Automatic Memory Optimizations for Improving MPI Derived Datatype Performance

Byna, Suren; Sun, Xian-He; Thakur, Rajeev; Gropp, William

doi:10.1007/11846802_36

Cited by 12 publications

(5 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Although better results have been reported by, for instance, Hoefler and Gottlieb (2010) and Lu et al (2004), this inconsistency seems to hamper the widespread use of derived datatypes. This, despite a considerable amount of work by Byna et al (2006), Ross et al (2003), Schneider et al (2013) and Wu et al (2004) among others addressing the issue. Schneider et al (2013) also demonstrate runtime compilation for MPI datatypes.…”

Section: Background and Related Workmentioning

confidence: 97%

DAME: Runtime-compilation for data movement

Prabhu

Gropp

2017

The International Journal of High Performance Computing Applica

Self Cite

View full text Add to dashboard Cite

Modern machines consist of multiple compute devices and complex memory hierarchies. For many applications, it is imperative that any data movement between and within the various compute devices be done as efficiently as possible in order to obtain maximum performance. However, hand-optimizing code for one architecture will likely sacrifice both performance portability and software maintainability. In addition, some optimization decisions are best made at runtime. This suggests that the problem ought to be tackled on two fronts. First, provide the programmer with a declarative language to describe data layouts and data motion. This would allow the runtime system to be tuned for each architecture by a specialist and free the programmer to concentrate on the application itself. Second, exploit the execution time information to optimize the data movement code further. MPI derived datatypes accomplish the former task and Just In Time (JIT) compilation can be used for the latter. In this paper, we present DAME-a language and interpreter designed to be used as the backend for MPI derived datatypes. We also present DAME-L and DAME-X, two JIT-enabled implementations of DAME, all of which have been integrated into MPICH. We evaluate their performance on DDTBench and two mini-applications written with MPI derived datatypes and obtain communication speedups of up to 20 3 and miniapplication speedups of up to 3 3 .

show abstract

Section: Background and Related Workmentioning

confidence: 97%

DAME: Runtime-compilation for data movement

Prabhu

Gropp

2017

The International Journal of High Performance Computing Applica

Self Cite

View full text Add to dashboard Cite

show abstract

“…The same is true for codes which perform "manual packing" of data before sending, a practice widespread in old HPC codes, due to the fact that a compiler can optimize the packing loops of specialized codes, e.g., utilize vector instructions to copy blocks, while a simple MPI implementation might only provide a non-specialized generic DDT interpreter. However, much effort has been directed into optimizing MPI DDT implementations, and using them often gives superior performance [21][22][23]. Fig.…”

Section: Mpi Derived Datatypes the Simplest Mpi Derivedmentioning

confidence: 99%

“…Several works have been focusing on optimizing MPI datatypes with different approaches: Traff et al [48] show that some complex derived datatypes can be transformed into simpler ones, improving the packing and unpacking performance with a more compact representation. Gropp et al [21] provide a classification of the MPI derived datatypes based on their memory access patterns, discussing how they can be efficiently implemented or automatically optimized [22]. Both approaches are orthogonal to this work because they propose optimizations that can be applied before offloading and can be integrated in the offloaded handlers.…”

Section: Related Workmentioning

confidence: 99%

Network-accelerated non-contiguous memory transfers

Girolamo

Taranov

Kurth

et al. 2019

Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis

View full text Add to dashboard Cite

Applications often communicate data that is non-contiguous in the send-or the receive-buffer, e.g., when exchanging a column of a matrix stored in row-major order. While non-contiguous transfers are well supported in HPC (e.g., MPI derived datatypes), they can still be up to 5x slower than contiguous transfers of the same size. As we enter the era of network acceleration, we need to investigate which tasks to offload to the NIC: In this work we argue that non-contiguous memory transfers can be transparently networkaccelerated, truly achieving zero-copy communications. We implement and extend sPIN, a packet streaming processor, within a Portals 4 NIC SST model, and evaluate strategies for NIC-offloaded processing of MPI datatypes, ranging from datatype-specific handlers to general solutions for any MPI datatype. We demonstrate up to 10x speedup in the unpack throughput of real applications, demonstrating that non-contiguous memory transfers are a first-class candidate for network acceleration.

show abstract

“…Although better results have been reported [3,7], this inconsistency seems to hamper the widespread use of derived datatypes. This, despite a considerable amount of work [2,12,13,15] addressing the issue. In [13], Schneider et al also demonstrate runtime compilation for MPI datatypes.…”

Section: Mpi Derived Datatypesmentioning

confidence: 99%

Dame

Prabhu

Gropp

2015

Proceedings of the 22nd European MPI Users' Group Meeting

Self Cite

View full text Add to dashboard Cite

In order to achieve high performance on modern and future machines, applications need to make effective use of the complex, hierarchical memory system. Writing performanceportable code continues to be challenging since each architecture has unique memory access characteristics. In addition, some optimization decisions can only reasonably be made at runtime. This suggests that a two-pronged approach to address the challenge is required. First, provide the programmer with a means to express memory operations declaratively which will allow a runtime system to transparently access the memory in the best way and second, exploit runtime information. MPI's derived datatypes accomplish the former although their performance in current MPI implementations shows scope for improvement. JIT-compilation can be used for the latter. In this work, we present DAME-a language and interpreter that is used as the backend for MPI's derived datatypes. We also present DAME-L and DAME-X, two JIT-enabled implementations of DAME. All three implementations have been integrated into MPICH. We evaluate the performance of our implementations using DDTBench and two mini-applications written with MPI derived datatypes and obtain communication speedups of up to 20x and mini-application speedup of 3x. CCS Concepts •Software and its engineering → Runtime environments; Just-in-time compilers; Parallel programming languages;

show abstract

Automatic Memory Optimizations for Improving MPI Derived Datatype Performance

Cited by 12 publications

References 9 publications

DAME: Runtime-compilation for data movement

DAME: Runtime-compilation for data movement

Network-accelerated non-contiguous memory transfers

Dame

Contact Info

Product

Resources

About