CPU March Madness Ends with Intel Nehalem EX Launch

By Michael Feldman

March 31, 2010

In what has become one of the busiest months ever for CPU introductions, Intel got its final say on Tuesday with the launch of its much-anticipated Nehalem EX. The new processor line encompasses the Xeon 7500 and 6500 series and will be the basis for shared memory SMP systems from dozens of server makers, including HPC stalwarts like Cray, SGI and Bull.

Nehalem EX’s biggest claim to fame is its 8-core footprint (a first for Intel), although 6-core and 4-core variants are also available. Since the new chip is intended for the “expandable” server segment, platforms from 2 to 8 sockets are supported natively, with the 4-socket server being the sweet spot. Custom implementations can expand the CPU count much further, however. SGI’s Altix UV 1000, for example, can aggregate as many as to 256 CPUs into a single cache coherent NUMA system.

Up to 16 memory slots can be attached per Nehalem EX socket for a maximum memory capacity of 1 TB on a 4-socket server. Four memory channels per socket should keep RAM access humming along nicely for most data-intensive applications. (If a better bandwidth-to-compute ratio needed, it’s probably worth looking into the 6-core and 4-core variants.) In any case, the large 24 MB of L3 cache should help data access in general, although, to be fair, only the top-bin parts provide that much L3. All the others come with either 18 MB or 12 MB.

For traditional enterprise applications, Nehalem EX is being positioned to bring mission-critical computing to the mainstream. Business intelligence, ERP, OLTP, and basically any heavy-duty enterprise application that fits into a virtualized server environment is fair game. Thanks to the impressive core count, memory support and new RAS features of the new chip, Intel is aiming is to grab market share from the RISC CPU-based SMP platforms that own much of this mission-critical space today.

For HPC, the EX silicon is positioned to support technical applications that can benefit from large amounts of shared memory, high CPU counts, or both. Here we’re talking about memory-intensive or I/O-intensive applications like CAE, EDA, life science simulations, seismic modeling codes — really any HPC workload that relies on large datasets and can benefit from a shared memory model. In a scale-out cluster environment, SMP servers also have a place since they reduce the number of nodes, and thus simplify cluster management, network infrastructure, and software costs.

While the Xeon 7500 series is the mainstream product line for Nehalem EX, the 6500 series is a 2-socket-only offshoot that was designed specifically for the HPC market. As such, the 6500 seems to be positioned halfway between the 7500 and the Xeon dual-socket CPUs, the 5500 and 5600 (Nehalem and Westmere EP, respectively). Basically the 6500 is aimed at lower cost dual-socket servers for applications that need larger memory footprints and higher bandwidth than EP-based platforms can muster.

For the time being, though, all the HPC platforms announced in conjunction with the EX launch are using 7500 silicon. Here’s a rundown of two of the more interesting systems introduced by HPC vendors this week.

SGI Adds Mini-SMP to Altix UV Stable

SGI’s use of Nehalem EX in its Altix UV (Ultraviolet) line, was well documented here in HPCwire and elsewhere, when the machines were previewed back in November 2009. The mid-range UV 100 (up to 96 CPU sockets and 6 TB of memory) and the top-of-the line UV 1000 (up to 256 sockets and 16 TB of memory) are based on SGI’s NUMAlink interconnect and custom chipset that extends the new CPU’s SMP capabilities significantly beyond its natural 8-socket limit. As such, the UV 100 and 1000 represent the largest x86-based SMP systems on the market today.

SGI decided to use the official Nehalem EX launch this week to introduce an entry-level Ultraviolet, the UV 10. It’s a 4-socket rackmount server with a top core count of 32 and maximum memory capacity of 512 GB (or 1 TB if you want to pay for outrageously expensive 16 GB DIMMs). Although, the new system is much smaller than the UV 100 and 1000 models, it comes with a lot of I/O expansion capacity. Ten PCIe I/O expansion slots are available, aimed principally at storage and networking cards. ”In terms of a quad-socket server, this is one of the highest-end options available in the marketplace,” said Geoffrey Noer, SGI’s senior director of product marketing

The UV10 doesn’t use the special chipset or NUMAlink interconnect of its larger siblings, but is still able to run the same software stack. The cut-down UV can be used as an application development machine for a larger UV production system or as a dedicated machine for bite-sized shared memory applications. SGI is also positioning the UV 10 as a service node for large conventional clusters, and is intending to sell them as such with the Altix ICE line. A list price of $33,250 is quoted for a mid-range configuration: 4 Xeon X7542 processors (6-core, 2.66 GHz), 32 GB of memory, and a SATA boot drive.

HLRN (Germany), CALMIP (France), the Institute of Low Temperature Science, Hokkaido (Japan), and the University of Tennessee (US) have UV 100 or 1000 machines on order, with the first systems slated to begin shipping in the second quarter of the year. The UV 10 systems are available immediately.

Cray and Bull Team Up on SMP Offering

Nehalem EX has also spurred Bull and Cray to jump on the SMP bandwagon. In this case though, it’s a two-for-one deal. Both vendors are using the same Bull-designed hardware for these systems and adding their own branding, software stacks, and custom support on top. The Cray-branded CX1000-S is under its new line of entry-level and midrange HPC machines that the company introduced last week, while the Bull supernodes fit into the bullx product line under its Extreme Computing portfolio announced last year. According to Ian Miller, senior vice president of the productivity solutions group and marketing at Cray, they very much liked the Bull technology. And since each company has more or less staked out different geographical territories — Bull in much of Europe and Cray everywhere else — the partnership just made sense to everyone involved.

The CX1000-S and bullx supernodes scale to 16 sockets (128 cores) and 1 TB of memory. Since 16 sockets is beyond the 8-processor reach of the Nehalem design, a custom-built node extension technology is employed. The implementation is somewhat unconventional though. Instead of an integrated node controller, the engineers opted for a custom “coherency switch” that aggregates up to four quad-socket Nehalem EX-based servers into one.

To back up a little, there are actually two variations of these machines: a compute node and a management node. In the Cray stable, the product set is split into the CX1000-SC (compute) and the CX1000-SM (management) models. The matching offerings from Bull are the S6010 and S6030.

The compute node is an L-shaped 1.5U box that houses up to four Nehalem EX chips hooked together via the native QPI links. Multiple boxes can be jigsawed together to construct larger SMP nodes (up to a maximum 6U, 16-socket configuration) via the aforementioned coherency switch.

The management node is a 3U box that comes with six PCIe slots, and in that sense is similar to the I/O-expandable UV 10. And like the SGI offering, this machine is meant to be used as a service node for a cluster or as a standalone compute node for apps that need lots of I/O capacity. It can also be expanded into larger SMP configurations via the magic coherency switch.

The SMP building block nodes will be shipping soon, but the coherency switch hardware won’t be available until later in the second quarter. Cray has quoted a starting price of under $100,000 for the CX1000 line in general. Pricing on the bullx supernodes is not public.

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