Intel Speeds NAMD by 1.8x: Saves Xeon Processor Users Millions of Compute Hours

By Rob Farber

August 12, 2020

Potentially saving datacenters millions of CPU node hours, Intel and the University of Illinois at Urbana–Champaign (UIUC) have collaborated to develop AVX-512 optimizations for the NAMD scalable molecular dynamics code. These optimizations will be incorporated into release 2.15 with patches available for earlier versions. Users of the Frontera supercomputer at the Texas Advanced Computing Center (TACC) are already using the NAMD patch to help accelerate research into the pathology and treatment of COVID-19.[i]

Benchmarks show that Intel’s optimized version of this workhorse application can run up to 1.8x faster on Intel Xeon Scalable processors compared to the non-optimized version.  Note that the 1.8x is coming from the Intel Xeon Gold 6148 results (specifically 0.66 compared to 1.17). The performance when combining both the software optimization and hardware performance of newer processors such as an Intel Xeon Platinum 9282 results in NAMD running 3x faster.

Figure 1: Speedups on the NAMD million atom STMV (Satellite Tobacco Mosaic Virus) benchmark.

Considered a reference by the NAMD scientific community, the million atom STMV (Satellite Tobacco Mosaic Virus) benchmark is indicative of the relative performance improvement scientists can expect for their own simulations. A 1.8x speedup in the NAMD code means that scientists can potentially obtain their molecular modeling results 1.8x faster or have their runs tie up 1.8x fewer nodes.

David Hardy is a senior research programmer in the Theoretical and Computational Biophysics Group of the Beckman Institute at the University of Illinois at Urbana-Champaign and is the lead developer of NAMD. He says, “NAMD development continues to benefit from a long-standing relationship with Intel, providing critical insights into current and upcoming technology, together with software engineering expertise and code contributions to improve performance.”

Optimizing a top HPC application

NAMD (short for Nanoscale Molecular Dynamics) is a widely utilized application in the HPC space. [ii] Researchers use NAMD to simulate detailed atomic scale models that can illuminate molecular structural details of how a virus works and what aspects of its structure offer potential targets for vaccines and treatments.

John Cazes (director of High Performance Computing, TACC) sees a tremendous benefit for NAMD users at TACC: “From Stampede to Stampede2 and now on Frontera, NAMD has been a major tool for scientists using TACC resources. We track application usage regularly and NAMD is always in the top 5, if not number one, each quarter. The Intel and UIUC improvements to NAMD will benefit our users on both Stampede2 and Frontera. We look forward to updating our installs when version 2.15 with AVX-512 support is released.”

Million node hour savings scales to large systems

Intel believes the significant performance boost from the optimized NAMD code will allow researchers to achieve longer timescales in the simulation of relevant molecules associated with disease, and by extension enable them to better understand aspects of infection with atom-level detail.

More specifically, the Intel optimized code speeds the computation of non-bonded atomic operations, which consume most of the NAMD runtime. The optimization uses a “tile” algorithm [iii] that increases the computational efficiency of each computational node by exploiting the Intel AVX-512 vector units and larger cache of the Intel Xeon Scalable processors. The tile algorithm does not affect NAMD’s scaling behavior across nodes, so researchers essentially get a 1.8x more powerful supercomputer simply by using the optimized code on large simulations.

Benchmarks confirm the speedup potential. Shown below is the strong scaling of NAMD on the TACC Frontera supercomputer running two different large synthetic benchmark systems, each assembled from “gluing” the STMV periodic cube (1,066,628 atoms) into an array of STMV to produce a bigger periodic system. The smaller system consists of a 5x2x2 array of 20 STMV (totaling about 21M atoms), and the larger system consists of a 7x6x5 array of 210 STMV (totaling about 224M atoms). Shown in Figure 2 are the performance results of the three different versions of NAMD (AVX2, AVX-512, and AVX-512 tiles, as compared in Figure 1) running the two synthetic benchmark systems, confirming the speedup available from the latest AVX-512 Tiles optimization and demonstrating NAMD’s scalability on Frontera for large simulations. The three NAMD versions use the Charm++ runtime system with the UCX communication layer, all built with the Intel 2019 C++ compiler.

Figure 2: Scaling results for the three different versions of NAMD (AVX2, AVX-512, and AVX-512 tiles) running two different benchmark systems (20 STMV and 210 STMV) on the TACC Frontera supercomputer. These results confirm the speedup available from the latest AVX-512 Tiles optimization.

Understanding the impact

To understand the impact of the optimized NAMD code, consider the efforts of a research team from the University of California San Diego (UCSD) led by Rommie Amaro (professor of Chemistry and Biochemistry at UCSD).

TACC reports that Amaro and her team have already used about 2.3 million Frontera node hours for molecular dynamics simulations and modeling, the most among any researchers using the system to study COVID-19. [iv]

The project provides an example of the magnitude of compute resources involved in a single research study. According to the Intel benchmarks, the optimizations have the potential for million-node hour savings for just this one project. Mike Brown, principal engineer in HPC at Intel Corporation, adds, “The savings are cumulative. When added with the many research projects in molecular biophysics running NAMD on AVX-512 capable clusters, both at TACC and other institutions, the impact to science can be very significant.”

The societal impact of such large-scale molecular studies can be huge. Amaro is a corresponding author of another recent study in which simulations on the National Science Foundation (NSF)-funded Frontera supercomputer reveal surprising information about the atomic makeup of the coronavirus’s exterior shield. This shell is the part of the virus we are trying to destroy when we wash our hands for 20 seconds. Contact by rubbing with soap destroys this viral shield.

Once inside the body, this shell camouflages the virus so it can hide itself from defending antibodies in our immune system.

The study, published June 12, 2020 discovered that shield’s sugary coating of molecules, called glycans, also prime the coronavirus to infect a cell by changing the shape of an internal spike protein.

Amaro noted, “That was really surprising to see. It’s one of the major results of our study. It suggests that the role of glycans in this case is going beyond shielding to potentially having these chemical groups actually being involved in the dynamics of the spike protein,” she said.

She likened the action of the glycan to pulling the trigger of a gun. “When that bit of the spike goes up, the finger is on the trigger of the infection machinery.” Amaro said. “When it gets like that, all it has to do is come up against an ACE2 receptor in the human cell, and then it’s going to bind super tightly and the cell is basically infected.”

Figure 3: The sugary coating of molecules called glycans (deep blue) that shield the SARS-CoV-2 spike from detection by the human immune system. (Credit: Lorenzo Casalino (UCSD) et al.)

Visualizing the model dynamics provides a wealth of information

Scientists have to study and visualize the dynamics of these atomic level models so they can understand what the computer model is telling them.

The key word is “visualize” as scientists also require powerful visual capabilities that can display the tremendous amount of data generated by optimized models running on world class supercomputers. Molecular dynamics simulations performed with NAMD (and other packages) are all about depicting the atoms of the system in motion. However, it’s not just a question of modeling the system, it’s also about visualizing the dynamics. Both visualization and simulation are closely related.

Amaro describes the dynamics of a good viral model as a “computational microscope” that lets scientists study “the wiggles and jiggles of the atoms”. [v] The latter reference pays homage to Physicist Richard Feynman who famously said, “Everything that living things do can be understood in terms of the jiggling and wiggling of atoms.”[vi]

To give others a sense of the jiggly nature of the atoms in these simulations, Amaro’s team created a moving visualization of the coronavirus spiked-protein, which can give readers a sense. This visualization is based on research they performed at TACC.

VMD: A visualization tool for NAMD users

Intel’s NAMD optimization work dovetails nicely with their open source SDVis (Software Defined Visualization) initiative with various industry collaborators which allows scientists to visualize huge data sets using CPUs – even on exascale supercomputers. [vii] Many major visualization tools such as ParaView, VisIt, and VMD (Visual Molecular Dynamics) have adopted the Intel SDVis libraries which include Embree, OSPRay, and OpenSWR.

For leadership research such as the coronavirus envelope simulations, the SDVis solutions are the right tool at the right time as they were designed to handle extremely large, data intensive visualizations. VMD in particular is pertinent as it was designed to be used in conjunction with NAMD, for the setup and analysis of the atomic scale systems,[viii] and has a demonstrated ability to render large models comprising hundreds of millions of atoms. [ix]

CPU-based rendering is becoming the norm in the HPC community. At TACC for example, Paul Navrátil (Director of Visualization, TACC) points out that, “CPU-based SDVis will be our primary visual analysis mode on Frontera.” Looking to the future, please see the article “Preparing for the Arrival of Aurora with CPU-based Interactive Visualization.” For general information on SDVis, please consult sdvis.org.

Summary

Intel NAMD optimizations offer significant improvements on Intel Xeon Scalable processor workload that will free millions of node hours of time for additional research. Benchmarks show that the optimized version with AVX-512 support can run up to 1.8x faster on Intel Xeon Scalable processors compared to the non-optimized version. The impact of these optimizations is cumulative.

UIUC and Intel continue to collaborate on software optimizations and algorithms to best exploit leading supercomputers, including Aurora, the planned U.S. exascale supercomputer to be located at Argonne National Laboratory. There is hope that NAMD will get even faster, as Mike Brown notes the CPU optimization work for NAMD is not done.

Besides NAMD, other Intel optimization efforts include LAMMPS (a molecular dynamics code), GROMACS (a package for protein, lipids, and nucleic acids simulation), AMBER (another molecular dynamics code), and others [x] to facilitate scientific innovation on the largest CPU-based supercomputers.

About the Author

Rob Farber is a global technology consultant and author with an extensive background in HPC, AI, and teaching. Rob can be reached at info@techenablement.com

[i] While NAMD 2.13 does include optimizations for vector computing (in part from previous collaboration between Intel and UIUC), the CPU algorithm that considers each interaction between pairs of atoms does not make the best use of AVX-512 vector units.

[ii] Intersect360

[iii] https://link.springer.com/chapter/10.1007%2F978-3-319-46079-6_14

[iv] https://www.hpcwire.com/2020/06/16/researchers-use-frontera-to-investigate-covid-19s-insidious-sugar-coating/

[v] https://beta.nsf.gov/science-matters/once-considered-too-high-risk-supercomputer-simulations-wiggling-and-jiggling-atoms

[vi] American Scientist 

[vii] https://dl.acm.org/doi/10.5555/3293524.3293526

[viii] https://en.wikipedia.org/wiki/NAMD

[ix] http://www.ks.uiuc.edu/Research/vmd/

[x] https://venturebeat.com/2020/06/08/covid-19-hpc-consortiums-supercomputing-effort-is-helping-scientists-better-understand-covid-19/

 

 

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!

Rockport Networks Launches 300 Gbps Switchless Fabric, Reveals 396-Node Deployment at TACC

October 27, 2021

Rockport Networks emerged from stealth this week with the launch of its 300 Gbps switchless networking architecture focused on the needs of the high-performance computing and the advanced-scale AI market. Early customers Read more…

AWS Adds Gaudi-Powered, ML-Optimized EC2 DL1 Instances, Now in GA

October 27, 2021

As machine learning becomes a dominating use case for local and cloud computing, companies are racing to provide solutions specifically optimized and accelerated for AI applications. Now, Amazon Web Services (AWS) is int Read more…

Fireside Chat with LBNL’s Advanced Quantum Testbed Director

October 26, 2021

Last week, Irfan Siddiqi led a “fireside chat” with a few media and analysts to introduce the Department of Energy’s relatively new Advanced Quantum Testbed (AQT), which is based at Lawrence Berkeley National Labor Read more…

Graphcore Introduces Larger-Than-Ever IPU-Based Pods

October 22, 2021

After launching its second-generation intelligence processing units (IPUs) in 2020, four years after emerging from stealth, Graphcore is now boosting its product line with its largest commercially-available IPU-based sys Read more…

Quantum Chemistry Project to Be Among the First on EuroHPC’s LUMI System

October 22, 2021

Finland’s CSC has just installed the first module of LUMI, a 550-peak petaflops system supported by the European Union’s EuroHPC Joint Undertaking. While LUMI -- pictured in the header -- isn’t slated to complete i Read more…

AWS Solution Channel

Royalty-free stock illustration ID: 577238446

Putting bitrates into perspective

Recently, we talked about the advances NICE DCV has made to push pixels from cloud-hosted desktops or applications over the internet even more efficiently than before. Read more…

Killer Instinct: AMD’s Multi-Chip MI200 GPU Readies for a Major Global Debut

October 21, 2021

AMD’s next-generation supercomputer GPU is on its way – and by all appearances, it’s about to make a name for itself. The AMD Radeon Instinct MI200 GPU (a successor to the MI100) will, over the next year, begin to power three massive systems on three continents: the United States’ exascale Frontier system; the European Union’s pre-exascale LUMI system; and Australia’s petascale Setonix system. Read more…

Rockport Networks Launches 300 Gbps Switchless Fabric, Reveals 396-Node Deployment at TACC

October 27, 2021

Rockport Networks emerged from stealth this week with the launch of its 300 Gbps switchless networking architecture focused on the needs of the high-performance Read more…

AWS Adds Gaudi-Powered, ML-Optimized EC2 DL1 Instances, Now in GA

October 27, 2021

As machine learning becomes a dominating use case for local and cloud computing, companies are racing to provide solutions specifically optimized and accelerate Read more…

Fireside Chat with LBNL’s Advanced Quantum Testbed Director

October 26, 2021

Last week, Irfan Siddiqi led a “fireside chat” with a few media and analysts to introduce the Department of Energy’s relatively new Advanced Quantum Testb Read more…

Killer Instinct: AMD’s Multi-Chip MI200 GPU Readies for a Major Global Debut

October 21, 2021

AMD’s next-generation supercomputer GPU is on its way – and by all appearances, it’s about to make a name for itself. The AMD Radeon Instinct MI200 GPU (a successor to the MI100) will, over the next year, begin to power three massive systems on three continents: the United States’ exascale Frontier system; the European Union’s pre-exascale LUMI system; and Australia’s petascale Setonix system. Read more…

D-Wave Embraces Gate-Based Quantum Computing; Charts Path Forward

October 21, 2021

Earlier this month D-Wave Systems, the quantum computing pioneer that has long championed quantum annealing-based quantum computing (and sometimes taken heat fo Read more…

LLNL Prepares the Water and Power Infrastructure for El Capitan

October 21, 2021

When it’s (ostensibly) ready in early 2023, El Capitan is expected to deliver in excess of two exaflops of peak computing power – around four times the powe Read more…

Intel Reorgs HPC Group, Creates Two ‘Super Compute’ Groups

October 15, 2021

Following on changes made in June that moved Intel’s HPC unit out of the Data Platform Group and into the newly created Accelerated Computing Systems and Graphics (AXG) business unit, led by Raja Koduri, Intel is making further updates to the HPC group and announcing... Read more…

Quantum Workforce – NSTC Report Highlights Need for International Talent

October 13, 2021

Attracting and training the needed quantum workforce to fuel the ongoing quantum information sciences (QIS) revolution is a hot topic these days. Last week, the U.S. National Science and Technology Council issued a report – The Role of International Talent in Quantum Information Science... Read more…

Enter Dojo: Tesla Reveals Design for Modular Supercomputer & D1 Chip

August 20, 2021

Two months ago, Tesla revealed a massive GPU cluster that it said was “roughly the number five supercomputer in the world,” and which was just a precursor to Tesla’s real supercomputing moonshot: the long-rumored, little-detailed Dojo system. Read more…

Esperanto, Silicon in Hand, Champions the Efficiency of Its 1,092-Core RISC-V Chip

August 27, 2021

Esperanto Technologies made waves last December when it announced ET-SoC-1, a new RISC-V-based chip aimed at machine learning that packed nearly 1,100 cores onto a package small enough to fit six times over on a single PCIe card. Now, Esperanto is back, silicon in-hand and taking aim... Read more…

US Closes in on Exascale: Frontier Installation Is Underway

September 29, 2021

At the Advanced Scientific Computing Advisory Committee (ASCAC) meeting, held by Zoom this week (Sept. 29-30), it was revealed that the Frontier supercomputer is currently being installed at Oak Ridge National Laboratory in Oak Ridge, Tenn. The staff at the Oak Ridge Leadership... Read more…

Intel Reorgs HPC Group, Creates Two ‘Super Compute’ Groups

October 15, 2021

Following on changes made in June that moved Intel’s HPC unit out of the Data Platform Group and into the newly created Accelerated Computing Systems and Graphics (AXG) business unit, led by Raja Koduri, Intel is making further updates to the HPC group and announcing... Read more…

Ahead of ‘Dojo,’ Tesla Reveals Its Massive Precursor Supercomputer

June 22, 2021

In spring 2019, Tesla made cryptic reference to a project called Dojo, a “super-powerful training computer” for video data processing. Then, in summer 2020, Tesla CEO Elon Musk tweeted: “Tesla is developing a [neural network] training computer... Read more…

Intel Completes LLVM Adoption; Will End Updates to Classic C/C++ Compilers in Future

August 10, 2021

Intel reported in a blog this week that its adoption of the open source LLVM architecture for Intel’s C/C++ compiler is complete. The transition is part of In Read more…

Hot Chips: Here Come the DPUs and IPUs from Arm, Nvidia and Intel

August 25, 2021

The emergence of data processing units (DPU) and infrastructure processing units (IPU) as potentially important pieces in cloud and datacenter architectures was Read more…

AMD-Xilinx Deal Gains UK, EU Approvals — China’s Decision Still Pending

July 1, 2021

AMD’s planned acquisition of FPGA maker Xilinx is now in the hands of Chinese regulators after needed antitrust approvals for the $35 billion deal were receiv Read more…

Leading Solution Providers

Contributors

Intel Unveils New Node Names; Sapphire Rapids Is Now an ‘Intel 7’ CPU

July 27, 2021

What's a preeminent chip company to do when its process node technology lags the competition by (roughly) one generation, but outmoded naming conventions make i Read more…

HPE Wins $2B GreenLake HPC-as-a-Service Deal with NSA

September 1, 2021

In the heated, oft-contentious, government IT space, HPE has won a massive $2 billion contract to provide HPC and AI services to the United States’ National Security Agency (NSA). Following on the heels of the now-canceled $10 billion JEDI contract (reissued as JWCC) and a $10 billion... Read more…

Quantum Roundup: IBM, Rigetti, Phasecraft, Oxford QC, China, and More

July 13, 2021

IBM yesterday announced a proof for a quantum ML algorithm. A week ago, it unveiled a new topology for its quantum processors. Last Friday, the Technical Univer Read more…

The Latest MLPerf Inference Results: Nvidia GPUs Hold Sway but Here Come CPUs and Intel

September 22, 2021

The latest round of MLPerf inference benchmark (v 1.1) results was released today and Nvidia again dominated, sweeping the top spots in the closed (apples-to-ap Read more…

10nm, 7nm, 5nm…. Should the Chip Nanometer Metric Be Replaced?

June 1, 2020

The biggest cool factor in server chips is the nanometer. AMD beating Intel to a CPU built on a 7nm process node* – with 5nm and 3nm on the way – has been i Read more…

Julia Update: Adoption Keeps Climbing; Is It a Python Challenger?

January 13, 2021

The rapid adoption of Julia, the open source, high level programing language with roots at MIT, shows no sign of slowing according to data from Julialang.org. I Read more…

Frontier to Meet 20MW Exascale Power Target Set by DARPA in 2008

July 14, 2021

After more than a decade of planning, the United States’ first exascale computer, Frontier, is set to arrive at Oak Ridge National Laboratory (ORNL) later this year. Crossing this “1,000x” horizon required overcoming four major challenges: power demand, reliability, extreme parallelism and data movement. Read more…

D-Wave Embraces Gate-Based Quantum Computing; Charts Path Forward

October 21, 2021

Earlier this month D-Wave Systems, the quantum computing pioneer that has long championed quantum annealing-based quantum computing (and sometimes taken heat fo Read more…

  • arrow
  • Click Here for More Headlines
  • arrow
HPCwire