Spark Blockmatrix multiply와 CARMA algorithm matrix multiply 직접 구현해 비교

관련논문 그리드 기반 알고리즘 SUMMA

SUMMA: Scalable Universal Matrix Multiplication Algorithm, 1997, 400
- http://www.netlib.org/lapack/lawnspdf/lawn96.pdf

2.5D 알고리즘

Memory-Efficient Matrix Multiplication in the BSP Model, 1999, 54
- https://link.springer.com/content/pdf/10.1007%2FPL00008264.pdf
Improving communication performance in dense linear algebra via topology aware collectives, 2011, 50
- http://delivery.acm.org/10.1145/2070000/2063487/a77-solomonik.pdf?ip=1.209.175.113&id=2063487&acc=ACTIVE%20SERVICE&key=0EC22F8658578FE1%2EB574870CA11B57BF%2E4D4702B0C3E38B35%2E4D4702B0C3E38B35&CFID=962351038&CFTOKEN=84353517&__acm__=1500722754_5745844e4c10eafae95deae14e54dbcb
Communication-optimal parallel 2.5D matrix multiplication and LU factorization algorithms, 2011, 114
- http://www.netlib.org/lapack/lawnspdf/lawn248.pdf

BFS/DFS 기반 알고리즘

Communication-Optimal Parallel Algorithm for Strassen’s Matrix Multiplication, 2012, 75
- https://arxiv.org/pdf/1202.3173.pdf

Recursive algorithm asymptotically minimizes the communication costs for any hierarchy of caches and any matrix dimension, without any machine-dependent parameters appearing in the algorithm, making the algorithm cache-oblivious.

Cache-Oblivious Algorithms, 1999, 975
- https://www.cs.cmu.edu/~guyb/realworld/slidesF10/FLP99.pdf

BFS/DFS algorithms are cache-, processor- and network-oblivious.

Network-Oblivious Algorithms, 2014, 23
- https://arxiv.org/pdf/1404.3318.pdf
Oblivious algorithms for multicores and network of processors, 2010, 84
- http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=5470354
Resource oblivious sorting on multicores., 2010, 37
- http://www.cs.utexas.edu/~vlr/papers/icalp10.pdf

ddps-lab / distributed-matrix-completion

Spark Blockmatrix multiply와 CARMA algorithm matrix multiply 직접 구현해 비교 #11