arXiv subjects

Anna Pirova

Publications and source records attributed to Anna Pirova.

4 recordsLinked to original sources

arXiv2026

Rethinking Sparse Formats for RISC-V: A Hierarchical Approach to High-Performance SpMV

The sparse matrix-vector multiplication (SpMV) algorithm is a fundamental computational kernel of linear algebra and serves as a building block for numerous applications, primarily iterative solvers for systems of linear equations used in scientific and engineering simulations. This paper compares vectorized implementations of the SpMV algorithm across eight established sparse matrix storage formats and proposes a novel modification of the CSR format, Hierarchical CSR (HCSR), which enhances SpMV performance on RISC-V processors. Our SpMV implementations utilize RVV 1.0 intrinsics and are publicly available as an open-source C++ library named RVVLASparse. Computational experiments conducted on SpacemiT K1 and K3 RISC-V boards demonstrate that selecting an appropriate matrix storage format accelerates SpMV computations by an average of 1.6x, while the proposed HCSR format achieves the shortest execution time among all considered formats across a broad class of sparse matrices.

cs.DC↗

arXiv2025

Performance optimization of BLAS algorithms with band matrices for RISC-V processors

The rapid development of RISC-V instruction set architecture presents new opportunities and challenges for software developers. Is it sufficient to simply recompile high-performance software optimized for x86-64 onto RISC-V CPUs? Are current compilers capable of effectively optimizing C and C++ codes or is it necessary to use intrinsics or assembler? Can we analyze and improve performance without well-developed profiling tools? Do standard optimization techniques work? Are there specific RISC-V features that need to be considered? These and other questions require careful consideration. In this paper, we present our experience optimizing four BLAS algorithms for band matrix operations on RISC-V processors. We demonstrate how RISC-V-optimized implementations of OpenBLAS algorithms can be significantly accelerated through improved vectorization of computationally intensive loops. Experiments on Lichee Pi 4A and Banana Pi BPI-F3 devices using RVV 0.7.1 and RVV 1.0 vector instruction sets respectively, show speedups of 1.5x to 10x depending on the operation compared to the OpenBLAS baseline. In particular, the successful use of vector register grouping with RVV can lead to significant performance improvements.

cs.DC↗

arXiv2022

Comparative statistical study of two local clustering coefficient formulations as tropical cyclone markers for climate networks

We introduce a new formulation of local clustering coefficient for weighted correlation networks. This new formulation is based upon a definition introduced previously in the neuroscience context and aimed at compensating for spurious correlations caused by indirect interactions. We modify this definition further by replacing Pearson's pairwise correlation coefficients and three-way partial correlation coefficients by the respective Kendall's rank correlations. This reduces statistical sample size requirements to compute the correlations, which translates into the possibility of using shorter time windows and hence into a shorter response time of the real-time climate network analysis. We construct evolving climate networks of mean sea level pressure fluctuations and analyze anomalies of local clustering coefficient in these networks. We develop a broadly applicable statistical methodology to study association between spatially inhomogeneous georeferenced multivariate time series and binary-valued spatiotemporal data (or other data reducible to this representation) and use it to compare the newly proposed formulation of local clustering coefficient (for weighted correlation networks) to the conventional one (for unweighted graphs) in terms of the association of these measures in climate networks to tropical cyclones. Thus we substantiate the previously made observation that tropical cyclones are associated with anomalously high values of local clustering coefficient, and confirm that the new formulation shows a stronger association.

physics.data-an↗

arXiv2021

Local clustering coefficient based on three-way partial correlations in climate networks as a new marker of tropical cyclone

We introduce a new network marker for climate network analysis. It is based upon an available special definition of local clustering coefficient for weighted correlation networks, which was previously introduced in the neuroscience context and aimed at compensating for uninformative correlations caused by indirect interactions. We modify this definition further by replacing Pearson's pairwise correlation coefficients and Pearson's three-way partial correlation coefficients by the respective Kendall's rank correlations. This reduces statistical sample size requirements to compute the correlations, which translates into the possibility of using shorter time windows and hence into shorter response time of the real-time climate network analysis. We compare this proposed network marker to the conventional local clustering coefficient based on unweighted networks obtained by thresholding the correlation matrix. We show several examples where the new marker is found to be better associated to tropical cyclones than the unweighted local clustering coefficient.

physics.data-an↗