Comparing Active Memory Expansion and Virtual Memory
MAR 7, 20269 MIN READ
Generate Your Research Report Instantly with AI Agent
Patsnap Eureka helps you evaluate technical feasibility & market potential.
Active Memory Expansion vs Virtual Memory Background and Goals
Memory management has evolved significantly since the early days of computing, driven by the persistent challenge of bridging the performance gap between processor speed and memory capacity. Traditional virtual memory systems emerged in the 1960s as a foundational solution, enabling programs to utilize more memory than physically available through disk-based swapping mechanisms. However, as applications have grown increasingly memory-intensive and performance-critical, the limitations of conventional virtual memory have become more pronounced.
Active Memory Expansion represents a paradigm shift in memory management philosophy, moving beyond passive disk-based swapping to intelligent, real-time memory optimization. This technology leverages advanced compression algorithms, predictive caching, and dynamic memory allocation to maximize the effective utilization of available physical memory. Unlike traditional virtual memory that relies heavily on slow disk I/O operations, Active Memory Expansion operates primarily within the memory hierarchy, utilizing faster storage tiers and sophisticated algorithms to maintain performance while expanding effective memory capacity.
The fundamental distinction lies in their operational approaches. Virtual memory systems typically employ a reactive strategy, swapping memory pages to disk when physical memory becomes scarce, often resulting in significant performance penalties. Active Memory Expansion, conversely, employs proactive strategies including real-time compression, intelligent prefetching, and adaptive memory allocation to prevent performance degradation before it occurs.
The primary technical objective of this comparative analysis centers on evaluating the effectiveness of these two approaches across multiple dimensions including latency reduction, throughput optimization, and resource utilization efficiency. Key performance indicators include memory access times, compression ratios, power consumption, and overall system responsiveness under varying workload conditions.
Contemporary computing environments demand memory solutions that can handle increasingly complex workloads while maintaining consistent performance characteristics. The goal extends beyond simple capacity expansion to encompass intelligent memory management that adapts to application behavior patterns and system resource availability. This analysis aims to establish clear performance benchmarks and identify optimal deployment scenarios for each technology, ultimately providing strategic guidance for enterprise-level memory architecture decisions in modern computing infrastructures.
Active Memory Expansion represents a paradigm shift in memory management philosophy, moving beyond passive disk-based swapping to intelligent, real-time memory optimization. This technology leverages advanced compression algorithms, predictive caching, and dynamic memory allocation to maximize the effective utilization of available physical memory. Unlike traditional virtual memory that relies heavily on slow disk I/O operations, Active Memory Expansion operates primarily within the memory hierarchy, utilizing faster storage tiers and sophisticated algorithms to maintain performance while expanding effective memory capacity.
The fundamental distinction lies in their operational approaches. Virtual memory systems typically employ a reactive strategy, swapping memory pages to disk when physical memory becomes scarce, often resulting in significant performance penalties. Active Memory Expansion, conversely, employs proactive strategies including real-time compression, intelligent prefetching, and adaptive memory allocation to prevent performance degradation before it occurs.
The primary technical objective of this comparative analysis centers on evaluating the effectiveness of these two approaches across multiple dimensions including latency reduction, throughput optimization, and resource utilization efficiency. Key performance indicators include memory access times, compression ratios, power consumption, and overall system responsiveness under varying workload conditions.
Contemporary computing environments demand memory solutions that can handle increasingly complex workloads while maintaining consistent performance characteristics. The goal extends beyond simple capacity expansion to encompass intelligent memory management that adapts to application behavior patterns and system resource availability. This analysis aims to establish clear performance benchmarks and identify optimal deployment scenarios for each technology, ultimately providing strategic guidance for enterprise-level memory architecture decisions in modern computing infrastructures.
Market Demand for Advanced Memory Management Solutions
The global memory management solutions market is experiencing unprecedented growth driven by the exponential increase in data processing requirements across industries. Enterprise applications, cloud computing platforms, and high-performance computing systems are generating massive workloads that strain traditional memory architectures. Organizations are increasingly seeking advanced memory management technologies to address performance bottlenecks, reduce operational costs, and improve system efficiency.
Data-intensive applications such as artificial intelligence, machine learning, and real-time analytics are creating substantial demand for innovative memory expansion solutions. These applications require rapid access to large datasets while maintaining cost-effectiveness, making both active memory expansion and virtual memory technologies critical components of modern computing infrastructure. The growing adoption of containerization and virtualization technologies further amplifies the need for sophisticated memory management capabilities.
Cloud service providers represent a significant market segment driving demand for advanced memory management solutions. These organizations require scalable, efficient memory architectures to support diverse workloads while optimizing resource utilization and minimizing infrastructure costs. The shift toward edge computing and distributed systems is creating additional requirements for memory management solutions that can operate effectively across heterogeneous computing environments.
Enterprise customers are increasingly prioritizing memory management solutions that offer transparent scalability, reduced total cost of ownership, and improved application performance. The market demand encompasses both hardware-based active memory expansion technologies and software-driven virtual memory optimization solutions, with organizations often seeking hybrid approaches that combine multiple techniques.
The semiconductor industry's focus on memory-centric computing architectures is creating new opportunities for advanced memory management technologies. Emerging applications in autonomous vehicles, Internet of Things devices, and augmented reality systems require memory solutions that can handle complex workloads while meeting strict latency and power consumption requirements.
Financial services, healthcare, and telecommunications sectors are driving significant demand for memory management solutions that can support real-time processing requirements while ensuring data security and regulatory compliance. These industries require memory architectures that can scale dynamically based on workload demands while maintaining consistent performance levels across different operational scenarios.
Data-intensive applications such as artificial intelligence, machine learning, and real-time analytics are creating substantial demand for innovative memory expansion solutions. These applications require rapid access to large datasets while maintaining cost-effectiveness, making both active memory expansion and virtual memory technologies critical components of modern computing infrastructure. The growing adoption of containerization and virtualization technologies further amplifies the need for sophisticated memory management capabilities.
Cloud service providers represent a significant market segment driving demand for advanced memory management solutions. These organizations require scalable, efficient memory architectures to support diverse workloads while optimizing resource utilization and minimizing infrastructure costs. The shift toward edge computing and distributed systems is creating additional requirements for memory management solutions that can operate effectively across heterogeneous computing environments.
Enterprise customers are increasingly prioritizing memory management solutions that offer transparent scalability, reduced total cost of ownership, and improved application performance. The market demand encompasses both hardware-based active memory expansion technologies and software-driven virtual memory optimization solutions, with organizations often seeking hybrid approaches that combine multiple techniques.
The semiconductor industry's focus on memory-centric computing architectures is creating new opportunities for advanced memory management technologies. Emerging applications in autonomous vehicles, Internet of Things devices, and augmented reality systems require memory solutions that can handle complex workloads while meeting strict latency and power consumption requirements.
Financial services, healthcare, and telecommunications sectors are driving significant demand for memory management solutions that can support real-time processing requirements while ensuring data security and regulatory compliance. These industries require memory architectures that can scale dynamically based on workload demands while maintaining consistent performance levels across different operational scenarios.
Current State and Challenges in Memory Expansion Technologies
Memory expansion technologies have reached a critical juncture where traditional approaches face significant scalability and performance limitations. Current implementations primarily rely on two distinct paradigms: active memory expansion through hardware-based solutions and virtual memory systems managed by operating systems. The landscape reveals substantial disparities in adoption rates, with virtual memory achieving near-universal deployment while active memory expansion remains largely experimental.
Virtual memory systems dominate contemporary computing architectures, implemented across virtually all modern operating systems and processor designs. This technology enables applications to access memory spaces exceeding physical RAM capacity through sophisticated page management algorithms. However, performance degradation becomes pronounced when systems heavily rely on disk-based swap mechanisms, creating bottlenecks that severely impact application responsiveness and overall system throughput.
Active memory expansion technologies present a more nascent but promising alternative, encompassing solutions such as memory compression, tiered memory architectures, and intelligent caching systems. These approaches demonstrate superior performance characteristics compared to traditional virtual memory swapping, yet face significant implementation barriers including hardware compatibility constraints, software ecosystem fragmentation, and substantial development costs.
Geographic distribution of technological advancement shows concentrated development in North America and East Asia, with major semiconductor manufacturers and cloud service providers driving innovation. European initiatives focus primarily on energy-efficient implementations, while emerging markets emphasize cost-effective deployment strategies for enterprise applications.
The primary technical challenges confronting both paradigms include latency optimization, power consumption management, and seamless integration with existing memory hierarchies. Active memory expansion faces additional hurdles in standardization efforts, requiring coordination among hardware manufacturers, operating system developers, and application vendors. Virtual memory systems struggle with predictive algorithms for page replacement and efficient handling of memory-intensive workloads in cloud computing environments.
Current market dynamics reveal increasing pressure for memory expansion solutions driven by artificial intelligence workloads, big data analytics, and edge computing requirements. These applications demand memory capacities that exceed traditional physical limitations while maintaining performance levels incompatible with conventional virtual memory approaches, creating urgent demand for hybrid solutions combining both technological paradigms.
Virtual memory systems dominate contemporary computing architectures, implemented across virtually all modern operating systems and processor designs. This technology enables applications to access memory spaces exceeding physical RAM capacity through sophisticated page management algorithms. However, performance degradation becomes pronounced when systems heavily rely on disk-based swap mechanisms, creating bottlenecks that severely impact application responsiveness and overall system throughput.
Active memory expansion technologies present a more nascent but promising alternative, encompassing solutions such as memory compression, tiered memory architectures, and intelligent caching systems. These approaches demonstrate superior performance characteristics compared to traditional virtual memory swapping, yet face significant implementation barriers including hardware compatibility constraints, software ecosystem fragmentation, and substantial development costs.
Geographic distribution of technological advancement shows concentrated development in North America and East Asia, with major semiconductor manufacturers and cloud service providers driving innovation. European initiatives focus primarily on energy-efficient implementations, while emerging markets emphasize cost-effective deployment strategies for enterprise applications.
The primary technical challenges confronting both paradigms include latency optimization, power consumption management, and seamless integration with existing memory hierarchies. Active memory expansion faces additional hurdles in standardization efforts, requiring coordination among hardware manufacturers, operating system developers, and application vendors. Virtual memory systems struggle with predictive algorithms for page replacement and efficient handling of memory-intensive workloads in cloud computing environments.
Current market dynamics reveal increasing pressure for memory expansion solutions driven by artificial intelligence workloads, big data analytics, and edge computing requirements. These applications demand memory capacities that exceed traditional physical limitations while maintaining performance levels incompatible with conventional virtual memory approaches, creating urgent demand for hybrid solutions combining both technological paradigms.
Existing Active and Virtual Memory Implementation Solutions
01 Memory management using page tables and address translation
Virtual memory systems utilize page tables to translate virtual addresses to physical addresses, enabling efficient memory management. This approach allows the operating system to map virtual address spaces to physical memory locations, supporting memory protection and isolation between processes. Address translation mechanisms include multi-level page tables and translation lookaside buffers (TLBs) to accelerate the translation process and improve system performance.- Memory management using page tables and address translation: Virtual memory systems utilize page tables to translate virtual addresses to physical addresses, enabling efficient memory management. This approach allows the operating system to map virtual address spaces to physical memory locations, supporting memory protection and isolation between processes. Page table structures can be hierarchical or multi-level to optimize translation speed and reduce memory overhead. Hardware components such as Translation Lookaside Buffers (TLBs) are employed to cache recent address translations and improve performance.
- Dynamic memory expansion and compression techniques: Active memory expansion involves dynamically increasing available memory through compression algorithms and intelligent data management. Memory compression techniques reduce the physical memory footprint by compressing inactive or less frequently accessed pages. These methods enable systems to effectively expand usable memory capacity without adding physical RAM. Decompression occurs transparently when compressed data is accessed, maintaining system performance while maximizing memory utilization.
- Virtual memory swapping and paging mechanisms: Swapping mechanisms transfer data between physical memory and secondary storage to manage memory constraints. Page replacement algorithms determine which memory pages to swap out when physical memory is full, optimizing for performance and minimizing page faults. These systems implement various strategies such as least recently used or clock algorithms to select pages for eviction. Prefetching and predictive loading techniques can be employed to anticipate memory access patterns and reduce latency.
- Memory pooling and resource allocation optimization: Memory pooling techniques aggregate available memory resources across multiple sources to create larger virtual memory spaces. These approaches enable efficient allocation and deallocation of memory blocks, reducing fragmentation and improving utilization. Dynamic memory allocation strategies adapt to changing workload requirements, balancing performance and resource consumption. Memory pools can be shared across processes or isolated for security and stability purposes.
- Hardware-assisted virtualization and memory management units: Modern processors incorporate specialized hardware features to accelerate virtual memory operations and support virtualization. Memory Management Units provide hardware-level address translation and protection mechanisms, reducing software overhead. Extended page tables and nested paging enable efficient virtualization by allowing hypervisors to manage guest operating system memory. Hardware support for large pages and multiple page sizes optimizes memory access patterns for different workload types.
02 Dynamic memory expansion and compression techniques
Active memory expansion involves dynamically increasing available memory through compression algorithms and memory reclamation strategies. These techniques compress inactive memory pages to create additional space without requiring physical memory installation. The system monitors memory usage patterns and selectively compresses or decompresses pages based on access frequency, balancing performance with memory availability.Expand Specific Solutions03 Memory swapping and paging mechanisms
Virtual memory systems implement swapping mechanisms to move data between physical memory and secondary storage devices. When physical memory becomes scarce, less frequently used pages are swapped out to disk storage, freeing space for active processes. Advanced paging algorithms determine which pages to swap based on usage patterns, access history, and priority levels to minimize performance impact.Expand Specific Solutions04 Memory allocation and management in virtualized environments
Virtualization platforms require sophisticated memory management to allocate and share physical memory among multiple virtual machines. These systems implement memory overcommitment, ballooning, and sharing techniques to optimize memory utilization. Memory management units coordinate between hypervisor and guest operating systems to ensure efficient resource allocation while maintaining isolation and security.Expand Specific Solutions05 Memory caching and hierarchical storage management
Multi-level memory hierarchies combine different storage technologies to balance speed and capacity. Caching mechanisms store frequently accessed data in faster memory tiers while less critical data resides in slower, larger capacity storage. Intelligent algorithms manage data placement across memory tiers, predicting access patterns and migrating data to optimize overall system performance and response times.Expand Specific Solutions
Key Players in Memory Technology and System Architecture
The active memory expansion and virtual memory technology landscape represents a mature market segment currently experiencing significant transformation driven by evolving computing demands. The industry has progressed beyond early developmental stages, with established players like IBM, Intel, AMD, and Microsoft having developed sophisticated memory management solutions over decades. Market leaders including Samsung Electronics, Micron Technology, and Huawei Technologies demonstrate advanced technical capabilities in both hardware and software implementations. Technology maturity varies across implementations, with traditional virtual memory systems being well-established while active memory expansion techniques represent emerging innovations. Companies like VMware, Oracle, and SAP focus on software-based solutions, while hardware manufacturers such as Apple and Honor integrate these technologies into consumer devices. The competitive landscape shows consolidation around major technology providers, with research institutions like Peking University contributing to theoretical advances, indicating a market transitioning from foundational development to optimization and specialized applications.
International Business Machines Corp.
Technical Solution: IBM has developed comprehensive memory management solutions that integrate both active memory expansion and virtual memory technologies. Their approach includes intelligent memory tiering systems that dynamically move data between different storage layers based on access patterns and performance requirements. IBM's Power Systems utilize advanced memory compression techniques and hardware-assisted virtualization to optimize memory utilization. Their solutions feature real-time memory analytics that can predict memory access patterns and proactively manage memory allocation between physical RAM, compressed memory pools, and storage-backed virtual memory. The company's memory management framework supports both transparent huge pages and fine-grained memory mapping, enabling efficient handling of diverse workload requirements from enterprise databases to AI applications.
Strengths: Enterprise-grade reliability and comprehensive memory analytics capabilities. Weaknesses: Higher complexity and cost compared to simpler memory management solutions.
Advanced Micro Devices, Inc.
Technical Solution: AMD's memory management approach focuses on their Infinity Cache technology combined with smart virtual memory optimization. Their solution provides large cache hierarchies that effectively bridge the gap between active memory and virtual memory systems. AMD's approach includes advanced memory compression algorithms and intelligent prefetching mechanisms that predict memory access patterns to optimize data placement. Their EPYC processors feature enhanced memory controllers that support both traditional virtual memory paging and active memory expansion through hardware-assisted compression. The system dynamically adjusts memory allocation strategies based on workload characteristics, providing seamless scaling from physical memory to virtual memory backing stores. AMD's solution particularly excels in multi-threaded applications where memory bandwidth and latency optimization are critical for performance.
Strengths: High memory bandwidth and efficient cache hierarchies for improved performance. Weaknesses: Limited ecosystem support compared to Intel and requires specific AMD processor architectures.
Core Innovations in Memory Expansion Patent Analysis
Memory expansion method and apparatus in a virtual memory system
PatentInactiveUS5317706A
Innovation
- An extended memory circuit is introduced for a virtual memory system, allowing additional memory banks to be logically mapped within the original virtual address space using an extended address control word and multiplexer to select most significant bits for bank selection, along with an extended memory refresh circuit to maintain memory integrity.
Method and apparatus for enabling physical memory larger than corresponding virtual memory
PatentInactiveUS5860141A
Innovation
- A method and apparatus that enable a physical memory space larger than the corresponding virtual memory by dividing main memory into uniquely identifiable buffers, creating unique identifiers for these buffers, and mapping them within the virtual address space, using a Least-Recently-Used (LRU) queue to manage buffer switching and library functions for interface operations between applications and the operating system.
Performance Benchmarking and Evaluation Methodologies
Performance evaluation of Active Memory Expansion (AME) and Virtual Memory systems requires comprehensive benchmarking methodologies that capture the nuanced differences between these memory management approaches. Traditional memory benchmarks often fall short when evaluating AME systems due to their unique characteristics of proactive data movement and intelligent prefetching mechanisms.
Synthetic benchmarking suites such as SPEC CPU, Stream, and LMbench provide foundational performance metrics for memory subsystem evaluation. However, these benchmarks primarily focus on sequential access patterns and may not adequately stress the adaptive capabilities of AME systems. Custom benchmark development becomes essential to evaluate AME-specific features like predictive page migration, dynamic compression ratios, and intelligent caching algorithms.
Real-world application benchmarking offers more realistic performance insights. Database workloads using TPC-C and TPC-H benchmarks effectively demonstrate how AME and virtual memory systems handle large dataset operations with varying access patterns. Similarly, machine learning workloads with frameworks like TensorFlow and PyTorch reveal performance differences in handling large model parameters and training datasets.
Memory pressure testing methodologies must account for the different response mechanisms between AME and virtual memory. While virtual memory systems rely on demand paging and swap mechanisms, AME systems employ proactive memory management strategies. Evaluation frameworks should measure not only throughput and latency but also energy consumption, as AME systems often trade computational overhead for improved memory utilization efficiency.
Latency characterization requires sophisticated measurement techniques that can capture microsecond-level differences in memory access patterns. Hardware performance counters, combined with software profiling tools, provide detailed insights into cache miss rates, page fault frequencies, and memory bandwidth utilization across different workload scenarios.
Scalability evaluation becomes particularly critical when comparing these systems under varying memory constraints and workload intensities. Multi-threaded benchmarks and concurrent access patterns help identify performance bottlenecks and optimization opportunities specific to each memory management approach.
Synthetic benchmarking suites such as SPEC CPU, Stream, and LMbench provide foundational performance metrics for memory subsystem evaluation. However, these benchmarks primarily focus on sequential access patterns and may not adequately stress the adaptive capabilities of AME systems. Custom benchmark development becomes essential to evaluate AME-specific features like predictive page migration, dynamic compression ratios, and intelligent caching algorithms.
Real-world application benchmarking offers more realistic performance insights. Database workloads using TPC-C and TPC-H benchmarks effectively demonstrate how AME and virtual memory systems handle large dataset operations with varying access patterns. Similarly, machine learning workloads with frameworks like TensorFlow and PyTorch reveal performance differences in handling large model parameters and training datasets.
Memory pressure testing methodologies must account for the different response mechanisms between AME and virtual memory. While virtual memory systems rely on demand paging and swap mechanisms, AME systems employ proactive memory management strategies. Evaluation frameworks should measure not only throughput and latency but also energy consumption, as AME systems often trade computational overhead for improved memory utilization efficiency.
Latency characterization requires sophisticated measurement techniques that can capture microsecond-level differences in memory access patterns. Hardware performance counters, combined with software profiling tools, provide detailed insights into cache miss rates, page fault frequencies, and memory bandwidth utilization across different workload scenarios.
Scalability evaluation becomes particularly critical when comparing these systems under varying memory constraints and workload intensities. Multi-threaded benchmarks and concurrent access patterns help identify performance bottlenecks and optimization opportunities specific to each memory management approach.
Energy Efficiency Considerations in Memory Expansion
Energy efficiency represents a critical differentiator between active memory expansion and virtual memory systems, with implications extending far beyond simple performance metrics. The fundamental architectural differences between these approaches create distinct energy consumption patterns that significantly impact system design decisions, particularly in mobile devices, data centers, and edge computing environments where power constraints are paramount.
Active memory expansion technologies, including Intel Optane and emerging storage-class memory solutions, demonstrate superior energy efficiency in specific workload scenarios. These systems eliminate the traditional storage hierarchy bottleneck by providing byte-addressable persistent memory with access latencies approaching DRAM levels. The energy advantage stems from reduced data movement between storage tiers and elimination of frequent page swapping operations that characterize virtual memory systems under memory pressure.
Virtual memory systems exhibit energy inefficiency primarily through their reliance on mechanical or flash-based storage devices for page swapping. Traditional hard disk drives consume substantial power during seek operations and spin-up cycles, while solid-state drives, despite lower mechanical overhead, still require significantly more energy per access compared to main memory. The energy cost multiplies when considering the CPU overhead required for page fault handling, translation lookaside buffer misses, and memory management unit operations.
The energy profile comparison reveals that active memory expansion maintains relatively consistent power consumption regardless of working set size, as data remains accessible through direct memory interfaces. Virtual memory systems demonstrate exponential energy increase as working sets exceed physical memory capacity, triggering frequent page swapping activities. This characteristic makes active memory expansion particularly attractive for memory-intensive applications with unpredictable access patterns.
However, virtual memory systems retain energy advantages in scenarios with high temporal locality and predictable access patterns. Modern virtual memory implementations incorporate sophisticated prefetching algorithms and page replacement policies that minimize unnecessary storage accesses. Additionally, the ability to completely power down unused memory pages provides energy savings opportunities unavailable in active memory expansion systems where expanded memory regions remain continuously powered.
The emergence of hybrid approaches combining both technologies presents optimal energy efficiency potential, allowing systems to dynamically allocate workloads based on access patterns and energy constraints while maintaining the flexibility and cost-effectiveness that has made virtual memory ubiquitous in modern computing systems.
Active memory expansion technologies, including Intel Optane and emerging storage-class memory solutions, demonstrate superior energy efficiency in specific workload scenarios. These systems eliminate the traditional storage hierarchy bottleneck by providing byte-addressable persistent memory with access latencies approaching DRAM levels. The energy advantage stems from reduced data movement between storage tiers and elimination of frequent page swapping operations that characterize virtual memory systems under memory pressure.
Virtual memory systems exhibit energy inefficiency primarily through their reliance on mechanical or flash-based storage devices for page swapping. Traditional hard disk drives consume substantial power during seek operations and spin-up cycles, while solid-state drives, despite lower mechanical overhead, still require significantly more energy per access compared to main memory. The energy cost multiplies when considering the CPU overhead required for page fault handling, translation lookaside buffer misses, and memory management unit operations.
The energy profile comparison reveals that active memory expansion maintains relatively consistent power consumption regardless of working set size, as data remains accessible through direct memory interfaces. Virtual memory systems demonstrate exponential energy increase as working sets exceed physical memory capacity, triggering frequent page swapping activities. This characteristic makes active memory expansion particularly attractive for memory-intensive applications with unpredictable access patterns.
However, virtual memory systems retain energy advantages in scenarios with high temporal locality and predictable access patterns. Modern virtual memory implementations incorporate sophisticated prefetching algorithms and page replacement policies that minimize unnecessary storage accesses. Additionally, the ability to completely power down unused memory pages provides energy savings opportunities unavailable in active memory expansion systems where expanded memory regions remain continuously powered.
The emergence of hybrid approaches combining both technologies presents optimal energy efficiency potential, allowing systems to dynamically allocate workloads based on access patterns and energy constraints while maintaining the flexibility and cost-effectiveness that has made virtual memory ubiquitous in modern computing systems.
Unlock deeper insights with Patsnap Eureka Quick Research — get a full tech report to explore trends and direct your research. Try now!
Generate Your Research Report Instantly with AI Agent
Supercharge your innovation with Patsnap Eureka AI Agent Platform!