CPU-GPU Collaborative Memory Stress Test for LPDDR

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current memory test methods for LPDDR SDRAM, such as MBIST, Uboot, and Android/Linux-based tests, face limitations in scalability, processing speed, and comprehensive memory coverage, with IO access and GPU testing deficiencies, leading to suboptimal stress testing and memory evaluation.

Innovation Solution

A memory test method involving a CPU and GPU collaboration, where the CPU runs a test program to access the memory through a bus, controlling the GPU to perform memory access based on its accessible capacity, utilizing OpenCL or OpenGL interfaces to maximize memory bus occupancy and implement a high-access-load stress test.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional memory test methods (MBIST, Uboot, Android/Linux-based tests) are used, then the testing process is simple to implement, but the memory coverage and stress testing effectiveness are insufficient

Engineering Contradiction:
Improvememory test effectivenessVSAvoidtest system complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent combines CPU and GPU into a collaborative testing system where both processors work simultaneously on memory testing. The CPU executes test programs while the GPU performs parallel memory operations, creating a merged testing architecture that achieves comprehensive memory coverage and high-stress testing while maintaining reasonable system complexity through coordinated operation.

Inventive Principle:
Principle #5Merging (Combining)

2Productivity

If CPU alone is used for memory testing, then the test program execution is straightforward, but the memory bus occupancy and access load are insufficient

Engineering Contradiction:
Improvememory bus occupancyVSAvoidprocessing architecture complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent introduces a new dimension of parallel processing by adding GPU operations to the traditional CPU-based testing approach. The GPU executes memory access operations in parallel with the CPU, utilizing a different processing dimension that simultaneously increases memory bus occupancy and access load without simply scaling up the CPU workload.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

3Reliability

If high stress testing is implemented, then the memory evaluation comprehensiveness is improved, but the test time and processing overhead increase

Engineering Contradiction:
Improvememory evaluation comprehensivenessVSAvoidtest execution time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent implements continuous memory testing by having both CPU and GPU perform memory operations simultaneously without idle periods. The GPU executes memory access tasks in parallel with the CPU throughout the entire test duration, ensuring continuous utilization of the memory bus and eliminating wasted test time while maintaining comprehensive evaluation.

Inventive Principle:
Principle #20Continuity of useful action

Data Source

PatentUS11860748B2Memory test method, memory test apparatus, device and storage medium
Publication Date: 2024.01.02 CHANGXIN MEMORY TECH INC
  • US11860748B2 patent drawing
  • US11860748B2 patent drawing
  • US11860748B2 patent drawing

AI summary

A memory test method, a memory test apparatus, a device and a storage medium are provided. The memory test method includes: obtaining a central processing unit (CPU) accessible space of a memory to-be-tested; obtaining a graphics processing unit (GPU) accessible space of the memory to-be-tested; and driving a CPU to run a test program based on the accessible space of the CPU, to access the memory to-be-tested through a bus of memory to-be-tested, when the CPU runs the test program, the CPU controls a GPU to access the memory to-be-tested based on the accessible space of the GPU through the bus of memory to-be-tested.