SLATE Update: Making Math Libraries Exascale-ready

By John Russell

August 9, 2018

Practically-speaking, achieving exascale computing requires enabling HPC software to effectively use accelerators – mostly GPUs at present – and that remains something of a challenge. Consider Summit, the U.S. supercomputer at ORNL, which captured the top spot on the Top500 list in June. Summit has 4,356 nodes, each with two IBM 22-core Power9 CPUs and six Nvidia Tesla V100 GPUs. It’s the GPUs that provide most of the performance speedup, and math libraries, in particular, must be able to take advantage of them to speed up HPC applications.

The SLATE project – Software for Linear Algebra Targeting Exascale – is intended to help solve the accelerator-readiness problem. Last week the U.S. Exascale Computing Project (ECP) posted a video interview with Jakub Kurzak, co-PI on SLATE, updating progress. It’s brief, breezy and worth watching given how foundational math libraries are for HPC applications. SLATE is intended to replace the 20-plus year-old Scalable Linear Algebra PACKage (ScaLAPACK) library, currently the industry standard for dense linear algebra operations in distributed memory environments.

“The main motivation for rewriting ScaLAPACK [is] it is very hard to imagine an accelerated ScaLAPACK,” says Kurzak. “If you look at where HPC is going, if you look at the big machine here, Summit, you see immediately the need. To put a number on it, something like 98 percent of the Summit’s performance is in its GPUs.” If codes are not GPU-accelerated, “you won’t reach exascale,” he says.

As described on the SLATE website:

“SLATE aims to extract the full performance potential and maximum scalability from modern, many-node HPC machines with large numbers of cores and multiple hardware accelerators per node. For typical dense linear algebra workloads, this means getting close to the theoretical peak performance and scaling to the full size of the machine (i.e., thousands to tens of thousands of nodes). This is to be accomplished in a portable manner by relying on standards like MPI and OpenMP.

“SLATE functionalities will first be delivered to the ECP applications that most urgently require SLATE capabilities (e.g., EXascale Atomistics with Accuracy, Length, and Time [EXAALT], NorthWest computational Chemistry for Exascale [NWChemEx], Quantum Monte Carlo PACKage [QMCPACK], General Atomic and Molecular Electronic Structure System [GAMESS], CANcer Distributed Learning Environment [CANDLE]) and to other software libraries that rely on underlying dense linear algebra services (e.g., Factorization Based Sparse Solvers and Preconditioners [FBSS]). SLATE will also fill the void left by ScaLAPACK’s inability to utilize hardware accelerators, and it will ease the difficulties associated with ScaLAPACK’s legacy matrix layout and Fortran API.”

These are ambitious goals. Kurzak and co-PI Jack Dongarra, both of the University of Tennessee’s Innovative Computing Laboratory (ICL), lead a group of roughly eight researchers dedicated to the ECP project. In the video, Kurzak is interviewed by Mike Bernhardt, ECP communications manager, and they discuss what’s been accomplished, what’s expected in the next year or so, and some of the challenges.

 

Presented here, slightly edited, are a few of Kurzak’s comments.

“We’ve spent a lot of time laying out the foundations making sure the architecture is solid. In terms of functionality we haven’t released all that much, but we have released some routines for basic linear algebra operations. If you want to multiply to really large matrices right now and get GPU acceleration, SLATE has these kinds of routines. We [also] released a batch of matrix norms routines. Now we’re working on a really exciting batch of routines for solving linear systems. I think our user base should explode when we release the linear solvers at the end of this quarter,” he says.

“[By] the end of 2019 SLATE should be a solid replacement for ScaLAPACK. At least for the most important parts of ScaLAPACK. It should offer a viable replacement for GPU acceleration. That being said we designed the package to be much more flexible than ScaLAPACK so we should be able to go way beyond [its] capabilities as we go beyond 2019. There’s a lot of exciting things I think we can do algorithmically in SLATE and cater to many more applications in terms of what kinds of problems we can solve, what sizes, what types of matrices.”

Kurzak notes SLATE is the first major project at ICL to be implemented in C++. “That’s a bit barrier to adoption initially, but I have to say it’s been a blessing [because] I think the choice of the C++ language, the shift from C, is probably going to be one of the key technologies that will contribute to SLATE’s success.”

Perhaps not surprisingly, recruitment and retention are among SLATE’s most difficult challenges.

“You want somebody that does know C++ well, somebody who definitely knows MPI, and oh yes knows multithreading too, and yes, knows GPU programming too, and yes, knows linear algebra. That is a long list of requirements. The assumption is we’ll hire somebody who does not know everything but will pick it up on the job. Nevertheless the barrier to entry is pretty high.”

Interestingly, enthusiasm is the number one factor he is looking for.

Link to ECP post: https://www.exascaleproject.org/video-highlight-ecps-slate-project-aims-to-provide-basic-dense-matrix-operations/

Link SLATE site: https://www.exascaleproject.org/project/slate-software-linear-algebra-targeting-exascale/

Link to SLATE poster: https://www.exascaleproject.org/wp-content/uploads/2018/01/ECP-Meeting-Poster-SLATE.pdf

Link to video: https://www.youtube.com/watch?v=wS5aPAcaNbY

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!

Nvidia Debuts Turing Architecture, Focusing on Real-Time Ray Tracing

August 16, 2018

From the SIGGRAPH professional graphics conference in Vancouver this week, Nvidia CEO Jensen Huang unveiled Turing, the company's next-gen GPU platform that introduces new RT Cores to accelerate ray tracing and new Tenso Read more…

By Tiffany Trader

HPC Coding: The Power of L(o)osing Control

August 16, 2018

Exascale roadmaps, exascale projects and exascale lobbyists ask, on-again-off-again, for a fundamental rewrite of major code building blocks. Otherwise, so they claim, codes will not scale up. Naturally, some exascale pr Read more…

By Tobias Weinzierl

STAQ(ing) the Quantum Computing Deck

August 16, 2018

Quantum computers – at least for now – remain noisy. That’s another way of saying unreliable and in diverse ways that often depend on the specific quantum technology used. One idea is to mitigate noisiness and perh Read more…

By John Russell

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

Super Problem Solving

You might think that tackling the world’s toughest problems is a job only for superheroes, but at special places such as the Oak Ridge National Laboratory, supercomputers are the real heroes. Read more…

NREL ‘Eagle’ Supercomputer to Advance Energy Tech R&D

August 14, 2018

The U.S. Department of Energy (DOE) National Renewable Energy Laboratory (NREL) has contracted with Hewlett Packard Enterprise (HPE) for a new 8-petaflops (peak) supercomputer that will be used to advance early-stage R&a Read more…

By Tiffany Trader

STAQ(ing) the Quantum Computing Deck

August 16, 2018

Quantum computers – at least for now – remain noisy. That’s another way of saying unreliable and in diverse ways that often depend on the specific quantum Read more…

By John Russell

NREL ‘Eagle’ Supercomputer to Advance Energy Tech R&D

August 14, 2018

The U.S. Department of Energy (DOE) National Renewable Energy Laboratory (NREL) has contracted with Hewlett Packard Enterprise (HPE) for a new 8-petaflops (peak Read more…

By Tiffany Trader

CERN Project Sees Orders-of-Magnitude Speedup with AI Approach

August 14, 2018

An award-winning effort at CERN has demonstrated potential to significantly change how the physics based modeling and simulation communities view machine learni Read more…

By Rob Farber

Intel Announces Cooper Lake, Advances AI Strategy

August 9, 2018

Intel's chief datacenter exec Navin Shenoy kicked off the company's Data-Centric Innovation Summit Wednesday, the day-long program devoted to Intel's datacenter Read more…

By Tiffany Trader

SLATE Update: Making Math Libraries Exascale-ready

August 9, 2018

Practically-speaking, achieving exascale computing requires enabling HPC software to effectively use accelerators – mostly GPUs at present – and that remain Read more…

By John Russell

Summertime in Washington: Some Unexpected Advanced Computing News

August 8, 2018

Summertime in Washington DC is known for its heat and humidity. That is why most people get away to either the mountains or the seashore and things slow down. H Read more…

By Alex R. Larzelere

NSF Invests $15 Million in Quantum STAQ

August 7, 2018

Quantum computing development is in full ascent as global backers aim to transcend the limitations of classical computing by leveraging the magical-seeming prop Read more…

By Tiffany Trader

By the Numbers: Cray Would Like Exascale to Be the Icing on the Cake

August 1, 2018

On its earnings call held for investors yesterday, Cray gave an accounting for its latest quarterly financials, offered future guidance and provided an update o Read more…

By Tiffany Trader

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