Computational Storage Shared File System Accelerator Data Access

Solution Overview

Problem

Existing computational storage systems face inefficiencies in data access and transfer when offloading computational tasks to accelerators, leading to increased time and energy consumption.

Innovation Solution

A computational storage system is designed with a shared file system accessible by both the host and the accelerator, allowing data to be stored and accessed directly by the accelerator, reducing the need for data transfer between the host and the accelerator.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If data is transferred between host and accelerator for computational tasks, then computational tasks can be performed on external data, but data transfer time and energy consumption increase

Engineering Contradiction:

Improvecomputational task execution capabilityVSAvoiddata transfer time

Core Design Contradiction:

Adaptability or versatilityVSLoss of time

Solution Approach 1:

The patent combines storage and computation resources by allowing accelerators to directly access storage devices through a shared file system, merging what were previously separate data transfer and computation operations into a unified access model that eliminates intermediate data movement

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The shared file system acts as an intermediary layer between the host, accelerators, and storage devices, providing a common interface that enables direct accelerator-access to stored data without requiring traditional host-mediated data transfer protocols

Inventive Principle:

Principle #24Intermediary (Mediator)

2Adaptability or versatility

If data is transferred between host and accelerator for computational tasks, then computational tasks can be performed on external data, but energy consumption increases

Engineering Contradiction:

Improvecomputational task execution capabilityVSAvoiddata transfer energy consumption

Core Design Contradiction:

Adaptability or versatilityVSUse of energy by moving object

Solution Approach 1:

The patent combines storage and computation resources by allowing accelerators to directly access storage devices through a shared file system, merging what were previously separate data transfer and computation operations into a unified access model that eliminates intermediate data movement

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The shared file system acts as an intermediary layer between the host, accelerators, and storage devices, providing a common interface that enables direct accelerator-access to stored data without requiring traditional host-mediated data transfer protocols

Inventive Principle:

Principle #24Intermediary (Mediator)

3Productivity

If accelerators are programmed to perform operations on transferred data, then computational tasks can be executed, but data movement overhead increases

Engineering Contradiction:

Improvecomputational task executionVSAvoiddata movement overhead

Core Design Contradiction:

ProductivityVSLoss of substance

Solution Approach 1:

The patent combines storage and computation resources by allowing accelerators to directly access storage devices through a shared file system, merging what were previously separate data transfer and computation operations into a unified access model that eliminates intermediate data movement

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The patent extracts the data transfer step from the computational workflow by enabling accelerators to read data directly from storage devices, removing the redundant data movement operation that previously occurred between host memory and accelerator memory

Inventive Principle:

Principle #2Taking out (Extraction)

Data Source

PatentUS20250053342A1Computational storage and networked based system

Publication Date: 2025.02.13 MICRON TECHNOLOGY INC

AI summary

Methods, systems, and apparatuses related to computational storage are described. For example, storage accessible to an accelerator may be shared between and, accessible to either of, a host and the accelerator. A computational storage system may include storage providing a portion of a shared file system accessible by a host and by accelerator logic of the computational storage system. Host interface logic may be configured to receive a storage command from the host to store data on the storage at a time the data is created. The host interface logic may be further configured to receive a storage command from the host for the accelerator logic to perform a computational task using the stored data on the storage. The accelerator logic can perform the computational task using the stored data on the storage.