NDLI: Implementing a classic: zero-copy all-to-all communication with mpi datatypes

Please wait, while we are loading the content...

HPC for the human brain project

LAWS: locality-aware work-stealing for multi-socket multi-core architectures

DTail: a flexible approach to DRAM refresh management

Evaluation of methods to integrate analysis into a large-scale shock shock physics code

21st century computer architecture keynote at 2014 international conference on supercomputing (ICS)

MT-MPI: multithreaded MPI for many-core environments

An optimal distributed load balancing algorithm for homogeneous work units

Revealing applications' access pattern in collective I/O for cache management

Palm: easing the burden of analytical performance modeling

The future of supercomputing

Acceleration of derivative calculations with application to radial basis function: finite-differences on the intel mic architecture

Galaxy: a high-performance energy-efficient multi-chip architecture using photonic interconnects

Scaling up matrix computations on shared-memory manycore systems with 1000 CPU cores

Effective automatic computation placement and data allocation for parallelization of regular programs

Last-level cache deduplication

Input-adaptive parallel sparse fast fourier transform for stream processing

Implementing a classic: zero-copy all-to-all communication with mpi datatypes

Addressing bandwidth contention in SMT multicores through scheduling

Supporting storage configuration for I/O intensive workflows

An end-to-end analysis of file system features on sparse virtual disks

An efficient two-dimensional blocking strategy for sparse matrix-vector multiplication on GPUs

A performance perspective on energy efficient HPC links

Collective memory transfers for multi-core chips

On the conditions for efficient interoperability with threads: an experience with PGAS languages using cray communication domains

Block value based insertion policy for high performance last-level caches

Thread-cooperative, bit-parallel computation of levenshtein distance on GPU

Value influence analysis for message passing applications

An adaptive cross-architecture combination method for graph traversal

Understanding the impact of threshold voltage on MLC flash memory performance and reliability

Improving performance by matching imbalanced workloads with heterogeneous platforms

A programming system for xeon phis with runtime SIMD parallelization

Verifying micro-architecture simulators using event traces

Scalable analysis of multicore data reuse and sharing

HOMR: a hybrid approach to exploit maximum overlapping in MapReduce over high performance interconnects

Multi-stage coordinated prefetching for present-day processors

Load balancing n-body simulations with highly non-uniform density

Scalable performance analysis of exascale MPI programs through signature-based clustering algorithms

Accelerating cache coherence mechanism with speculation

DWC: dynamic write consolidation for phase change memory systems

Long-term resource fairness: towards economic fairness on pay-as-you-use computing systems

Unified on-chip memory allocation for SIMT architecture

Reducing energy consumption of NoC by router bypassing

Hardware-assisted scalable flow control of shared receive queue

Automating and optimizing data transfers for many-core coprocessors

Parallelizing and optimizing sparse tensor computations

Implementing a classic: zero-copy all-to-all communication with mpi datatypes

Content Provider	ACM Digital Library
Author	Träff, Jesper Larsson Rougier, Antoine Hunold, Sascha
Abstract	We investigate the use of the derived datatype mechanism of MPI (the Message-Passing Interface) in the implementation of the classic all-to-all communication algorithm of Bruck et al.\ (1997). Through a series of improvements to the canonical implementation of the algorithm we gradually eliminate initial and final processor-local data reorganizations, culminating in a \emph{zero-copy} version that contains no explicit, process-local data movement or copy operations: all necessary data movements are implied by MPI derived datatypes, and carried out as part of the communication operations. We furthermore show how the improved algorithm can be used to solve irregular all-to-all communication problems (that are not too irregular). The Bruck algorithm serves as a vehicle to demonstrate descriptive and performance advantages with MPI datatypes in the implementation of complex algorithms, and discuss shortcomings and inconveniences in the current MPI datatype mechanism. In particular, we use and implement three new derived datatypes (bounded vector, circular vector, and bucket) not in MPI that might be useful in other contexts. We also discuss the role of persistent collectives which are currently not found in MPI for amortizing type creation (and other) overheads, and implement a persistent variant of the \texttt{MPI\_Alltoall} collective. On two small systems we experimentally compare the algorithmic improvements to the Bruck et al.\ algorithm when implemented on top of MPI, showing the zero-copy version to perform significantly better than the initial, straight-forward implementation. One of our variants has also been implemented inside \texttt{mvapich}, and we show it to perform better than the \texttt{mvapich} implementation of the Bruck et al.\ algorithm for the range of processes and problem sizes where it is enabled. The persistent version of \texttt{MPI\_Alltoall} has no overhead and outperforms all other variants, and in particular improves upon the standard implementation by 50\% to 15\% across the full range of problem sizes considered.
Starting Page	135
Ending Page	144
Page Count	10
File Format	PDF
ISBN	9781450326421
DOI	10.1145/2597652.2597662
Language	English
Publisher	Association for Computing Machinery (ACM)
Publisher Date	2014-06-10
Publisher Place	New York
Access Restriction	Subscribed
Subject Keyword	Message-passing interface (mpi) Derived datatypes All-to-all collective communication
Content Type	Text
Resource Type	Article

Central Library (ISO-9001:2015 Certified)
Indian Institute of Technology Kharagpur
Kharagpur, West Bengal, India | PIN - 721302

See location in the Map
03222 282435
Mail: support@ndl.gov.in

Sl.	Authority	Responsibilities	Communication Details
1	Ministry of Education (GoI), Department of Higher Education	Sanctioning Authority	https://www.education.gov.in/ict-initiatives
2	Indian Institute of Technology Kharagpur	Host Institute of the Project: The host institute of the project is responsible for providing infrastructure support and hosting the project	https://www.iitkgp.ac.in
3	National Digital Library of India Office, Indian Institute of Technology Kharagpur	The administrative and infrastructural headquarters of the project	Dr. B. Sutradhar bsutra@ndl.gov.in
4	Project PI / Joint PI	Principal Investigator and Joint Principal Investigators of the project	Dr. B. Sutradhar bsutra@ndl.gov.in Prof. Saswat Chakrabarti will be added soon
5	Website/Portal (Helpdesk)	Queries regarding NDLI and its services	support@ndl.gov.in
6	Contents and Copyright Issues	Queries related to content curation and copyright issues	content@ndl.gov.in
7	National Digital Library of India Club (NDLI Club)	Queries related to NDLI Club formation, support, user awareness program, seminar/symposium, collaboration, social media, promotion, and outreach	clubsupport@ndl.gov.in
8	Digital Preservation Centre (DPC)	Assistance with digitizing and archiving copyright-free printed books	dpc@ndl.gov.in
9	IDR Setup or Support	Queries related to establishment and support of Institutional Digital Repository (IDR) and IDR workshops	idr@ndl.gov.in