Medium-Temperature-Oxidized GeOx Resistive-Switching Random-Access Memory and Its Applicability in Processing-in-Memory Computing

Abstract

Processing-in-memory (PIM) is emerging as a new computing paradigm to replace the existing von Neumann computer architecture for data-intensive processing. For the higher end-user mobility, low-power operation capability is more increasingly required and components need to be renovated to make a way out of the conventional software-driven artificial intelligence. In this work, we investigate the hardware performances of PIM architecture that can be presumably constructed by resistive-switching random-access memory (ReRAM) synapse fabricated with a relatively larger thermal budget in the full Si processing compatibility. By introducing a medium-temperature oxidation in which the sputtered Ge atoms are oxidized at a relatively higher temperature compared with the ReRAM devices fabricated by physical vapor deposition at room temperature, higher device reliability has been acquired. Based on the empirically obtained device parameters, a PIM architecture has been conceived and a system-level evaluations have been performed in this work. Considerations include the cycle-to-cycle variation in the GeOx ReRAM synapse, analog-to-digital converter resolution, synaptic array size, and interconnect latency for the system-level evaluation with the Canadian Institute for Advance Research-10 dataset. A fully Si processing-compatible and robust ReRAM synapse and its applicability for PIM are demonstrated. Graphical Abstract

Aloe Vera-Inspired Cognitive Computing: Unveiling the Power of Pavlovian Conditioning and Pattern Recognition with a Synaptic RRAM Device

“…The synaptic conductance values extracted from the LTP/LTD plot in Figure 6c are used as the weight values for the image recognition task using off-chip training. 48 Moreover, we also improved the uniformity in the pulse characteristics in the Ag/ aloe vera/FTO device which is crucial for the application in ANNs. 49 All the neuromorphic pattern recognition simulations were executed using the open source Pytorch package.…”

Bipolar Resistive Switching in TiO2 Artificial Synapse Mimicking Pavlov’s Associative Learning

“…49 The off-chip training procedure has been adapted for the NN implementation. 50 The normalized conductance values (normalized to [−1, 1]) extracted from the long-term potentiation/depression curve (see Figure 5b) are used as the synaptic weights for the designed NN. The NN simulations are implemented using the open-source Pytorch package 51 and the input data set for the image recognition used from the popular Modified National Institute of Standards and Technology (MNIST) data set.…”

“…The image recognition capabilities of the fabricated synaptic device are also tested using a multilayer feed-forward Neural network (NN) design (with one hidden layer) that has been conceived, as shown in Figure a . The off-chip training procedure has been adapted for the NN implementation . The normalized conductance values (normalized to [−1, 1]) extracted from the long-term potentiation/depression curve (see Figure b) are used as the synaptic weights for the designed NN.…”

Superionic Silver Halide Solid Electrolyte: Dielectric Property and Iontronic Memtransistor Application for Bioinspired Computing

“…In order to assess the performance of our device for neuromorphic applications, we have simulated a single hidden layer ANN for off-chip digit recognition using the MNIST dataset. [36,37] The simulated ANN has 784 input nodes (corresponding to the linearized 28 × 28 images in MNIST dataset), 128 hidden nodes, and 10 output nodes (inset of Figure 7a). The Py-Torch simulation package is used for the ANN simulations.…”

“…Artificial Neural Network Training and Pattern Recognition: A single hidden layer ANN was simulated for off-chip digit recognition using the MNIST dataset for assessing the performance of the device for the neuromorphic computing. [36,37] The simulated ANN has 784 input nodes (corresponding to the linearized 28 × 28 images in MNIST dataset), 128 hidden nodes, and 10 output nodes (inset of Figure 7a). The Pytorch simulation package was used for the ANN simulations.…”