Sign up for our newsletter and get the latest HPC news and analysis.
Send me information from insideHPC:


Intel HPC Technology: Fueling Discovery and Insight with a Common Foundation

This guest post explores how Intel HPC technology and the company’s scalable and efficient HPC tech portfolio accelerates today’s diverse workloads. 

Intel HPC Technology

To remain competitive, companies, academic institutions, and government agencies must tap the data available to them to empower scientific breakthroughs and drive greater business agility. Many of these organizations already use HPC systems to support common workloads like simulation and data-intensive modeling, and analytics systems which mine for meaningful and actionable insights. Addressing the demands of ever-growing HPC scenarios —  including AI-related workloads — requires HPC infrastructure to keep pace.

As IT budgets tighten, HPC professionals face a lot of pressure to reap the greatest benefit from their HPC investments. To derive the greatest return on their capital expenditures they must think holistically about their chosen HPC solution. If one HPC system ingredient creates a bottleneck, overall system performance will suffer. Wherever possible, HPC professionals should maximize their existing infrastructure investments, and plan system upgrades with future-proofed HPC technologies as stakeholders’ projects require it.

Compute infrastructure must meet big-data challenges that address an organization’s needs today, but also have the flexibility to scale up to increasingly complex workloads in the future.

A modern HPC implementation must meet big-data challenges now and have the flexibility to scale up as increasingly-complex hybrid workloads place greater demands on HPC systems. For example, one step of a research project may involve simulation, while a second step requires deep learning capability. In a hybrid scenario like this, a balanced multi-node system can excel at both data-intensive workloads like AI, plus compute-intensive activities like modeling. Intel’s portfolio of technologies offer a common infrastructure foundation across processors, networking, storage, fabric, memory, and software to support demanding workloads like these.

Intel Xeon Scalable processors offer up to 28 cores on tap, six memory channels, and 48 PCIe lanes for enhanced performance and increases in throughput. Combined with HPC advances like Intel Advanced Vector Extensions (Intel AVX 512), Intel Xeon Scalable processors deliver up to 1.73x HPC performance compared to prior generations.1 Intel AVX-512 extensions are optimized for HPC, data analytics, security, and cryptography workloads and offer up to double the number of FLOPS per clock cycle compared to AVX2.2

A modern HPC implementation must meet big-data challenges now and have the flexibility to scale up as increasingly-complex hybrid workloads place greater demands on HPC systems.

Intel Omni-Path Architecture (Intel OPA) provides a low latency interconnect for multi-node environments. Intel OPA supports 100Gb/s while reducing the number of switches needed, thereby lowering overall fabric costs.3

Intel Field Programmable Gate Arrays (Intel FPGAs) complement Intel Xeon processor-based platforms, offering inherent reconfigurability and flexible acceleration for low-latency applications such as data analytics and AI.

Intel Optane SSDs feature Intel’s new 3D XPoint technology that combines memory and storage functionality to speed data access.

Intel 3D NAND SSDs offer higher performance, reliability, and support for increased densities.4 Together, these characteristics make Intel 3D NAND SSDs a very effective replacement for older hard disk drive-based storage solutions.

These technologies, combined with Intel’s ecosystem of ISVs, OEMs and system providers offer a solid foundation for HPC now, and a growth path for scale-up as end-user needs dictate it.

To find out more about Intel HPC technology to accelerate your research and business insights, visit Intel.com/hpc where you will find solution briefs, educational resources, case studies and more.

1 Performance improvements of up to 1.73x for HPC workloads when compared to the previous generation. Up to 1.73x claim based on LAMMPS: LAMMPS is a classical molecular dynamics code and an acronym for Large-scale Atomic/Molecular Massively Parallel Simulator. It is used to simulate the movement of atoms to develop better therapeutics, improve alternative energy devices, develop new materials, and more. E5-2697 v4: 2S Intel® Xeon® processor E5-2697 v4, 2.3GHz, 36 cores, Intel® Turbo Boost Technology and Intel® Hyperthreading Technology on, BIOS 86B0271.R00, 8x16GB 2400MHz DDR4, Red Hat Enterprise Linux* 7.2 kernel 3.10.0-327. Gold 6148: 2S Intel® Xeon® Gold 6148 processor, 2.4GHz, 40 cores, Intel® Turbo Boost Technology and Intel® Hyperthreading Technology on, BIOS 86B.01.00.0412.R00, 12x16GB 2666MHz DDR4, Red Hat Enterprise Linux* 7.2 kernel 3.10.0-327.

2 As measured by Intel® Xeon® Processor Scalable Family with Intel® AVX-512 compared to an Intel® Xeon® E5 v4 with Intel® AVX2.2

3 Configurations: Intel® Omni-Path Architecture: Configuration assumes a 750-node cluster, and number of switch chips required is based on a full bisectional bandwidth (FBB) Fat-Tree configuration.  Intel® OPA uses one fully-populated 768-port director switch, and Mellanox EDR solution uses a combination of director switches and edge switches.  Includes hardware acquisition costs (server and fabric), 24×7 3-year support (Mellanox Gold support), and 3-year power and cooling costs.  Mellanox component pricing from www.kernelsoftware.com, with prices as of August 15, 2017.  Intel® OPA pricing is also from www.kernelsoftware.com on August 15, 2017.  Mellanox power data based on Mellanox CS7500 Director Switch, Mellanox SB7700/SB7790 Edge switch, and Mellanox ConnectX-5 VPI adapter card product briefs posted on www.mellanox.com as of August 15, 2017.  Intel OPA power data based on product briefs posted on www.intel.com as of April 4, 2017.  Power and cooling costs based on $0.1071 per kWh, and assumes server power costs and server cooling cost are equal and additive. * Other names and brands may be claimed as property of others.

4 Capacity. Comparing Intel® SSD DC P3520 Series 2TB with Intel® SSD DC P4510 Series 8TB drive. These are representative highest density offerings in 2017.

 

 

Leave a Comment

*

Resource Links: