Rapid yet accurate Tile-circuit and device modeling for Analog In-Memory Computing
Authors:
J. Luquin,
C. Mackin,
S. Ambrogio,
A. Chen,
F. Baldi,
G. Miralles,
M. J. Rasch,
J. Büchel,
M. Lalwani,
W. Ponghiran,
P. Solomon,
H. Tsai,
G. W. Burr,
P. Narayanan
Abstract:
Analog In-Memory Compute (AIMC) can improve the energy efficiency of Deep Learning by orders of magnitude. Yet analog-domain device and circuit non-idealities -- within the analog ``Tiles'' performing Matrix-Vector Multiply (MVM) operations -- can degrade neural-network task accuracy. We quantify the impact of low-level distortions and noise, and develop a mathematical model for Multiply-ACcumulat…
▽ More
Analog In-Memory Compute (AIMC) can improve the energy efficiency of Deep Learning by orders of magnitude. Yet analog-domain device and circuit non-idealities -- within the analog ``Tiles'' performing Matrix-Vector Multiply (MVM) operations -- can degrade neural-network task accuracy. We quantify the impact of low-level distortions and noise, and develop a mathematical model for Multiply-ACcumulate (MAC) operations mapped to analog tiles. Instantaneous-current IR-drop (the most significant circuit non-ideality), and ADC quantization effects are fully captured by this model, which can predict MVM tile-outputs both rapidly and accurately, as compared to much slower rigorous circuit simulations. A statistical model of PCM read noise at nanosecond timescales is derived from -- and matched against -- experimental measurements. We integrate these (statistical) device and (deterministic) circuit effects into a PyTorch-based framework to assess the accuracy impact on the BERT and ALBERT Transformer networks. We show that hardware-aware fine-tuning using simple Gaussian noise provides resilience against ADC quantization and PCM read noise effects, but is less effective against IR-drop. This is because IR-drop -- although deterministic -- is non-linear, is changing significantly during the time-integration window, and is ultimately dependent on all the excitations being introduced in parallel into the analog tile. The apparent inability of simple Gaussian noise applied during training to properly prepare a DNN network for IR-drop during inference implies that more complex training approaches -- incorporating advances such as the Tile-circuit model introduced here -- will be critical for resilient deployment of large neural networks onto AIMC hardware.
△ Less
Submitted 5 May, 2025;
originally announced June 2025.
Ground-state energy estimation of the water molecule on a trapped ion quantum computer
Authors:
Yunseong Nam,
Jwo-Sy Chen,
Neal C. Pisenti,
Kenneth Wright,
Conor Delaney,
Dmitri Maslov,
Kenneth R. Brown,
Stewart Allen,
Jason M. Amini,
Joel Apisdorf,
Kristin M. Beck,
Aleksey Blinov,
Vandiver Chaplin,
Mika Chmielewski,
Coleman Collins,
Shantanu Debnath,
Andrew M. Ducore,
Kai M. Hudek,
Matthew Keesan,
Sarah M. Kreikemeier,
Jonathan Mizrahi,
Phil Solomon,
Mike Williams,
Jaime David Wong-Campos,
Christopher Monroe
, et al. (1 additional authors not shown)
Abstract:
Quantum computing leverages the quantum resources of superposition and entanglement to efficiently solve computational problems considered intractable for classical computers. Examples include calculating molecular and nuclear structure, simulating strongly-interacting electron systems, and modeling aspects of material function. While substantial theoretical advances have been made in mapping thes…
▽ More
Quantum computing leverages the quantum resources of superposition and entanglement to efficiently solve computational problems considered intractable for classical computers. Examples include calculating molecular and nuclear structure, simulating strongly-interacting electron systems, and modeling aspects of material function. While substantial theoretical advances have been made in mapping these problems to quantum algorithms, there remains a large gap between the resource requirements for solving such problems and the capabilities of currently available quantum hardware. Bridging this gap will require a co-design approach, where the expression of algorithms is developed in conjunction with the hardware itself to optimize execution. Here, we describe a scalable co-design framework for solving chemistry problems on a trapped ion quantum computer, and apply it to compute the ground-state energy of the water molecule. The robust operation of the trapped ion quantum computer yields energy estimates with errors approaching the chemical accuracy, which is the target threshold necessary for predicting the rates of chemical reaction dynamics.
△ Less
Submitted 7 March, 2019; v1 submitted 26 February, 2019;
originally announced February 2019.
