Geodesic
Shared memory based MPI Reduce and Bcast algorithms
Numerical methods and programming, Tome 24 (2023) no. 4, pp. 339-351

Voir la notice de l'article provenant de la source Math-Net.Ru

Algorithms for implementing collective operations MPI_Bcast, MPI_Reduce, MPI_Allreduce using shared memory of multiprocessor servers are proposed. The algorithms create a shared memory segment and a system of queues in it, through which message blocks are transmitted. The software implementation is based on the Open MPI library as an isolated coll/sharm component. Unlike existing algorithms, interaction with the queuing system is organized with spinlock and focused on reducing the number of barrier synchronizations and atomic operations. When conducting experiments on a server with x86–64 architecture for the MPI_Bcast operation, the largest reduction in time was obtained by 6.5 times (85% less) and MPI_Reduce by 3.3 times (70% less) compared to the implementation in the coll/tuned component of the Open MPI library. Recommendations on the use of algorithms for different message sizes are suggested.
Keywords: Bcast; Reduce; Allreduce; collective operations; MPI; computer systems. 
@article{VMP_2023_24_4_a8,
     author = {A. A. Romanyuta and M. G. Kurnosov},
     title = {Shared memory based {MPI} {Reduce} and {Bcast} algorithms},
     journal = {Numerical methods and programming},
     pages = {339--351},
     publisher = {mathdoc},
     volume = {24},
     number = {4},
     year = {2023},
     language = {ru},
     url = {http://geodesic.mathdoc.fr/item/VMP_2023_24_4_a8/}
}
TY  - JOUR
AU  - A. A. Romanyuta
AU  - M. G. Kurnosov
TI  - Shared memory based MPI Reduce and Bcast algorithms
JO  - Numerical methods and programming
PY  - 2023
SP  - 339
EP  - 351
VL  - 24
IS  - 4
PB  - mathdoc
UR  - http://geodesic.mathdoc.fr/item/VMP_2023_24_4_a8/
LA  - ru
ID  - VMP_2023_24_4_a8
ER  - 
%0 Journal Article
%A A. A. Romanyuta
%A M. G. Kurnosov
%T Shared memory based MPI Reduce and Bcast algorithms
%J Numerical methods and programming
%D 2023
%P 339-351
%V 24
%N 4
%I mathdoc
%U http://geodesic.mathdoc.fr/item/VMP_2023_24_4_a8/
%G ru
%F VMP_2023_24_4_a8
A. A. Romanyuta; M. G. Kurnosov. Shared memory based MPI Reduce and Bcast algorithms. Numerical methods and programming, Tome 24 (2023) no. 4, pp. 339-351. http://geodesic.mathdoc.fr/item/VMP_2023_24_4_a8/