HPCwire: Is there any customer who operates both a Cray XT5 and a Cray XT5m system?

Bolding: There are customers running applications across both Cray XT5 and Cray  XT5m systems, although they don’t have both systems in-house today.  By the end of this year, we’ll have sites that have both in their data centers.

HPCwire: The XT5m is a compatible downward extension of the Cray XT5 architecture, but uses a 2D torus interconnect instead of the XT5’s 3D torus. What’s the strategy behind this change?

Bolding: It’s a cost-saving strategy for the midrange scale. With a Cray XT5m system consisting of 1-6 cabinets, customers tend not to have applications that require the full 3D topology as much as with a larger, high-end Cray XT5. We right-sized the Cray XT5m for midrange requirements, including the interconnect, allowing us to provide a price-competitive product in this space. We’ve done extensive studies on application performance on the Cray XT5m, and there has been minimal performance impact at six cabinets and below. Above that size, you need the Cray XT5’s 3D torus to maintain scalability. For most apps in the 1-6 cabinet range, performance degradation due to the topology is less than 5 percent even for applications running across several thousand cores of a Cray XT5m.

HPCwire: If a site maxes out on their XT5m, what’s the upgrade path to a Cray XT5 system?

Bolding: It’s very simple. Today, it just involves replacing the network mezzanine card and adding more cables to transform the 2D torus into a 3D torus. We’ll make the upgrade path even simpler in the future of our midrange systems.

HPCwire: Do you expect some users to take advantage of non-MPI programming models that are available on the XT5m, such as SHMEM, UPC and Co-Array Fortran?

Bolding: We do. We port those in the software today and will be making announcements of enhancements to the hardware support for some of these features in the next 12 months. We are committed to making more innovations and remaining a leader in HPC, and this requires providing our customers with multiple, high performance programming models.

HPCwire: What is the Data Virtualization Service that comes with the XT5 and XT5m, and why is it important?

Bolding: DVS is an important part of Cray systems. DVS is a flexible virtualization layer that Cray plans use to expand our software functionality and performance . One feature of DVS is that it can allow Cray to project various file systems onto the compute nodes (which are diskless on Cray XT5 systems). This allows Cray systems to act more like a standard commodity cluster if it needs to.  We support IO and storage functionalities that we haven’t in the past. We can share file systems with high-performance file systems on platforms other than clusters. Customers such as NERSC and Oak Ridge are doing very innovative things with file systems and DVS can play a role in providing the flexibility they need.  So, DVS helps us both with compute and IO/storage.

HPCwire: Where is the Cray XT5m product line headed in the future?

Bolding: We are going to continue driving into the midrange market, which is a segment that has excellent growth opportunities for Cray. We want to build a substantial and sustainable market presence there, especially with customers focused on scalability of applications. To grow the XT5m line we need to continue to be competitive on cost while improving price/performance, network scalability,  software features and ISV availability, and we’ll be doing all those things. We’ll also be improving the processor roadmap with new technologies coming into market place. Cray’s innovation, combined with AMD’s strong roadmap is a winning combination for the next few years for the entire XT family of systems. 
HPCwire: What should we look for next?

Bolding: At SC09, we plan to talk more specifically about our future plans for the Cray XT5m product line.

For more information on how the Cray XT5m is making petaflops performance affordable, download the AMD white paper here.

 

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!

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…

Nvidia Appoints Andy Grant as EMEA Director of Supercomputing, Higher Education, and AI

March 22, 2024

Nvidia recently appointed Andy Grant as Director, Supercomputing, Higher Education, and AI for Europe, the Middle East, and Africa (EMEA). With over 25 years of high-performance computing (HPC) experience, Grant brings a 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…

Houston We Have a Solution: Addressing the HPC and Tech Talent Gap

March 15, 2024

Generations of Houstonian teachers, counselors, and parents have either worked in the aerospace industry or know people who do - the prospect of entering the fi 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