Recent years have witnessed phenomenal growth in the computational capabilities and
applications of GPUs. However, this trend has also led to a dramatic increase in their power …

Despite industrial investment in both on-die GPUs and next generation interconnects, the
highest performing parallel accelerators ship** today continue to be discrete GPUs …

Systems from smartphones to supercomputers are increasingly heterogeneous, being
composed of both CPUs and GPUs. To maximize cost and energy efficiency, these systems …

Non-volatile memory (NVM) provides a scalable and power-efficient solution to replace
DRAM as main memory. However, because of relatively high latency and low bandwidth of …

The state-of-the-art approximate nearest neighbor search (ANNS) algorithms face a
fundamental tradeoff between query latency and accuracy, because of small main memory …

For extreme-scale high-performance computing systems, system-wide power consumption
has been identified as one of the key constraints moving forward, where DRAM main …

Spintronic memories are promising candidates for future on-chip storage due to their high
density, non-volatility and near-zero leakage. However, the energy consumed by read and …

Historically, GPU-based HPC applications have had a substantial memory bandwidth
advantage over CPU-based workloads due to using GDDR rather than DDR memory …

Graphics Processing Units (GPUs) have traditionally relied on the host CPU to initiate
access to the data storage. This approach is well-suited for GPU applications with known …

Non-volatile memory (NVM) provides a scalable solution to replace DRAM as main memory.
Because of relatively high latency and low bandwidth of NVM (comparing with DRAM), NVM …