# US 2025/0372145 A1 · Integration of in-memory analog computing architectures with systolic arrays

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2025/0372145 A1
- Document ID: US20250372145A1
- Status: Published application
- Kind: A1
- Application number: 19/225,634
- Filed: 2025-06-02
- Published: 2025-12-04
- Earliest claimed: 2024-06-03
- Assignees: University of South Carolina
- Applicants: University of South Carolina
- Inventors: Mohammed Essa; Ramtin Mohammadizand; Brendan C. Reidy
- Web page: https://patentgrep.com/patent/US20250372145A1

## CPC

- G11C 11/4063: Auxiliary circuits, e.g. for addressing, decoding, driving, writing, sensing or timing
- G06F 9/30014: {with variable precision}
- G06F 9/30036: {Instructions to perform operations on packed data, e.g. vector, tile or matrix operations}
- G06F 9/544: {Buffers; Shared memory; Pipes}
- G06F 15/8046: {Systolic arrays}
- G06N 3/048: Activation functions
- G06N 3/065: Analogue means
- G06N 3/082: modifying the architecture, e.g. adding, deleting or silencing nodes or connections
- G06N 3/084: Backpropagation, e.g. using gradient descent
- G06F 2209/543: Local
- G11C: STATIC STORES (semiconductor memory devices H10B)
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

The system architecture trained by a training component using a unified training component and method is a heterogeneous hardware that accelerates essential operations of artificial intelligence models by incorporating both systolic arrays and IMAC circuits. To leverage the strengths of systolic arrays for convolutional layers and the strengths of IMAC circuits for dense layers, the unified training component utilizes a training method with mixed-precision training techniques to train the different types of layers.

## Claims

### Claim 1

A hybrid computing device, comprising:
an in-memory analog computing (IMAC) architecture comprising a plurality of interconnected subarrays;
an analog-to-digital converter interconnecting the IMAC architecture with a memory unit; and
a systolic array operably connected to the memory unit.

### Claim 2 (depends on 1)

The device of claim 1, wherein the plurality of interconnected subarrays are linked by a plurality of programmable switch blocks.

### Claim 3 (depends on 1)

The device of claim 1, wherein each of the plurality of interconnected subarrays is made up of a plurality of memristive crossbars leading to a plurality of differential amplifiers and a plurality of analog neuron circuits.

### Claim 4 (depends on 1)

The device of claim 1, wherein the memory unit is a dynamic random access memory (DRAM).

### Claim 5 (depends on 4)

The device of claim 4, wherein the DRAM is a low-power double data rate (LPDDR) DRAM.

### Claim 6 (depends on 1)

The device of claim 1, wherein the systolic array comprises a plurality of processing elements (PEs).

### Claim 7 (depends on 6)

The device of claim 6, wherein the PEs comprise multiply-and-accumulate (MAC) units responsible for executing matrix-matrix, vector-vector, and matrix-vector multiplications.

### Claim 8 (depends on 1)

The device of claim 1, wherein the systolic array is a tensor processing unit (TPU).

### Claim 9 (depends on 1)

The device of claim 1, wherein the systolic array is a central processing unit (CPU) or a graphics processing unit (GPU) integrated with a systolic array.

### Claim 10 (depends on 1)

A method of using a unified training component to train the hybrid computing device of claim 1, comprising:
inserting a tanh activation function before a first dense fully connected (FC) layer and after a last convolutional layer to ensure that activations stay within a range of {−1, 1};
training a plurality of FC layers and a plurality of convolutional layers using identical data to produce a plurality of trained FC layers and a plurality of trained convolutional layers;
retraining an FC section of the IMAC architecture to produce a plurality of retrained FC layers;
modifying the plurality of retrained FC layers based on characteristics of weights and activation functions of the IMAC architecture.

### Claim 11 (depends on 10)

The method of claim 10, further comprising training the plurality of FC layers and the plurality of convolutional layers using a machine learning training method.

### Claim 12 (depends on 11)

The method of claim 11, wherein the machine learning training method is selected from the group consisting of: a backpropagation method, a reinforcement learning method, and an unsupervised learning method.

### Claim 13 (depends on 11)

The method of claim 11, further comprising freezing the plurality of trained convolutional layers after reaching a predetermined loss value.

### Claim 14 (depends on 11)

The method of claim 11, further comprising freezing the plurality of trained convolutional layers after reaching a predetermined training iteration.

### Claim 15 (depends on 10)

The method of claim 10, wherein retraining the FC section of the IMAC architecture utilizes ternary weights.

### Claim 16 (depends on 10)

The method of claim 10, wherein retraining the FC section comprises replacing the tanh activation function with a sign function to produce input values of −1 and 1 for the plurality of FC layers of the FC section.

### Claim 17 (depends on 10)

The method of claim 10, wherein retraining the FC section comprises retraining the entire FC section, starting with any untrained FC layers from the plurality of FC layers.

### Claim 18 (depends on 10)

The method of claim 10, wherein retraining the FC section comprises retraining only the plurality of trained FC layers.

### Claim 19 (depends on 10)

The method of claim 10, further comprising modifying the plurality of retrained FC layers by employing ternary synapses and sigmoid activation functions.

### Claim 20 (depends on 10)

The method of claim 10, further comprising modifying the plurality of retrained FC layers using RRAM-based synapses and neurons.
