Task 004: Neural Primitives

The abstract: 

In-memory computing architectures present a promising solution to address the memory- and the power-wall challenges by mitigating the bottleneck between processing units and storage. Such architectures incorporate computing functionalities inside memory arrays to make better use of the large internal memory bandwidth, thereby, avoiding frequent data movements. In-DRAM computing architectures offer high throughput and energy improvements in accelerating modern data-intensive applications like machine learning. In this talk, we present a vector addition methodology inside DRAM arrays through functional read enabled on local word-lines. The proposed primitive performs majority-based addition operations by storing data in transposed manner. Majority functions are achieved in DRAM cells by activating odd number of rows simultaneously. The proposed majority based bit-serial addition enables huge parallelism and high throughput. Moreover, we validate the robustness of the proposed in-DRAM computing methodology under process variations to ascertain its reliability. Energy evaluation of the proposed scheme shows 21.7X improvement compared to normal data read operations in standard DDR3-1333 interface. Further, compared to state-of-the-art in-DRAM compute proposals, the proposed scheme provides one of the fastest addition mechanisms with low area overhead (< 1% of DRAM chip area). Our system evaluation running the k-Nearest Neighbor (kNN) algorithm on the MNIST handwritten digit classification dataset shows 11.5X performance improvement compared to a conventional von-Neumann machine.

My bio:

Mustafa Ali received his B.Sc. ans M.Sc. degrees in Electrical Engineering from MTC, Cairo, Egypt in 2011, 2016 respectively. He achieved the 1st rank in undergraduate in MTC, 2011. He was honored the Duty Medal for excellent performance during his studies. He worked on flexible electronics applications using TFTs in his M.Sc from 2014 to 2016. Additionally, he worked as a TA and RA at MTC from 2013 to 2017. Mustafa was also a hardware and embedded systems engineer at Integreight, Inc. from 2012 to 2017. He joined the Nano-electronics Research Lab (NRL), Purdue University in Spring 2018 and he is currently pursuing his Ph.D. under the guidance of Prof. Roy. His research interest lies in accelerating Brain-inspired Computing and machine learning. He is also interested in in-memory computing based on CMOS and post-CMOS devices.