A5056531806- Parallel Computing and Optimization Techniques
- Advanced Data Storage Technologies
- Advanced Memory and Neural Computing
- Interconnection Networks and Systems
- Distributed systems and fault tolerance
- Cloud Computing and Resource Management
- Opportunistic and Delay-Tolerant Networks
- Energy Efficient Wireless Sensor Networks
- Mobile Ad Hoc Networks
- Ferroelectric and Negative Capacitance Devices
- Caching and Content Delivery
- Graphene research and applications
- Pharmacy and Medical Practices
University of Central Florida
2017-2021
As demands for memory-intensive applications continue to grow, the memory capacity of each computing node is expected grow at a similar pace. In high-performance (HPC) systems, per compute decided upon most demanding application that would likely run on such system, and hence average in future HPC systems significantly. However, since many with different demands, large percentage overall will be underutilized; modules can thought as private its corresponding node. Thus, are moving towards...
Memristors are emerging Non-Volatile Memories (NVMs) that promising for building future memory systems. Unlike DRAM, memristors non-volatile, i.e., they can retain data after power loss. In contrast to DRAM where each cell is associated with a pass transistor, memristor cells be implemented without such and hence enable high density ReRAM Moreover, leverage unique crossbar architecture improve the of modules. have been considered build centers both energy-efficiency capacity goals....
The exponential growth of data has driven technology providers to develop new protocols, such as cache coherent interconnects and memory semantic fabrics, help users facilities leverage advances in technologies satisfy these growing storage demands. Using fabric-attached memories (FAM) can be directly attached a system interconnect easily integrated with variety processing elements (PEs). Moreover, systems that support FAM smoothly upgraded allow multiple PEs share the pools using...
Many applications have growing demands for memory, particularly in the HPC space, making memory system a potential bottleneck of next-generation computing systems. Sharing across processor sockets and nodes becomes compelling argument given that technology is scaling at slower rate than technology. Moreover, as many rely on shared data, e.g., graph database workloads, having large number accessing allows efficient use resources avoids duplicating huge files, which can be infeasible graphs or...
Many modern applications have memory footprints that are increasingly large, driving system capacities higher and higher. However, due to the diversity of run on High-Performance Computing (HPC) systems, utilization can fluctuate widely from one application another, which results in underutilization issues when there many jobs with small footprints. Since chips collocated compute nodes, this necessitates need for message passing APIs be able share information between nodes.
Non-volatile memories (NVMs) have the characteristics of both traditional storage systems (persistent) and memory (byte-addressable). However, they suffer from high write latency a limited endurance. Researchers proposed hybrid that combine DRAM NVM, utilizing lower to hide some shortcomings NVM - improving system's performance by caching resident data in DRAM. this can nullify persistency cached pages, leading question trade-offs terms reliability. In paper, we propose Stealth-Persist,...
With many recent advances in interconnect technologies and memory interfaces, disaggregated systems are approaching industrial adoption. For instance, the Gen-Z consortium focuses on a new semantic protocol that enables fabric-attached memories (FAM), where other compute units can be directly attached to fabric interconnects. Decoupling of from becomes feasible option as rate data transfer increases due emergence novel technologies, such Silicon Photonic Interconnects.
The exponential growth of data has driven technology providers to develop new protocols, such as cache coherent interconnects and memory semantic fabrics, help users facilities leverage advances in technologies satisfy these growing storage demands. Using fabric-attached memories (FAM) can be directly attached a system interconnect easily integrated with variety processing elements (PEs). Moreover, systems that support FAM smoothly upgraded allow multiple PEs share the pools using...