Compilers and More: Exascale Programming Requirements

By Michael Wolfe

April 14, 2011

Programming at Exascale, Part 3

In an earlier column, I discussed six levels of parallelism that we’ll have in exascale systems: node, socket, core, vector, instruction, and pipeline levels, and said that to reach exascale performance, we need to take advantage of all these levels, since the final performance is the product of them all. In my most recent column, I argued that to be successful at that, we need to effectively expose, express, and exploit parallelism: expose it in the application and algorithms, express it in the language and program, and exploit it in the generated code and at runtime. Exposing parallelism is mostly a creative task, and thus must be done by humans. Expressing parallelism is where we mostly get sidetracked: what language, what kind of parallelism, how will it work with legacy software? Since parallel programming is all about performance, we need to focus on those aspects that would hinder performance, specifically locality and synchronization. Finally, successfully exploiting parallelism means mapping the parallelism exposed in the application and expressed in the program to the parallelism in the hardware. I discussed five dimensions of flexibility: scalability, dynamic parallelism, composability, load balancing, and productivity. In this column, the last of a three-part series, I’ll give my views on what programming at the exascale level is likely to require, and how we can get there from where we are today. My belief is that it will take some work, but it’s not a wholesale rewrite of 50 years of high performance expertise.

Exascale Programming: What It Won’t Be

What are the characteristics of a programming strategy for the coming exascale computers? It’s easier to say what it isn’t.

It’s not a library. Encapsulation is a well-known, often used, and important technique to building large systems. By design, encapsulation hides information about the implementation of the encapsulated object (data structure, algorithm, service) from the user of that object. Encapsulation will continue to be important for many reasons. But information hiding obscures not just the algorithm and data structures, but performance aspects, such as what kinds of parallelism are used within the encapsulated object and how that interacts with parallelism of the user of that object, or low level information such as how the data is laid out and how that affects locality in an algorithm. In particular, opaque low-level libraries (e.g., MPI for data distribution and message passing) hide too much information from the system, preventing any system-level tuning. That’s not to say a useful system won’t be built using MPI as the transport layer, but MPI or POSIX threads or other low-level libraries should not be directly used in the application.

It’s not a C++ class hierarchy or template library. Here, I’m again going out on a limb; there have been and continue to be many sets of useful C++ class libraries intended to raise the level of application programming. Take the C++ standard template library for vector; the intent of such a template is to allow a user to define a data structure and get the benefit of reusing any routines in the STL or from elsewhere built on the vector template. But you don’t really understand the performance of the vector datatype; that information hiding means you don’t know if accesses to vector V; are efficient or not. Compare that to an array access in a loop, with the corresponding vector access V[i]; the array access can often be optimized down to two instructions: load, and increment the pointer to the next address. Moreover, two-dimensional objects using the vector type (vector>) become even more opaque.

Or take Thrust, an STL-like implementation providing a high-level interface to GPU programming, built on CUDA. You can define two vectors in Thrust as

   using namespace thrust;    device_vector x(1000);    device_vector y(1000); 

Multiplying two such vectors and then accumulating the result can be done as:

   transform( x.begin(), x.end(), y.begin(), z.begin(), multiplies() );    r = reduce( z.begin(), z.end(), 0, plus() );

This is certainly easier (more productive?) to write than the equivalent CUDA C (or CUDA Fortran) code, but it’s still far easier to write the Fortran:

   r = sum( x(:) * y(:) ) 

Moreover, when the constructs are part of the language, the compiler can compose and optimize them together. As mentioned in my last column, in the Fortran case, the compiler can generate code for the multiply then accumulate the result without requiring an intermediate vector result. With the C++ library, the code for the transform method doesn’t know that its result will immediately be accumulated, so the method or (as in this case) the user has to provide a result vector. The only tool the compiler has to optimize class library calls is inlining, and it’s simply not enough to recover the performance lost by the abstraction. There have been some efforts to use run-time code generation, building the expression tree from the method calls, then generating the optimized (and composed) code from the whole expression tree; this was the technology behind Rapidmind, which is now being used in Intel’s Array Building Blocks (ArBB). Such mechanisms are promising, but what we really want is a way to define new data types and describe operations to the compiler in a way that the compiler can reason about them, compose them, reorder them, and so on; currently, the definition is basically in terms of C code, which is not expressive enough. There’s a research project just waiting to happen.

It’s not a domain-specific language. I really like the idea of DSLs, of embedding domain knowledge in the language and using that knowledge when generating and optimizing the code. However, languages, real languages, are big project; DSLs are (by definition) specialized, and hence don’t have a large enough user community to support production, maintenance and continuing development of the language and all the tools needed to support a language. We can’t expect language implementors (like PGI) to take on the development and continuing support of a plethora of languages, any more than we should expect user communities to each design, implement, and then continue to update, enhance, tune and optimize the language implementation with each new processor release from Intel. A possible alternative approach would be to implement a language to support DSLs, supported by a language vendor, including interfacing to debuggers, performance tools, editors, and so on. The various user communities would then be somewhat insulated from the details of a performance-oriented solution, and the vendor would avoid falling into the many-languages trap. There’s another potential research project.

It’s not OpenCL. OpenCL may be a necessary step towards heterogeneous programming, but it’s not the final answer. It’s very low level, “close to the metal”, as even the language designers admit. As with MPI, we may be able to build on OpenCL, but it’s not sufficient.

It’s not a whole new language. New languages have a high barrier to entry; most programmers avoid adopting a new language for fear that it will die, unless the language meets some need better than anything else, or until it has survived along enough to ameliorate the fear. But I think a new language is not called for here. We may benefit from some new features in existing languages, and maybe new ways to make programs in those languages, but most new languages really don’t add semantically much beyond managed memory.

It’s not easy. I’ve argued before that parallel programming is not easy, won’t be, and can’t be made easy. The idea of making parallel programming easy is silly.

It’s not just parallelism. Parallelism is an important aspect, perhaps the dominant aspect, but the key isn’t parallelism, it’s performance. A bad parallel algorithm doesn’t run fast just because it’s parallel. A bad implementation of a good parallel algorithm will also be slow. It’s quite easy to write slow parallel programs; this was the key failure (my opinion) of High Performance Fortran. So our programming mechanism will focus on performance, where parallelism is one aspect (locality and synchronization being two more).

Exascale Programming: What It Is

So what do we want and need when programming at exascale from whatever programming environment we get? Here is my bucket list:

  • It supports all levels of parallelism, from node parallelism down to vector and pipeline parallelism, effectively. Support is a big word here; it has to allow for a programming model that an application developer can use to think about what kinds of parallelism will map well at different levels, that a programmer can use to write a program that can be mapped well at different levels, and that the implementation (compiler and runtime) can use to exploit the parallelism. We have this today, clumsily, with different mechanisms for different levels; a bit more integration would take us a long way.
  • It can map an expression of program parallelism (a parallel loop, say) to different levels of hardware parallelism (across nodes, or to a vector unit) depending on the target. This will make it scalable up and down, from exascale to laptop. There was a great deal of work on the SISAL language to efficiently scalarize an implicitly parallel language, which turned out to be largely the dual of the parallelizing compiler problem. Such work will be part of this parallelism remapping. Remapping node-level parallelism may require changing the data distribution per node; today, this is done at the application level. We should be able to specify what parameters of the program depend on which aspects of the target machine, so the system can do the remapping.
  • It supports the programmer with lots of feedback. Vectorizing compilers have been very successful for over 35 years in delivering good vector performance from sequential loops because the compilers tell the programmer when they are successful, and more importantly, when and why they fail. This is essentially performance feedback. We are in the business of developing high performance applications, and we should be notified when we are using constructs that will restrict our performance. Static feedback and useful dynamic feedback will both be critical.
  • It supports dynamic parallelism, creating parallel tasks and threads when needed. There are many successful and useful implementations of dynamic parallelism, some limited (OpenMP) and some more aggressive (Cilk). Dynamic parallelism is somewhat at odds with locality and synchronization optimization. Using a work-stealing scheduler, an idle worker will steal a work item from the queue of another worker. However, that work item may have been placed on that worker’s queue because that’s where its data is, or because that work item depends on some other work item also assigned to that worker. However, without constructs for dynamic parallelism, we end up micromanaging thread-level parallelism in the constructs we do have.
  • It efficiently composes abstract operations, as I discussed in my previous column; whether these are native to the language, or abstract operations defined by a user or in a library, the implementation must be able to combine them naturally. Perhaps, when we define abstract operations, we need a mechanism to describe how they can compose with others. Many now-standard compiler optimizations fall into composition, such as loop vectorization and loop fusion. We need more investigation about what composing abstract operations means, beyond simply inlining.
  • It is self-balancing and self-tuning. This involves runtime introspection and behavior modification, and means the parameters or data and work distribution must be exposed to the system in order to be modified. Examples include changing the tile sizes for tiled nested loops when optimized for cache locality, or changing the data distributions when the work load is not uniform across the domain. Such behavior modification has been demonstrated in many systems, though not many integrated with the programming language and its implementation.
  • It must be resilient. The big systems are, many believe, going to be in partial failure mode much of the time. This presents challenges for the system manager and programmer. Expecting the entire system to be working, taking checkpoints and restoring from a failure point will not be efficient if failures are the norm. Some of the necessary features must be supported by the hardware (getting data off a node with a failed processor; early failure detection). Other features could be supported by some of the runtime features we develop for other reasons (redistributing data to working nodes; reserving some nodes to serve as online replacements). Such a system can survive and continue beyond many failures.

Most of these points (except for the last) have been researched and implemented in some form already, and could be reproduced with current technology (and enough motivation) in Fortran, C++, or whatever language you want. We have to extend the programming model to expose performance aspects and perhaps resilience aspects, so the user can guide how the system (compiler plus runtime) implements the program. We often get focused on either abstracting away so much that we lose sight of performance (as happened with High Performance Fortran), or we get so tied up with performance that we focus too much on details of each target machine (as happens today with OpenCL and CUDA). We need to let the programmer do the creative parts, and let the system do the mechanical work.

Final Note: This series of columns is an expanded form of the material from the PGI Exhibitor Forum presentation at SC10 in New Orleans. If you were there, you can tell me whether it’s more informative (or entertaining) in written or verbal form.

About the Author

Michael Wolfe has developed compilers for over 30 years in both academia and industry, and is now a senior compiler engineer at The Portland Group, Inc. (www.pgroup.com), a wholly-owned subsidiary of STMicroelectronics, Inc. The opinions stated here are those of the author, and do not represent opinions of The Portland Group, Inc. or STMicroelectronics, Inc.

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!

SC23: The Ethics of Supercomputing

November 29, 2023

Why should HPC practitioners care about ethics? And, what are our ethics in HPC? These questions were central to a lively discussion at the SC23 Birds-of-a-Feather (BoF) session: With Great Power Comes Great Responsib Read more…

Grace Hopper’s Big Debut in AWS Cloud While Gravaton4 Launches

November 29, 2023

Editors Note: Additional Coverage of the AWS-Nvidia 65 Exaflop ‘Ultra-Cluster’ and Gravitron4 can be found on our sister site Datanami. Amazon Web Services will soon be home to a new Nvidia-built supercomputer tha Read more…

Give a Little (on Tuesday), Get a Lot

November 28, 2023

HPC is built on open source. While building HPC systems with "open plumbing" has enormous advantages, there can also be some challenges. As illustrated in the classic XKCD comic, the entire dependency tree of many usefu Read more…

re:Invent 2023: AWS Talks a Little Quantum, Showcases Error Correction Progress

November 28, 2023

Quantum computing held sway in the last few minutes of AWS senior vice president Peter DeSantis’ keynote yesterday at the AWS re:Invent 2023 conference, being held in Las Vegas this week. While scarce on details, DeSan Read more…

Analyst Panel Says Take the Quantum Computing Plunge Now…

November 27, 2023

Should you start exploring quantum computing? Yes, said a panel of analysts convened at Tabor Communications HPC and AI on Wall Street conference earlier this year. Without doubt, the quantum computing landscape remai Read more…

AWS Solution Channel

Deploying AI/ML at the Edge with Omniflow’s Sustainable Smart Lamppost, NVIDIA, and AWS

Imagine a world where a lamppost does more than just illuminate streets; it actively contributes to a smarter, safer, and more sustainable community. Using Amazon Web Services (AWS) and NVIDIA technologies, Omniflow is turning this vision into a reality. Read more…

QCT Solution Channel

QCT and Intel Codeveloped QCT DevCloud Program to Jumpstart HPC and AI Development

Organizations and developers face a variety of issues in developing and testing HPC and AI applications. Challenges they face can range from simply having access to a wide variety of hardware, frameworks, and toolkits to time spent on installation, development, testing, and troubleshooting which can lead to increases in cost. Read more…

SC23 HPC Student Cluster Smackdown

November 21, 2023

Since 2007, the Student Cluster Competition (SCC) has provided an international multi-day contest for the best and brightest university HPC teams. This year, the in-person event was held at SC23 in Denver from November 1 Read more…

SC23: The Ethics of Supercomputing

November 29, 2023

Why should HPC practitioners care about ethics? And, what are our ethics in HPC? These questions were central to a lively discussion at the SC23 Birds-of-a-Fe Read more…

Grace Hopper’s Big Debut in AWS Cloud While Gravaton4 Launches

November 29, 2023

Editors Note: Additional Coverage of the AWS-Nvidia 65 Exaflop ‘Ultra-Cluster’ and Gravitron4 can be found on our sister site Datanami. Amazon Web Servic Read more…

Analyst Panel Says Take the Quantum Computing Plunge Now…

November 27, 2023

Should you start exploring quantum computing? Yes, said a panel of analysts convened at Tabor Communications HPC and AI on Wall Street conference earlier this y Read more…

SCREAM wins Gordon Bell Climate Prize at SC23

November 21, 2023

The first Gordon Bell Prize for Climate Modeling was presented at SC23 in Denver. The award went to a team led by Sandia National Laboratories that had develope Read more…

SC23 BOF: Inclusivity Progress and Challenges

November 21, 2023

New to SC23 was a series of talks on Inclusivity topics. Sponsored by the Inclusivity Committee and open to all conference attendees, these 90-minute birds-of-a Read more…

Supercomputing 2023: Odds and Ends from the Show

November 20, 2023

This year's fantastic Supercomputing 2023 was back in full form. Attendees seemed to be glad that the show was back in Denver, which was a preferred destination Read more…

Material Simulation with Quantum Accuracy Wins 2023 ACM Gordon Bell Prize

November 20, 2023

Accurately calculating interactions among electrons has been a significant obstacle to reliable material exploration and design through computer modeling. Recen Read more…

Shutterstock 1086444218

HPC Hardware Contracts: Backlash as Security Ignored in Performance Pursuit

November 16, 2023

The security of supercomputers has been grossly ignored in the pursuit of horsepower. Still, there is a growing realization that security is needed to prevent b Read more…

CORNELL I-WAY DEMONSTRATION PITS PARASITE AGAINST VICTIM

October 6, 1995

Ithaca, NY --Visitors to this year's Supercomputing '95 (SC'95) conference will witness a life-and-death struggle between parasite and victim, using virtual Read more…

SGI POWERS VIRTUAL OPERATING ROOM USED IN SURGEON TRAINING

October 6, 1995

Surgery simulations to date have largely been created through the development of dedicated applications requiring considerable programming and computer graphi Read more…

U.S. Will Relax Export Restrictions on Supercomputers

October 6, 1995

New York, NY -- U.S. President Bill Clinton has announced that he will definitely relax restrictions on exports of high-performance computers, giving a boost Read more…

Dutch HPC Center Will Have 20 GFlop, 76-Node SP2 Online by 1996

October 6, 1995

Amsterdam, the Netherlands -- SARA, (Stichting Academisch Rekencentrum Amsterdam), Academic Computing Services of Amsterdam recently announced that it has pur Read more…

Cray Delivers J916 Compact Supercomputer to Solvay Chemical

October 6, 1995

Eagan, Minn. -- Cray Research Inc. has delivered a Cray J916 low-cost compact supercomputer and Cray's UniChem client/server computational chemistry software Read more…

NEC Laboratory Reviews First Year of Cooperative Projects

October 6, 1995

Sankt Augustin, Germany -- NEC C&C (Computers and Communication) Research Laboratory at the GMD Technopark has wrapped up its first year of operation. Read more…

Sun and Sybase Say SQL Server 11 Benchmarks at 4544.60 tpmC

October 6, 1995

Mountain View, Calif. -- Sun Microsystems, Inc. and Sybase, Inc. recently announced the first benchmark results for SQL Server 11. The result represents a n Read more…

New Study Says Parallel Processing Market Will Reach $14B in 1999

October 6, 1995

Mountain View, Calif. -- A study by the Palo Alto Management Group (PAMG) indicates the market for parallel processing systems will increase at more than 4 Read more…

Leading Solution Providers

Contributors

SC23 Booth Videos

AMD @ SC23
AWS @ SC23
Altair @ SC23
CoolIT @ SC23
Cornelis Networks @ SC23
CoreHive @ SC23
DDC @ SC23
HPE @ SC23 with Justin Hotard
HPE @ SC23 with Trish Damkroger
Intel @ SC23
Intelligent Light @ SC23
Lenovo @ SC23
Penguin Solutions @ SC23
QCT Intel @ SC23
Tyan AMD @ SC23
Tyan Intel @ SC23
HPCwire LIVE from SC23 Playlist

CORNELL I-WAY DEMONSTRATION PITS PARASITE AGAINST VICTIM

October 6, 1995

Ithaca, NY --Visitors to this year's Supercomputing '95 (SC'95) conference will witness a life-and-death struggle between parasite and victim, using virtual Read more…

SGI POWERS VIRTUAL OPERATING ROOM USED IN SURGEON TRAINING

October 6, 1995

Surgery simulations to date have largely been created through the development of dedicated applications requiring considerable programming and computer graphi Read more…

U.S. Will Relax Export Restrictions on Supercomputers

October 6, 1995

New York, NY -- U.S. President Bill Clinton has announced that he will definitely relax restrictions on exports of high-performance computers, giving a boost Read more…

Dutch HPC Center Will Have 20 GFlop, 76-Node SP2 Online by 1996

October 6, 1995

Amsterdam, the Netherlands -- SARA, (Stichting Academisch Rekencentrum Amsterdam), Academic Computing Services of Amsterdam recently announced that it has pur Read more…

Cray Delivers J916 Compact Supercomputer to Solvay Chemical

October 6, 1995

Eagan, Minn. -- Cray Research Inc. has delivered a Cray J916 low-cost compact supercomputer and Cray's UniChem client/server computational chemistry software Read more…

NEC Laboratory Reviews First Year of Cooperative Projects

October 6, 1995

Sankt Augustin, Germany -- NEC C&C (Computers and Communication) Research Laboratory at the GMD Technopark has wrapped up its first year of operation. Read more…

Sun and Sybase Say SQL Server 11 Benchmarks at 4544.60 tpmC

October 6, 1995

Mountain View, Calif. -- Sun Microsystems, Inc. and Sybase, Inc. recently announced the first benchmark results for SQL Server 11. The result represents a n Read more…

New Study Says Parallel Processing Market Will Reach $14B in 1999

October 6, 1995

Mountain View, Calif. -- A study by the Palo Alto Management Group (PAMG) indicates the market for parallel processing systems will increase at more than 4 Read more…

  • arrow
  • Click Here for More Headlines
  • arrow
HPCwire