Joshua San Miguel

Joshua San Miguel

Auto-generated profile, based on publicly available information

Joshua San Miguel

University of Wisconsin-Madison, Madison, USA

Research Focus & Metrics

Academic Career Length

10Years (2017-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
27 November 2017 in IEEE Computer Architecture Letters
Verified
Journal Article
Closed access
JMJoshua San Miguel
Joshua San Miguel
KGKarthik Ganesan
Karthik Ganesan
MBMario Badr
Mario Badr
NJNatalie Enright Jerger
Natalie Enright Jerger
Energy-harvesting devices-which operate solely on energy collected from their environment-have brought forth a new paradigm of intermittent computing. These devices succumb to frequent power outages that would cause conventional systems to be stuck in a perpetual loop of restarting computation and never making progress. Ensuring forward progress in an intermittent execution model is difficult and requires saving state in non-volatile memory. In this work, we propose the EHmodelto explore the trade-offs associated with backing up data to maximize forward progress. In particular, we focus on the relationship between energy and forward progress and how they are impacted by backups/restores to derive insights for programmers and architects.

11 January 2021 in IEEE Design & Test
Verified
Journal Article
Closed access
DWDi Wu
Di Wu
RYRuokai Yin
Ruokai Yin
JMJoshua San Miguel
Joshua San Miguel
Stochastic Computing (SC) has shown great promise in achieving low hardware area and power consumption for neuromorphic architectures compared to traditional binary-encoded computation, due to its bit-serial data representation and extremely straightforward logic. With emerging deep learning models requiring more sophisticated nonlinear operations, we propose new designs for SC division and square root. Our designs are novel in their ability to leverage SC correlation via low-cost in-stream mechanisms that eliminate expensive bit stream regeneration. Our experiments show that, compared to state-of-the-art designs, our proposed division and square root units simultaneously achieve higher accuracy and consume less area.

11 March 2021 in IEEE Micro
Verified
Journal Article
Closed access
DWDi Wu
Di Wu
JLJingjie Li
Jingjie Li
RYRuokai Yin
Ruokai Yin
HHHsuan Hsiao
Hsuan Hsiao
YKYounghyun Kim
Younghyun Kim
JMJoshua San Miguel
Joshua San Miguel
General matrix multiplication (GEMM) is pervasive in various domains, such as signal processing, computer vision, and machine learning. Conventional binary architectures for GEMM exhibit poor scalability in area and energy efficiency, due to the spatial nature of number representation and computing. On the contrary, unary computing processes data in temporal domain with extremely simple logic. However, to date, there rarely exist efficient architectures for unary GEMM. In this work, we first present uGEMM, a hardware-efficient unary GEMM architecture enabled by universally-compatible arithmetic units, which simultaneously achieves input-insensitivity and high output accuracy. Next, we demonstrate that the proposed uGEMM can reliably early terminate the computation and offers dynamic energy-accuracy scaling for real-world applications via an accuracy-aware metric. Finally, to propel the future research for unary computing, we open source our unary computing simulator, UnarySim.

26 October 2020 in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
Verified
Journal Article
Closed access
MHMaedeh Hemmat
Maedeh Hemmat
JMJoshua San Miguel
Joshua San Miguel
ADAzadeh Davoodi
Azadeh Davoodi
In this work, we propose AirNN, a novel framework which enables dynamic approximation of an already-trained convolutional neural network (CNN) in hardware during inference. AirNN enables input-dependent approximation of the CNN to achieve energy saving without much degradation in its classification accuracy at runtime. For each input, AirNN uses only a fraction of the CNN’s weights based on that input (with the rest remaining 0) to conduct the inference. Consequently, energy saving is possible due to fewer number of fetches from off-chip memory as well as fewer multiplications for majority of the inputs. To achieve per-input approximation, we propose a clustering algorithm that groups similar weights in the CNN based on their importance, and design an iterative framework that decides dynamically how many clusters of weights should be fetched from off-chip memory for each individual input. We also propose new hardware structures to implement our framework on top of a recently proposed FPGA-based CNN accelerator. In our experiments with popular CNNs, we, on average, show 49% energy saving with less than 3% degradation in classification accuracy due to doing inference with only a fraction of the weights for the majority of the inputs. We also propose a greedy interleaving scheme, implemented in hardware, in order to improve the performance of the iterative procedure and compensate for its latency overhead.

26 July 2021
Verified
Conference Paper
Closed access
DWDi Wu
Di Wu
JLJingjie Li
Jingjie Li
SBSetareh Behroozi
Setareh Behroozi
YKYounghyun Kim
Younghyun Kim
JMJoshua San Miguel
Joshua San Miguel
Linear multiply-accumulate (MAC) operations have been the main focus of prior efforts in improving the energy efficiency of neural network inference due to their dominant contribution to energy consumption in traditional models. On the other hand, nonlinear operations, such as division, exponentiation, and logarithm, that are becoming increasingly significant in emerging neural network models, have been largely underexplored. In this paper, we propose UNO, a low-area, low-energy processing element that virtualizes the Taylor approximation of nonlinear operations on top of off-the-shelf linear MAC units already present in inference hardware. Such virtualization approximates multiple nonlinear operations in a unified, MAC-compatible manner to achieve dynamic run-time accuracy-energy scaling. Compared to the baseline, our scheme reduces the energy consumption by up to 38.4% for individual operations and increases the energy efficiency by up to 274.5% for emerging neural network models with negligible inference loss

27 April 2024
Verified
Conference Paper
Open access
ZPZhewen Pan
Zhewen Pan
JMJoshua San Miguel
Joshua San Miguel
DWDi Wu
Di Wu
In recent years, hardware architectures optimized for general matrix multiplication (GEMM) have been well studied to deliver better performance and efficiency for deep neural networks. With trends towards batched, low-precision data, e.g., FP8 format in this work, we observe that there is growing untapped potential for value reuse. We propose a novel computing paradigm, value-level parallelism, whereby unique products are computed only once, and different inputs subscribe to (select) their products via temporal coding. Our architecture, Carat, employs value-level parallelism and transforms multiplication into accumulation, performing GEMMs with efficient multiplier-free hardware. Experiments show that, on average, Carat improves iso-area throughput and energy efficiency by 1.02× and 1.06× over a systolic array and 3.2× and 4.3× when scaled up to multiple nodes.