Instruction Control for Early Cache Reads Under Address Dependency

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The speed of memory access and storage has not kept pace with the increasing computing power of processors, leading to bottlenecks in data supply and limiting processing efficiency due to address dependency relationships between memory access instructions.

Innovation Solution

An instruction control apparatus and method that decodes memory access instructions, caches them, and transmits them early to a data cache if dependency on the destination end storage circuit exists, sending blocking signals to manage data access and release, allowing read operations to precede write operations.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If memory access instructions are executed sequentially with address dependency checking, then data accuracy is ensured, but processing efficiency deteriorates due to waiting for previous instructions to complete

Engineering Contradiction:
Improveprocessing efficiencyVSAvoidwaiting time for dependency release
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The patent applies preliminary action by allowing the read operation of a memory access instruction to be executed in advance before the write operation of the preceding instruction completes. The instruction transmission controller transmits the current memory access instruction to the data cache unit early, and the data cache unit performs the read operation ahead of time, storing data in a blocked state. This eliminates the waiting time for dependency release while ensuring data accuracy is maintained through the blocking mechanism that prevents premature data use.

Inventive Principle:
Principle #10Preliminary action

2Speed

If data reading is performed in advance to hide memory access delay, then execution speed improves, but data accuracy may deteriorate if data is used before write operation completes

Engineering Contradiction:
Improveexecution speed of memory access instructionVSAvoiddata accuracy
Core Design Contradiction:
SpeedVSReliability

Solution Approach 1:

The patent introduces a blocking signal as an intermediary mechanism between the early read operation and data usage. The instruction transmission controller sends a blocking signal to the data cache unit when transmitting the current memory access instruction. This blocking signal acts as a mediator that allows the read operation to complete and data to be stored in the data cache unit, but prevents the data from being used or forwarded until the write operation completes and the blocking signal is cleared. This resolves the contradiction by enabling early execution while guaranteeing data accuracy.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Reliability

If memory access instructions wait for dependency release before execution, then data dependency accuracy is maintained, but data supply speed deteriorates

Engineering Contradiction:
Improvedata dependency accuracyVSAvoiddata supply speed
Core Design Contradiction:
ReliabilityVSSpeed

Solution Approach 1:

The patent applies preliminary action by separating the read operation from the write operation in time. The instruction transmission controller transmits the current memory access instruction to the data cache unit in advance, allowing the read operation to execute before the preceding instruction's write operation completes. The data cache unit stores the read data in a blocked state, maintaining data dependency accuracy. Meanwhile, the early execution of the read operation hides the memory access delay, significantly improving data supply speed to meet processor requirements.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentUS20260064418A1Instruction control method, data caching method, and related products
Publication Date: 2026.03.05 SHANGHAI CAMBRICON INFORMATION TECH CO LTD
  • US20260064418A1 patent drawing
  • US20260064418A1 patent drawing
  • US20260064418A1 patent drawing

AI summary

This disclosure discloses an instruction control method, instruction control apparatus, processor, chip, and board card. The processor can be included as a computing apparatus in a combination processing apparatus, which may also include interface devices and other processing apparatus. The computing apparatus interacts with other processing apparatuses to jointly complete computing operations specified by the user. The combined processing apparatus may also include a storage apparatus, which is connected to the computing apparatus and other processing apparatuses respectively, for storing data of the computing apparatus and other processing apparatuses. The disclosed solution provides an instruction control method that can enhance instruction level parallelism and improve processing efficiency.