Texas Instruments had 17 patents in artificial intelligence during Q1 2024. Texas Instruments Inc has developed neural network processors with configurable computing engines and post-processing engines for various input, weight, and output precisions. These processors also include controllers to fetch data elements, perform operations, and generate output data efficiently. Additionally, they have implemented a CNN-based signal processing method for encrypted data and a hardware accelerator for optimizing keypoint matching through matrix multiplication and SSE calculations. GlobalData’s report on Texas Instruments gives a 360-degree view of the company including its patenting strategy. Buy the report here.
Texas Instruments grant share with artificial intelligence as a theme is 29% in Q1 2024. Grant share is based on the ratio of number of grants to total number of patents.
Recent Patents
Application: Neural network processor (Patent ID: US20240103811A1)
The patent filed by Texas Instruments Inc. describes a neural network processor that includes an input data register, weights register, and a computing engine capable of performing multiplication and accumulation operations between input data elements and weight elements of various precisions. A controller is configured to receive indications of input and weight precisions, configure the computing engine accordingly, fetch input and weight elements, and perform MAC operations to generate intermediate output data elements. The processor supports a wide range of input and weight precisions, including binary, signed 4-bit, unsigned 4-bit, signed 8-bit, and unsigned 8-bit for input, and binary, ternary, 2-bit, 4-bit, and 8-bit for weights.
The neural network processor's computing engine includes multiple computation units with specific inputs and outputs for MAC operations, controlled by a controller based on input and weight precisions. The controller also manages data multiplexing, bit shifting, and accumulation operations within the computation units. Additionally, the processor supports various shift operations based on input and weight precisions, ensuring efficient processing for different neural network applications. The patent also covers methods for configuring the processor, fetching data, performing MAC operations, and storing intermediate output data, highlighting the versatility and adaptability of the neural network processor for diverse neural network tasks.
To know more about GlobalData’s detailed insights on Texas Instruments, buy the report here.
Data Insights
From
The gold standard of business intelligence.
Blending expert knowledge with cutting-edge technology, GlobalData’s unrivalled proprietary data will enable you to decode what’s happening in your market. You can make better informed decisions and gain a future-proof advantage over your competitors.