Cache Optimization With Shared CPU Cryptography for Direct IO Access

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current systems enabling direct cache access for external devices face inefficiencies due to the need for separate cryptographic engines per IO port, which limits confidentiality to point-to-point links and increases power and space requirements, while existing CPU cryptographic engines are not utilized for IO operations.

Innovation Solution

Implementing a cryptographic engine configured to operate in a direct cache access mode, which determines the destination of data transactions and bypasses memory writes when necessary, enabling direct data transfer to CPU cache or IO, and using message type identifiers to manage data delivery.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If separate cryptographic engines are implemented per IO port to enable direct cache access, then data confidentiality is improved, but power consumption and device space requirements increase

Engineering Contradiction:
Improvedata confidentialityVSAvoidpower consumption
Core Design Contradiction:
ReliabilityVSUse of energy by moving object

Solution Approach 1:

The patent makes the CPU cryptographic engine universal by enabling it to handle both traditional memory encryption/decryption operations and direct cache access operations. Instead of having dedicated cryptographic engines for each IO port, the single CPU cryptographic engine is configured to service multiple IO ports and handle encryption/decryption for both memory and direct cache access paths, thereby reducing overall power consumption while maintaining data confidentiality.

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

2Reliability

If separate cryptographic engines are implemented per IO port to enable direct cache access, then data confidentiality is improved, but device space requirements increase

Engineering Contradiction:
Improvedata confidentialityVSAvoiddevice space
Core Design Contradiction:
ReliabilityVSArea of stationary object

Solution Approach 1:

The CPU cryptographic engine is designed to perform multiple functions: it handles encryption and decryption for traditional memory access paths as well as for direct cache access paths from multiple IO ports. This multi-functional design eliminates the need for separate cryptographic engines at each IO port, thereby reducing the overall device space required while maintaining security.

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

Solution Approach 2:

The patent merges the functions of multiple separate cryptographic engines (one per IO port) into a single shared CPU cryptographic engine. By combining these redundant cryptographic functions into one centralized resource, the device space required for cryptographic operations is significantly reduced while the security functionality is preserved across all IO ports.

Inventive Principle:
Principle #5Merging (Combining)

3Reliability

If traditional memory access path is used with cryptographic operations, then data confidentiality is maintained, but performance deteriorates due to multiple encryption/decryption operations

Engineering Contradiction:
Improvedata confidentialityVSAvoidperformance
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The patent segments the data access path into two distinct modes: traditional memory access path and direct cache access path. For time-critical or performance-sensitive operations, data can be transferred directly to cache with fewer cryptographic operations, while less time-sensitive data can use the traditional memory path. This segmentation allows the system to optimize performance for critical paths without compromising the security of less critical paths.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

For direct cache access operations, the patent enables the data to skip the intermediate memory storage step and go directly from the IO port to the CPU cache. This skipping of the memory write/read cycle reduces the number of cryptographic operations required (from up to ten down to eight), thereby improving performance for time-critical data transfers while maintaining security through the CPU cryptographic engine.

Inventive Principle:
Principle #21Skipping (Rushing through)

Data Source

PatentUS12380036B2Cache optimization mechanism
Publication Date: 2025.08.05 INTEL CORP
  • US12380036B2 patent drawing
  • US12380036B2 patent drawing
  • US12380036B2 patent drawing

AI summary

An apparatus includes a central processing unit (CPU), including a plurality of processing cores, each having a cache memory, a fabric interconnect coupled to the plurality of processing cores and cryptographic circuitry, coupled to the fabric interconnect including mesh stop station to receive memory data and determine a destination of the memory data and encryption circuitry to encrypt/decrypt the memory data based on a destination of the memory data.