• Publications
  • Influence
ISAAC: A Convolutional Neural Network Accelerator with In-Situ Analog Arithmetic in Crossbars
TLDR
This work explores an in-situ processing approach, where memristor crossbar arrays not only store input weights, but are also used to perform dot-product operations in an analog manner.
GenCache: Leveraging In-Cache Operators for Efficient Sequence Alignment
TLDR
The basic principles in GenCache can also be exploited by 3rd generation sequence aligners, and hardware and software techniques interact synergistically to target both memory and compute bottlenecks, while not affecting the outputs of the application.
Newton: Gravitating Towards the Physical Limits of Crossbar Acceleration
TLDR
This work introduces new techniques that apply at different levels of the tile hierarchy, some leveraging heterogeneity and others relying on divide-and-conquer numeric algorithms to reduce computations and ADC pressure, and places constraints on how a workload is mapped to tiles, thus helping reduce resource-provisioning in tiles.
Multilayer design of QCA multiplexer
TLDR
High level synthesis of digital design using the proposed multiplexer is explored that establishes the significant improvement in digital design with the layered structure over that of conventional design approaches.
ISAAC
TLDR
This work explores an in-situ processing approach, where memristor crossbar arrays not only store input weights, but are also used to perform dot-product operations in an analog manner.
Memory: The Dominant Bottleneck in Genomic Workloads
TLDR
It is made an argument that a balanced sequence alignment pipeline is almost entirely constrained by memory bottlenecks, and calls for more investment in memory systems research.
Computer Vision-based Social Distancing Surveillance with Automated Camera Calibration for Large-scale Deployment
TLDR
A computer vision-based AI-assisted solution to aid compliance with social distancing norms, which performs satisfactorily under different test scenarios, processes video feed at real-time speed, as well as addresses data privacy regulations by blurring faces of detected people, making it ideal for deployments.
GenCache
OrderLight: Lightweight Memory-Ordering Primitive for Efficient Fine-Grained PIM Computations
TLDR
This work proposes a novel lightweight memory ordering primitive for PIM use cases, OrderLight, which moves away from core-centric ordering enforcement and considerably reduces the overheads of enforcing correctness and demonstrates that OrderLight delivers 5.5 × to 8.
...
...