ChASE: a distributed hybrid CPU-GPU eigensolver for large-scale hermitian eigenvalue problems

@article{Wu2022ChASEAD,
  title={ChASE: a distributed hybrid CPU-GPU eigensolver for large-scale hermitian eigenvalue problems},
  author={Xinzhe Wu and Davor Davidovic and Sebastian Achilles and Edoardo Di Napoli},
  journal={Proceedings of the Platform for Advanced Scientific Computing Conference},
  year={2022}
}
As modern massively parallel clusters are getting larger with beefier compute nodes, traditional parallel eigensolvers, such as direct solvers, struggle keeping the pace with the hardware evolution and being able to scale efficiently due to additional layers of communication and synchronization. This difficulty is especially important when porting traditional libraries to heterogeneous computing architectures equipped with accelerators, such as Graphics Processing Unit (GPU). Recently, there鈥β

Figures and Tables from this paper

References

SHOWING 1-10 OF 43 REFERENCES

GPU-acceleration of the ELPA2 distributed eigensolver for dense symmetric and hermitian eigenproblems

Parallel Shift-Invert Spectrum Slicing on Distributed Architectures with GPU Accelerators

This work will examine the performance of parallel shift-invert spectrum slicing on modern GPU clusters using state-of-the-art linear algebra software and results are a method which utilizes more floating point operations than traditional eigensolvers, but in a way which allows for the expression of massive concurrency leading to an overall improvement in time-to-solution on large computing resources.

A novel hybrid CPU鈥揋PU generalized eigensolver for electronic structure calculations based on fine-grained memory aware tasks

A generalized eigensolver featuring novel algorithms of increased computational intensity, decomposition of the computation into fine-grained memory aware tasks, and their hybrid execution are developed, which are state-of-the-art in high-performance computing.

Solving the Bethe-Salpeter equation on massively parallel architectures

PFEAST: A High Performance Sparse Eigenvalue Solver Using Distributed-Memory Linear Solvers

This paper highlights a recent development within the software package that allows the dominant computational task, solving a set of complex linear systems, to be performed with a distributed memory solver.

ChASE: Chebyshev Accelerated Subspace iteration Eigensolver for sequences of Hermitian eigenvalue problems

Novel to ChASE is the computation of the spectral estimates that enter in the filter and an optimization of the polynomial degree which further reduces the necessary FLOPs.

Towards dense linear algebra for hybrid GPU accelerated manycore systems

Task鈥恇ased, GPU鈥恆ccelerated and robust library for solving dense nonsymmetric eigenvalue problems

The StarNEig library is built on top of the StarPU runtime system and targets both shared and distributed memory machines and implements a ScaLAPACK compatibility layer which should assist new users in the transition to Star NEig.

Development of a High-Performance Eigensolver on a Peta-Scale Next-Generation Supercomputer System

A high-performance, highly scalable eigenvalue solver is introduced with the goal of realizing the K-computer system, which is a next-generation supercomputer system.

A Parallel Eigensolver for Dense Symmetric Matrices Based on Multiple Relatively Robust Representations

We present a new parallel algorithm for the dense symmetric eigenvalue/eigenvector problem that is based upon the tridiagonal eigensolver, Algorithm $\mbox{\sf MR}^3$, recently developed by Dhillon