Parallel Programming Is Here – Are You Ready?

By Nicole Hemsoth

June 23, 2009

Whether you’re simulating the extreme conditions inside an exploding star or designing an ergonomically innovative office chair, it’s a good bet that a high performance computing (HPC) system and some brain-bending programming will be involved.

The HPC system may be a supercomputer like the 1.6 petaflop Jaguar behemoth at Oak Ridge National Laboratory, or a cluster powered by off-the-shelf multicore components. Whatever the scale of the hardware and the scope of the application, developers will have to learn how to deal with the complexities of parallel programming to get the most out of their computational resources.

The need for parallel programming is being driven by advances in multicore architectures. This rapid and accelerating technology trend is creating an array of HPC systems that range from dual and quad core systems to supercomputers and clusters with tens, hundreds and thousands of cores. These platforms perform at teraflop and petaflop speeds on terabytes of data. Capable of tackling some of today’s most complex and pressing problems in engineering and science, these HPC systems are composed of a computational ecosystem that includes: scalable multicore architectures; fast, flexible, mammoth memories that can support many simultaneous threads; and high bandwidth I/O and communications.

Developers who have honed their parallel programming skills are ready to create applications that reach new levels of scalability, performance, safety and reliability. In particular, parallelism can be exploited in mechanical computer-aided engineering (MCAE) applications code for structural analysis and fluid dynamics, in computational chemistry and computational physics simulations and modeling, and industrial applications that run the gamut from oil and gas exploration to the design of high end golf equipment. For example, in the world of MCAE, Dale Layfield, engineer in Sun Microsystem’s ISV Engineering organization, points to the benefits realized by applying parallelization to NASTRAN, a venerable finite element analysis (FEA) program that has been around for about 40 years.

“NASTRAN is a highly compute and I/O intensive structural analysis program,” explains Layfield. “It lends itself well to being broken into smaller components and spreading those components across distributed computer clusters which substantially reduces throughput time. Distributed memory parallelism (DMP) helps eliminate the I/O bottleneck by dividing the analysis across a network of separate nodes. Multithreaded SMP (symmetric multiprocessing) allows you to make best use of the processing power within each node. SMP combined with DMP gives you the most bang for your buck.”

Like NASTRAN, many of the other complex applications designed to run on HPC systems rely on parallel programming methodologies to handle the increasing number of computationally intensive jobs involving massive amounts of data and memory.

As David Conover, Chief Technologist, Mechanical Products for ANSYS notes, “Among the major benefits of parallel programming are faster turnaround time and the ability to create higher fidelity simulations and modeling to solve engineering design challenges. Engineers applying finite element methods can create models with much higher spatial resolutions and more geometric detail. And they can build models that include entire assemblies, rather than just one small component. Then they can analyze the interactions between those components at a high level of detail. Because the users are able to perform more simultaneous tasks of increased complexity, the entire engineering process is far more productive. You just can’t achieve this level of functionality with applications that rely on sequential processes.”

By creating larger high fidelity models with greater geometric detail and subjecting them to detailed simulations of the physical forces that they will encounter in real life, engineers can reduce the need for expensive and time-consuming physical testing — the “build and break” approach. In addition, parallelization allows engineers to run more simulations in order to make design decisions earlier in the project lifecycle.

To achieve the speedup in applications performance, parallel programming uses threads that allow multiple operations to occur simultaneously. In an article in the May 20, 2009 HPCwire titled,“Parallel Programming: Some Fundamentals Concepts,” authors Shameen Akhter and Jason Roberts, both of Intel, commented, “The entire concept of parallel programming centers on the design, development and deployment of threads within an application and the coordination between threads and their respective operations.”

In short, parallel programming allows you to write scalable, flexible code that harnesses more HPC CPU resources and maximizes memory and I/O. It also allows users of the code — whether it’s you, a member of your organization’s engineering or scientific staff, or a customer – to solve problems that could not be solved using sequential programs, and solve them more quickly.

Parallel programming is not easy

However, as computer science professor Andrew S. Tanenbaum stated at the USENIX ’08 conference, “Sequential programming is really hard…the difficulty is that parallel programming is a step beyond that.”

Bronson Messer, a computational astrophysicist at Oak Ridge National Laboratory (ORNL), concurs. He points out that to do computing at the large scales he and his colleagues encounter daily, the application developer needs to understand the entire HPC ecosystem which includes multicore CPUs, high speed file and connective systems, and terabytes of memory that have to be swapped in and out at blinding speeds.

“Everything has to play together,” Messer says. “If there is a weak link at this scale, it will almost immediately be exposed. Your parallel code may run on a quad-core or eight-core system, but when you move up to thousands and tens of thousands of processors, your application may be dead in the water. Debugging code on this many processors is an unsolved problem.”

Messer also comments that building robustness and fault tolerance into the code is another major hurdle as the rate of data collection escalates. For example, the Sloan Digital Sky Survey telescope in Sun Spot, New Mexico is precisely mapping a swath of space some five billion light years in diameter, generating terabytes, even petabytes of data every night. And when CERN’s Large Hadron Collider finally comes on line, it will generate 700 megabytes of data every second.

These parallel programming speed bumps not only apply to code written for the huge supercomputers that are the workhorses of government labs and academia. Developers creating algorithms for the rapidly growing population of HPC grids, clusters and clouds that are infiltrating the enterprise are running into similar problems. And within industry the pressure is even more intense as companies seek to gain a competitive edge through the use of HPC.

When asked what he thought was the most difficult task facing developers working with this new programming paradigm, Scott J. Lasica, VP Technical Services Worldwide for HPC toolmaker Rogue Wave Software, was very clear. “Today’s developers need to learn to do multithreading, which, in my opinion, is one of the hardest — if not the hardest — task associated with software programming. Given the level of complexity we’re dealing with, it’s very easy to make mistakes and very hard to figure out where things went wrong.”

What’s a developer to do?

Lasica points out that fortunately there are a lot of tools available to help developers write multithreaded code in languages like C++ and Java — even Fortran. For example, a Java(TM) application can be dropped into an application server and the server will take care of the threading. Various new debugging tools also help ease the bumpy road to parallelization. But Lasica says that a thorough grounding in the intricacies of multithreading is essential for developers dealing with today’s complex distributed systems.

Reza Sadeghi, CTO of MSC Software agrees. And he also prescribes a major mind shift for today’s developers. “Developers tend to think serially, not in terms of what they can do with multiple CPUs,” he explains. “And even if they are thinking parallel, they are still in the realm of dual, quad or eight cores. But the new HPC systems are raising the bar to encompass hundreds and thousands of cores as well as multicore sub architectures. It’s a whole new way of building algorithms and solving complex loops. By adopting this different mindset, backed up by learning all you can about parallelism and multithreading, you can make optimum use of the many diagnostic tools that are now available and build successful HPC applications.”

Advanced programming models also help ease the developer’s path. Among the most popular are OpenMP for shared memory programming, and MPI (message passing interface) for distributed memory programming.

ORNL’s Messer adds that given the rapid pace of technology, it is important for developers to create algorithms that will scale far beyond their current systems. “If you know apriori that your algorithm won’t scale, you have an immediate problem,” he says. “With today’s multicore HPC systems, you are dealing with a deeper and more complicated memory hierarchy in addition to the problems inherent in multithreading. Despite advances in OS, compilers and programming models, you still may have to manage some of that hierarchy yourself. The results are worth it.”

Continuing education is key

Addison Snell, general manager of Tabor Research, comments that developers need to familiarize themselves with how to optimize software on multicore HPC systems. “I’m not sure the latest generation of software engineers has been trained to cope with advanced parallelism – there is a serious question of readiness in the software community,” he says.

It is certain that as the world of high performance computing heats up, and multicore, multithreaded systems move into the enterprise, those individuals who are familiar with parallel programming will command a favorable position in today’s rough and tumble job market. Application developers should be very familiar with the principles of parallel programming, including how to handle multithreading. They should also be acquainted with parallel tools, and be able to build thread-safe component interfaces. Also, both test engineers and field engineers should have parallel debugging skills and be familiar with parallel analysis and profiling tools.

In order to help developers and engineers meet the challenges posed by parallel programming, Sun Microsystems is offering a series of seminars called “An Introduction to Parallel Programming” discussing parallel programming as a fundamental of application development. Log on weekly to access each of these seven modules presented by mathematician and Sun senior staff engineer Ruud van der Pas. http://www.sun.com/solutions/hpc/development.jsp.

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