COSMO: Computing with Stochastic Numbers in Memory

Abstract

Stochastic computing (SC) reduces the complexity of computation by representing numbers with long streams of independent bits. However, increasing performance in SC comes with either an increase in area or a loss in accuracy. Processing in memory (PIM) computes data in-place while having high memory density and supporting bit-parallel operations with low energy consumption. In this article, we propose COSMO, an architecture for co mputing with s tochastic numbers in me mo ry, which enables SC in memory. The proposed architecture is general and can be used for a wide range of applications. It is a highly dense and parallel architecture that supports most SC encodings and operations in memory. It maximizes the performance and energy efficiency of SC by introducing several innovations: (i) in-memory parallel stochastic number generation, (ii) efficient implication-based logic in memory, (iii) novel memory bit line segmenting, (iv) a new memory-compatible SC addition operation, and (v) enabling flexible block allocation. To show the generality and efficiency of our stochastic architecture, we implement image processing, deep neural networks (DNNs), and hyperdimensional (HD) computing on the proposed hardware. Our evaluations show that running DNN inference on COSMO is 141× faster and 80× more energy efficient as compared to GPU.

Stochastic-HD: Leveraging Stochastic Computing on the Hyper-Dimensional Computing Pipeline

“…Unlike AnalogPIM designs, we sample the output in the time domain. For example, to generate a dot product with 4 bits of precision, we use the same search circuits but latch the output at 16 different time instants (Gupta et al, 2022 ). Figure 5 has an example of a query search for a hypervector and how the hypervectors are laid out in memory.…”