Difference between revisions of "HPX-5"

From Modelado Foundation Wiki
Jump to: navigation, search
 
Line 2: Line 2:
 
== Announcement ==
 
== Announcement ==
  
'''''HPX-5 version 2.0 Runtime System Release Announced by Indiana University!'''''
+
'''''Indiana University announces HPX-5 version 4.0 runtime system software!'''''
  
The Center for Research in Extreme Scale Technologies (CREST) at Indiana University is pleased to announce the release of version 2.0 of the HPX-5 runtime system for petascale/exascale computing. Building on CREST’s commitment to developing new approaches for achieving the highest levels of performance on current and next-generation supercomputing platforms, HPX-5 is provided to support the international high-performance computing community in addressing significant challenges involved in achieving exascale computing.
+
The Center for Research in Extreme Scale Technologies (CREST) at Indiana University is pleased to announce the release of version 4.0 of HPX-5, a state-of-the-art runtime system for extreme-scale computing. Version 4.0 of the HPX-5 runtime system represents a significant maturation of the sequence of HPX-5 releases to date for efficient scalable general purpose high performance computing. It incorporates new optimization for performance, features associated with the ParalleX execution model, and programmer services including C++ bindings and collectives.
  
HPX-5 is a reduction to practice of the revolutionary ParalleX execution model, which establishes roles and responsibilities between layers in an exascale system. It is implemented in portable C99 and is organized around a cooperative lightweight thread scheduler, a global address space, an active-message parcel transport, and a group of globally addressable local synchronization object classes. Internally, the infrastructure is built on scalable concurrent data structures to minimize shared-memory synchronization overhead. The global address space and parcel transport are based on the innovative Photon network transport library, which supports low-level access to network hardware and provides RDMA with remote completion events for low overhead signaling. An alternative ISend/IRecv network layer is included for portability, along with a reference MPI implementation. HPX-5 is compatible with Linux running on Intel x86 and Xeon Phi processors and various ARM core platforms (including both ARMv7 and ARMv8/Aarch64). A pre-release version of HPX-5 v2.0 is available for Mac OS X 10.10+.
+
HPX-5 is a realization of the ParalleX execution model, which establishes the runtime's roles and responsibilities with respect to other interoperating system layers, and explicitly includes a performance model that provides an analytic framework for performance and optimization. As an Asynchronous Multi-Tasking (AMT) software system, HPX-5 is event-driven, enabling the migration of continuations and the movement of work to data, when appropriate, based on sophisticated local control synchronization objects (e.g., futures, dataflow) and active messages. ParalleX compute complexes, embodied as lightweight, first-class threads, can block, perform global mutable side-effects, employ non-strict firing rules, and serve as continuations. HPX-5 employs an active global address space in which virtually addressed objects can migrate across the physical system without changing address. First-class named processes can span and share nodes.
  
“HPX-5 is a useful environment for exploring dynamic adaptive execution for high scalability computation as well as critical support for truly dynamic end-user science and engineering problems” said Thomas Sterling, professor at Indiana University and creator of the ParalleX execution model.  
+
HPX-5 is an evolving runtime system used both to enable dynamic adaptive parallel applications and to conduct path-finding experimentation to quantify effects of latency, overhead, contention, and parallelism of its integral mechanisms. These performance parameters determine a trade-off space within which dynamic control is performed for best performance. It is an area of active research driven by complex applications and advances in HPC architecture. HPX-5 employs dynamic and adaptive resource management and task scheduling to achieve the significant improvements in efficiency and scalability necessary to deploy many classes of parallel applications on the largest (current and future) supercomputers in the nation and world. Although still under development, HPX-5 is portable to a diverse set of systems, is reliable and programmable, scales across multi-core and multi-node systems, and delivers efficiency improvements for irregular, time-varying problems.  
  
== Introduction ==
+
HPX-5 is written primarily in portable C99 and is released under an open source BSD license. Future major releases will be delivered semi-annually, and correctness and performance bug fixes will be made available as required. To support active engagement with the larger developer community, active development branches are available. HPX-5 will also be disseminated through the OpenHPC consortium led by the Linux Foundation.
 
+
HPX-5 v2.0 (High Performance ParalleX) is the latest version of the HPX-5 runtime system that provides a unified programming model for parallel and distributed applications, allowing programs to run unmodified on systems from a single SMP to large clusters and supercomputers with thousands of nodes. HPX-5 is a realization of ParalleX, an abstract cross-cutting exascale execution model, which establishes roles and responsibilities between system layers.
+
 
+
The HPX-5 interface and library implementation is guided by the ParalleX execution model and is freely available, open source, portable, and performance-oriented. HPX-5 is a general-purpose runtime system for applications, targeted at conventional, widely available architectures. It provides a unified programming model for parallel and distributed applications. As a dynamic adaptive runtime system, it is event-driven and embodies the principles of multi-threaded computing while also providing a global name and address space and advanced synchronization constructs. For communication between nodes, it uses the Photon network layer which has been tuned for optimal single sided communications as an alternative to the Message Passing Interface (MPI).
+
  
 
{{Infobox project
 
{{Infobox project
Line 22: Line 18:
 
| imagecaption =
 
| imagecaption =
 
| download = http://hpx.crest.iu.edu/download
 
| download = http://hpx.crest.iu.edu/download
| team-members = Thomas Sterling, Andrew Lumsdaine, Kelsey Shephard, Jayashree Ajay Candadai, Matt Anderson, Luke Dalessandro, Daniel Kogler, Abhishek Kulkarni
+
| team-members = Thomas Sterling, Andrew Lumsdaine, Kelsey Shephard, Jayashree Candadai, Matt Anderson, Luke Dalessandro, Daniel Kogler, Abhishek Kulkarni
 
| pi = Ron Brightwell, Sandia
 
| pi = Ron Brightwell, Sandia
 
| co-pi = Andrew Lumsdaine
 
| co-pi = Andrew Lumsdaine
 
}}
 
}}
 
 
The ParalleX execution model itself is experimental and continues to evolve as driven by quantitative insights gathered from the Starvation-Latency-Overhead-Waiting for contention (SLOW) performance model. It aims to address the aggravating effects of asynchrony for extreme-scale machines and to improve the efficiency and scalability of scaling constrained applications. ParalleX utilizes lightweight concurrent threads managed using synchronization primitives such as dataflow and futures in order to alter the application flow structure from being message-passing to becoming message-driven. It also includes an advanced global address space model instead of relying on more conventional distributed memory structures.
 
  
 
== Audience ==
 
== Audience ==
Line 36: Line 29:
 
== Features in HPX-5 ==
 
== Features in HPX-5 ==
  
*  Fine grained execution through cooperative lightweight threads and unified access to a global address space.
+
*  Fine grained execution through blockable lightweight threads and unified access to a global address space.
*  High-performance PGAS implementation which supports low-level one-sided operations and two-sided active messages with continuations, and an experimental AGAS option that allows the binding of global to physical addresses to vary dynamically.
+
*  High-performance PGAS implementation which supports low-level one-sided operations and two-sided active messages with continuations, and an experimental AGAS option with active load balancing that allows the binding of global to physical addresses to vary dynamically.
*  Makes concurrency manageable with globally allocated lightweight control objects (LCOs) based synchronization (futures, gates, reductions) allowing thread execution or parcel instantiation to wait for events without execution resource consumption.
+
*  Makes concurrency manageable with globally allocated lightweight control objects (LCOs) based synchronization (futures, gates, reductions, dataflow) allowing thread execution or parcel instantiation to wait for events without execution resource consumption.
*  Higher level abstractions including asynchronous remote-procedure-call options, data parallel loop constructs, and system abstractions.
+
*  Higher level abstractions including asynchronous remote-procedure-call options, data parallel loop constructs, and system abstractions like timers.
Early implementation of ParalleX processes providing programmers with termination detection and per-process collectives.
+
Implementation of ParalleX processes providing programmers with termination detection and per-process collectives.
 
*  Photon networking library synthesizing RDMA-with-remote-completion directly on top of uGNI, IB verbs, or libfabric. For portability and legacy support, HPX-5 emulates RDMA-with-remote-completion using MPI point-to-point messaging.
 
*  Photon networking library synthesizing RDMA-with-remote-completion directly on top of uGNI, IB verbs, or libfabric. For portability and legacy support, HPX-5 emulates RDMA-with-remote-completion using MPI point-to-point messaging.
Distributed GPU and co-processors (Intel Xeon Phi) through experimental OpenCL support.
+
Programmer services including C++ bindings and collectives (prototype non-blocking network collectives for hierarchical process collective operation).
*  PAPI support for profiling. APEX policy engine (Autonomic Performance Environment for eXascale) support for runtime adaption.
+
*  Leverages distributed GPU and co-processors (Intel Xeon Phi) through experimental OpenCL support.
*  Migration of legacy applications through easy interoperability and co-existence with traditional runtimes like MPI. HPX-5 2.0 is also released along with several applications: LULESH, Wavelet AMR, HPCG and the ParalleX Graph Library.
+
*  PAPI support for profiling.
 +
*  Integration with APEX policy engine (Autonomic Performance Environment for eXascale) support for runtime adaption, RCR and LXK OS.
 +
*  Migration of legacy applications through easy interoperability and co-existence with traditional runtimes like MPI. HPX-5 4.0 is also released along with several applications: LULESH, Wavelet AMR, HPCG, CoMD and the ParalleX Graph Library.
  
 
== Timeline ==
 
== Timeline ==
Line 52: Line 47:
 
*  v1.0.0 released on 3rd May 2015.
 
*  v1.0.0 released on 3rd May 2015.
 
*  v2.0.0 released on 17th November 2015.
 
*  v2.0.0 released on 17th November 2015.
*  v3.0.0 is scheduled to be released in April/May 2016.
+
*  v3.0.0 released on 5th May 2016.
 +
*  v4.0.0 released on 11th November 2016.
 +
*  v5.0.0 is scheduled to be released in May 2017.
  
 
HPX-5 is developed with an agile process that includes continuous integration, regular point releases, and frequent regression tests for correctness and performance.  Users can submit issue reports to the development team through the HPX-5 web site(https://gitlab.crest.iu.edu/extreme/hpx/issues).  Future major releases of HPX-5 will be delivered semi-annually although bug fixes will be made available between major releases.
 
HPX-5 is developed with an agile process that includes continuous integration, regular point releases, and frequent regression tests for correctness and performance.  Users can submit issue reports to the development team through the HPX-5 web site(https://gitlab.crest.iu.edu/extreme/hpx/issues).  Future major releases of HPX-5 will be delivered semi-annually although bug fixes will be made available between major releases.
Line 60: Line 57:
 
== Quick Start Instructions ==
 
== Quick Start Instructions ==
  
If you plan to use HPX–5, we suggest to start with the latest released version (currently HPX–5 v2.0.0) which can be downloaded from https://hpx.crest.iu.edu/download.
+
If you plan to use HPX–5, we suggest to start with the latest released version (currently HPX–5 v4.0.0) which can be downloaded from https://hpx.crest.iu.edu/download.
 
Follow the installation directions located under hpx/INSTALL
 
Follow the installation directions located under hpx/INSTALL
  

Latest revision as of 19:42, 28 November 2016

Announcement

Indiana University announces HPX-5 version 4.0 runtime system software!

The Center for Research in Extreme Scale Technologies (CREST) at Indiana University is pleased to announce the release of version 4.0 of HPX-5, a state-of-the-art runtime system for extreme-scale computing. Version 4.0 of the HPX-5 runtime system represents a significant maturation of the sequence of HPX-5 releases to date for efficient scalable general purpose high performance computing. It incorporates new optimization for performance, features associated with the ParalleX execution model, and programmer services including C++ bindings and collectives.

HPX-5 is a realization of the ParalleX execution model, which establishes the runtime's roles and responsibilities with respect to other interoperating system layers, and explicitly includes a performance model that provides an analytic framework for performance and optimization. As an Asynchronous Multi-Tasking (AMT) software system, HPX-5 is event-driven, enabling the migration of continuations and the movement of work to data, when appropriate, based on sophisticated local control synchronization objects (e.g., futures, dataflow) and active messages. ParalleX compute complexes, embodied as lightweight, first-class threads, can block, perform global mutable side-effects, employ non-strict firing rules, and serve as continuations. HPX-5 employs an active global address space in which virtually addressed objects can migrate across the physical system without changing address. First-class named processes can span and share nodes.

HPX-5 is an evolving runtime system used both to enable dynamic adaptive parallel applications and to conduct path-finding experimentation to quantify effects of latency, overhead, contention, and parallelism of its integral mechanisms. These performance parameters determine a trade-off space within which dynamic control is performed for best performance. It is an area of active research driven by complex applications and advances in HPC architecture. HPX-5 employs dynamic and adaptive resource management and task scheduling to achieve the significant improvements in efficiency and scalability necessary to deploy many classes of parallel applications on the largest (current and future) supercomputers in the nation and world. Although still under development, HPX-5 is portable to a diverse set of systems, is reliable and programmable, scales across multi-core and multi-node systems, and delivers efficiency improvements for irregular, time-varying problems.

HPX-5 is written primarily in portable C99 and is released under an open source BSD license. Future major releases will be delivered semi-annually, and correctness and performance bug fixes will be made available as required. To support active engagement with the larger developer community, active development branches are available. HPX-5 will also be disseminated through the OpenHPC consortium led by the Linux Foundation.

HPX-5 Architecture
Architecture.png
Team Members Thomas Sterling, Andrew Lumsdaine, Kelsey Shephard, Jayashree Candadai, Matt Anderson, Luke Dalessandro, Daniel Kogler, Abhishek Kulkarni
PI Ron Brightwell, Sandia
Co-PIs Andrew Lumsdaine
Website http://hpx.crest.iu.edu
Download http://hpx.crest.iu.edu/download

Audience

HPX-5 is used for a broad range of scientific applications, helping scientists and developers write code that shows better performance on irregular applications and at scale when compared to more conventional programming models such as MPI. For the application developer, it provides dynamic adaptive resource management and task scheduling to reach otherwise unachievable efficiencies in time and energy and scalability. HPX-5 supports such applications with implementation of features like Active Global Address Space (AGAS), ParalleX Processes, Complexes (ParalleX Threads and Thread Management), Parcel Transport and Parcel Management, Local Control Objects (LCOs) and Localities. Fine-grained computation is expressed using actions. Computation is logically grouped into processes to provide quiescence and termination detection. LCOs are synchronization objects that manage local and distributed control flow and have a global address. The heart of HPX-5 is a lightweight thread scheduler that directly schedules lightweight actions by multiplexing them on a set of heavyweight scheduler threads.

Features in HPX-5

  • Fine grained execution through blockable lightweight threads and unified access to a global address space.
  • High-performance PGAS implementation which supports low-level one-sided operations and two-sided active messages with continuations, and an experimental AGAS option with active load balancing that allows the binding of global to physical addresses to vary dynamically.
  • Makes concurrency manageable with globally allocated lightweight control objects (LCOs) based synchronization (futures, gates, reductions, dataflow) allowing thread execution or parcel instantiation to wait for events without execution resource consumption.
  • Higher level abstractions including asynchronous remote-procedure-call options, data parallel loop constructs, and system abstractions like timers.
  • Implementation of ParalleX processes providing programmers with termination detection and per-process collectives.
  • Photon networking library synthesizing RDMA-with-remote-completion directly on top of uGNI, IB verbs, or libfabric. For portability and legacy support, HPX-5 emulates RDMA-with-remote-completion using MPI point-to-point messaging.
  • Programmer services including C++ bindings and collectives (prototype non-blocking network collectives for hierarchical process collective operation).
  • Leverages distributed GPU and co-processors (Intel Xeon Phi) through experimental OpenCL support.
  • PAPI support for profiling.
  • Integration with APEX policy engine (Autonomic Performance Environment for eXascale) support for runtime adaption, RCR and LXK OS.
  • Migration of legacy applications through easy interoperability and co-existence with traditional runtimes like MPI. HPX-5 4.0 is also released along with several applications: LULESH, Wavelet AMR, HPCG, CoMD and the ParalleX Graph Library.

Timeline

The HPX-5 source code is distributed with a liberal open-source license.

  • v1.0.0 released on 3rd May 2015.
  • v2.0.0 released on 17th November 2015.
  • v3.0.0 released on 5th May 2016.
  • v4.0.0 released on 11th November 2016.
  • v5.0.0 is scheduled to be released in May 2017.

HPX-5 is developed with an agile process that includes continuous integration, regular point releases, and frequent regression tests for correctness and performance. Users can submit issue reports to the development team through the HPX-5 web site(https://gitlab.crest.iu.edu/extreme/hpx/issues). Future major releases of HPX-5 will be delivered semi-annually although bug fixes will be made available between major releases.

The HPX-5 source code is released under the BSD open-source license and is distributed with a complete set of tests along with selected sample applications. HPX-5 is funded and supported by the DoD, DoE, and NSF, and is used actively in projects such as PSAAP, XPRESS, XSEDE. Further information and downloads for HPX-5 can be found at http://hpx.crest.iu.edu.

Quick Start Instructions

If you plan to use HPX–5, we suggest to start with the latest released version (currently HPX–5 v4.0.0) which can be downloaded from https://hpx.crest.iu.edu/download. Follow the installation directions located under hpx/INSTALL

Links