Reconfigurable Processor Direct Storage Access Bypassing Host Memory

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Traditional data processing systems with reconfigurable processors face inefficiencies in moving data between reconfigurable processors and external storage, requiring multiple operations through host memory, leading to high latency and resource utilization.

Innovation Solution

A data processing system enabling direct access to external storage by reconfigurable processors, bypassing the host processor and memory, allowing for low latency and high bandwidth data transfer.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of time

If data is moved through host memory and processor, then data can be transferred between reconfigurable processors and external storage, but latency increases and resource utilization decreases

Engineering Contradiction:
Improvedata transfer latencyVSAvoiddata transfer path complexity
Core Design Contradiction:
Loss of timeVSDevice complexity

Solution Approach 1:

The patent extracts the data transfer function from the host processor and memory subsystem, allowing reconfigurable processors to directly access external storage. This separation removes the bottleneck where data must pass through host memory, reducing latency and freeing host resources for other tasks.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent introduces a storage controller as an intermediary component that enables direct access between reconfigurable processors and external storage. This mediator manages the data transfer path independently of the host processor, facilitating efficient data movement without requiring host intervention.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Productivity

If multiple operations are used to move data through host memory, then data can be transferred to external storage, but resource utilization decreases

Engineering Contradiction:
Improvedata transfer efficiencyVSAvoidhost processor resource consumption
Core Design Contradiction:
ProductivityVSUse of energy by moving object

Solution Approach 1:

The patent enables reconfigurable processors to perform data transfer operations independently without requiring continuous host processor involvement. The storage controller and reconfigurable processors work together to manage data movement, reducing host resource consumption and improving overall system productivity.

Inventive Principle:
Principle #25Self-service

3Speed

If direct access path is created, then latency is reduced and bandwidth is increased, but system architecture complexity increases

Engineering Contradiction:
Improvedata transfer speedVSAvoidsystem architecture complexity
Core Design Contradiction:
SpeedVSDevice complexity

Solution Approach 1:

The storage controller serves multiple functions: it manages direct data access, handles memory mapping, coordinates data transfer, and interfaces with both reconfigurable processors and external storage. This multi-functionality consolidates complexity into a single component rather than distributing it across multiple subsystems.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentUS12430272B2Direct access to external storage from a reconfigurable processor
Publication Date: 2025.09.30 SAMBANOVA SYSTEMS INC
  • US12430272B2 patent drawing
  • US12430272B2 patent drawing
  • US12430272B2 patent drawing

AI summary

A data processing system is presented that includes multiple local buses, a host processor, a network interface controller (NIC) for connecting to external storage via a network, one or more reconfigurable processors, and a bus switch. The bus switch couples the multiple local busses, thereby operatively coupling the one or more reconfigurable processors, the host processor, and the NIC. The one or more reconfigurable processors are configured to implement a virtual function that uses a virtual address for a memory access operation. The host processor is configured to implement an application programming interface (API) that translates the virtual address into a physical address, and the NIC uses the physical address to initiate a direct data access operation at the external storage that moves data directly between the one or more reconfigurable processors and the external storage, wherein the data bypasses the host processor.