Convolution is the most time-consuming part in the computation of convolutional neural networks (CNNs). Due to the complex data dependency and the increase in the amount of model samples, the convolution suffers from high overhead on data movement. This ...

GPUs are currently the platform of choice for training neural networks. However, training a deep neural network (DNN) is a time-consuming process even on GPUs because of the massive number of parameters that have to be learned. As a result, accelerating ...

Neuromorphic computing offers the potential of an unprecedented level of parallelism at a local scale. Although in their infancy, current first-generation neuromorphic processing units (NPUs) deliver as many as 128K artificial neurons in a package ...

ParlayLib is a C++ library for developing efficient parallel algorithms and software on shared-memory multicore machines. It provides additional tools and primitives that go beyond what is available in the C++ standard library, and simplifies the task ...

In this paper we develop optimal algorithms in the binary-forking model for a variety of fundamental problems, including sorting, semisorting, list ranking, tree contraction, range minima, and ordered set union, intersection and difference. In the ...

This paper extends the Transactional Memory (TM) paradigm by proposing a new powerful abstraction, the transactional future. Transactional futures, as the name suggests, combine TM with futures, by allowing programmers to exploit intra-transaction ...

Sparse matrix-matrix multiplication (SpGEMM) is a widely used kernel in various graph, scientific computing and machine learning algorithms. It is well known that SpGEMM is a memory-bound operation, and its peak performance is expected to be bound by ...

There has been a significant amount of research on hardware and software support for efficient concurrent data structures; yet, the question of how to build correct, simple, and scalable data structures has not yet been definitively settled. In this ...