18081488. ISA EXTENSION FOR HIGH-BANDWIDTH MEMORY simplified abstract (Samsung Electronics Co., Ltd.)
Contents
ISA EXTENSION FOR HIGH-BANDWIDTH MEMORY
Organization Name
Inventor(s)
Mu-Tien Chang of Santa Clara CA (US)
Krishna T. Malladi of San Jose CA (US)
Dimin Niu of Sunnyvale CA (US)
Hongzhong Zheng of Los Gatos CA (US)
ISA EXTENSION FOR HIGH-BANDWIDTH MEMORY - A simplified explanation of the abstract
This abstract first appeared for US patent application 18081488 titled 'ISA EXTENSION FOR HIGH-BANDWIDTH MEMORY
Simplified Explanation
The abstract describes a method for processing commands in a high-bandwidth memory (HBM) system. The method involves sending a function-in-HBM (FIM) instruction to the HBM by a HBM memory controller of a GPU. A logic component of the HBM receives the FIM instruction and coordinates the execution of the instructions using the controller, an arithmetic logic unit (ALU), and a static random-access memory (SRAM) located on the logic component.
- The method involves processing commands in a high-bandwidth memory (HBM) system.
- A function-in-HBM (FIM) instruction is sent to the HBM by a HBM memory controller of a GPU.
- A logic component of the HBM receives the FIM instruction.
- The logic component coordinates the execution of the instructions using the controller, an arithmetic logic unit (ALU), and a static random-access memory (SRAM) located on the logic component.
Potential Applications
- High-performance computing systems
- Graphics processing units (GPUs)
- Data-intensive applications
- Artificial intelligence and machine learning systems
Problems Solved
- Efficient processing of commands in a high-bandwidth memory system
- Coordinating the execution of instructions in a GPU-based system
- Optimizing memory access and data processing in data-intensive applications
Benefits
- Improved performance and efficiency in processing commands
- Enhanced memory access and data processing capabilities
- Faster execution of instructions in high-bandwidth memory systems
- Increased overall system performance and throughput
Original Abstract Submitted
A method of processing in-memory commands in a high-bandwidth memory (HBM) system includes sending a function-in-HBM instruction to the HBM by a HBM memory controller of a GPU. A logic component of the HBM receives the FIM instruction and coordinates the instructions execution using the controller, an ALU, and a SRAM located on the logic component.
