The Taming of the GPU

By Gabor Samu

June 20, 2018

The media has been alight with articles regarding the groundbreaking Summit supercomputer recently unveiled at Oak Ridge National Laboratory. It sports a mind boggling 9,216 IBM POWER9 CPUs, 27,648 NVIDIA Tesla GPUs, underpinned with 250 petabytes of storage. This muscle will be put to good use running traditional HPC as well as AI workloads across a broad range of sciences.

Looking at the landscape of systems being built for HPC and now AI, there is one commonality – many are hybrid CPU-GPU systems. Whether we’re considering systems at the pinnacle of computing such as Summit, or commercial HPC and AI systems, GPUs have become a defacto method for accelerating code and providing copious amounts of floating point performance.

The early days of clustered computing saw the advent of workload and resource managers which were a means of taming environments by orchestrating access to, and bringing computing resources to bear, in a predictable manner – aligned with the needs of scientists and businesses alike. As environments have grown in scale to meet the growing thirst for HPC, GPUs and accelerated computing have stepped out on stage to take a bow.

Software developers have and continue to port and optimize applications to benefit from the capabilities provided by GPUs. According to a recent report from November 2017, a high percentage of HPC applications now offer GPU support.

“According to the latest HPC User Site Census data and additional research, of the 50 most popular application packages mentioned by HPC users, 34 offer GPU support (including two under current development), including 9 of the top 10.”

Indeed, the recent Top 500 list (November 2017) includes no less than 87 hybrid CPU-GPU systems (and more counting other types of accelerators).

So how do GPU-heavy systems impact the task of the workload and resource managers? Fundamentally, as GPUs are resources, workload schedulers have had to adapt too.

A wild west land grab

It’s not just large-scale supercomputers that face the challenges of compute supply versus user demands.  Commercial HPC environments are also now increasingly hybrid CPU-GPU based with potentially hundreds of users and millions of jobs per day in high-throughput computing use cases. These are complex environments and large investments requiring workload management software with sophisticated capabilities to reign in all the resources – so that users end up with GPU workloads running on the right servers.

Computing environments today can have some servers with GPUs, some without, varied GPU configurations including models and memory, and a different number of GPUs per node. Adding to this complexity, in a typical data center, servers can come and go so the configuration is not always static.

In general, workload schedulers require the administrator to specify in the configuration whether a given server is equipped with GPUs, often requiring additional information such as the GPU model, etc. Without this crucial information, the workload scheduler cannot effectively route jobs to nodes – potentially leading to a Wild West grab for resources.

Call in the Cavalry

IBM Spectrum LSF has been continuously innovating to address the needs of increasingly complex HPC environments of scale since 1992. Support for NVIDIA GPUs was first introduced in IBM Spectrum LSF in 2007 (figure 1). Continuing this long tradition of enhancements to NVIDIA GPU support, IBM Spectrum LSF now includes a new capability designed to dramatically simplify the administration of GPU servers and enables users to be more productive faster. With “zero config” for NVIDIA GPUs, IBM Spectrum LSF detects the presence of GPUs and automatically performs the necessary scheduler configuration – without any interaction from the administrator. IBM Spectrum LSF will help tame the GPU environment for you, allowing users with GPU ready codes to be productive from the moment the environment is setup.

IBM Resources

Follow @IBMSystems

IBM Systems on Facebook

Subscribe to HPCwire's Weekly Update!

Be the most informed person in the room! Stay ahead of the tech trends with industy updates delivered to you every week!

InfiniBand Still Tops in Supercomputing

July 19, 2018

In the competitive global HPC landscape, system and processor vendors, nations and end user sites certainly get a lot of attention--deservedly so--but more than ever, the network plays a crucial role. While fast, perform Read more…

By Tiffany Trader

HPC for Life: Genomics, Brain Research, and Beyond

July 19, 2018

During the past few decades, the life sciences have witnessed one landmark discovery after another with the aid of HPC, paving the way toward a new era of personalized treatments based on an individual’s genetic makeup Read more…

By Warren Froelich

WCRP’s New Strategic Plan for Climate Research Highlights the Importance of HPC

July 19, 2018

As climate modeling increasingly leverages exascale computing and researchers warn of an impending computing gap in climate research, the World Climate Research Programme (WCRP) is developing its new Strategic Plan – and high-performance computing is slated to play a critical role. Read more…

By Oliver Peckham

HPE Extreme Performance Solutions

Introducing the First Integrated System Management Software for HPC Clusters from HPE

How do you manage your complex, growing cluster environments? Answer that big challenge with the new HPC cluster management solution: HPE Performance Cluster Manager. Read more…

IBM Accelerated Insights

Are Your Software Licenses Impeding Your Productivity?

In my previous article, Improving chip yield rates with cognitive manufacturing, I highlighted the costs associated with semiconductor manufacturing, and how cognitive methods can yield benefits in both design and manufacture.  Read more…

U.S. Exascale Computing Project Releases Software Technology Progress Report

July 19, 2018

As is often noted the race to exascale computing isn’t just about hardware. This week the U.S. Exascale Computing Project (ECP) released its latest Software Technology (ST) Capability Assessment Report detailing progress so far. Read more…

By John Russell

InfiniBand Still Tops in Supercomputing

July 19, 2018

In the competitive global HPC landscape, system and processor vendors, nations and end user sites certainly get a lot of attention--deservedly so--but more than Read more…

By Tiffany Trader

HPC for Life: Genomics, Brain Research, and Beyond

July 19, 2018

During the past few decades, the life sciences have witnessed one landmark discovery after another with the aid of HPC, paving the way toward a new era of perso Read more…

By Warren Froelich

D-Wave Breaks New Ground in Quantum Simulation

July 16, 2018

Last Friday D-Wave scientists and colleagues published work in Science which they say represents the first fulfillment of Richard Feynman’s 1982 notion that Read more…

By John Russell

AI Thought Leaders on Capitol Hill

July 14, 2018

On Thursday, July 12, the House Committee on Science, Space, and Technology heard from four academic and industry leaders – representatives from Berkeley Lab, Argonne Lab, GE Global Research and Carnegie Mellon University – on the opportunities springing from the intersection of machine learning and advanced-scale computing. Read more…

By Tiffany Trader

HPC Serves as a ‘Rosetta Stone’ for the Information Age

July 12, 2018

In an age defined and transformed by its data, several large-scale scientific instruments around the globe might be viewed as a ‘mother lode’ of precious data. With names seemingly created for a ‘techno-speak’ glossary, these interferometers, cyclotrons, sequencers, solenoids, satellite altimeters, and cryo-electron microscopes are churning out data in previously unthinkable and seemingly incomprehensible quantities -- billions, trillions and quadrillions of bits and bytes of electro-magnetic code. Read more…

By Warren Froelich

Tsinghua Powers Through ISC18 Field

July 10, 2018

Tsinghua University topped all other competitors at the ISC18 Student Cluster Competition with an overall score of 88.43 out of 100. This gives Tsinghua their s Read more…

By Dan Olds

HPE, EPFL Launch Blue Brain 5 Supercomputer

July 10, 2018

HPE and the Ecole Polytechnique Federale de Lausannne (EPFL) Blue Brain Project yesterday introduced Blue Brain 5, a new supercomputer built by HPE, which displ Read more…

By John Russell

Pumping New Life into HPC Clusters, the Case for Liquid Cooling

July 10, 2018

High Performance Computing (HPC) faces some daunting challenges in the coming years as traditional, industry-standard systems push the boundaries of data center Read more…

By Scott Tease

Leading Solution Providers

SC17 Booth Video Tours Playlist

Altair @ SC17

Altair

AMD @ SC17

AMD

ASRock Rack @ SC17

ASRock Rack

CEJN @ SC17

CEJN

DDN Storage @ SC17

DDN Storage

Huawei @ SC17

Huawei

IBM @ SC17

IBM

IBM Power Systems @ SC17

IBM Power Systems

Intel @ SC17

Intel

Lenovo @ SC17

Lenovo

Mellanox Technologies @ SC17

Mellanox Technologies

Microsoft @ SC17

Microsoft

Penguin Computing @ SC17

Penguin Computing

Pure Storage @ SC17

Pure Storage

Supericro @ SC17

Supericro

Tyan @ SC17

Tyan

Univa @ SC17

Univa

  • arrow
  • Click Here for More Headlines
  • arrow
Do NOT follow this link or you will be banned from the site!
Share This