Abstract-With the SIMT execution model, GPUs can hide memory latency through massive multithreading for many applications that have regular memory access patterns. To support applications with irregular memory access patterns, cache hierarchies have been introduced to GPU architectures to capture temporal and spatial locality and mitigate the effect of irregular accesses. However, GPU caches exhibit poor efficiency due to the mismatch of the throughput-oriented execution model and its cache hierarchy design, which limits system performance and energy-efficiency.The massive amount of memory requests generated by GPUs cause cache contention and resource congestion. Existing CPU cache management policies that are designed for multicore systems, can be suboptimal when directly applied to GPU caches. We propose a specialized cache management policy for GPGPUs. The cache hierarchy is protected from contention by the bypass policy based on reuse distance. Contention and resource congestion are detected at runtime. To avoid oversaturating on-chip resources, the bypass policy is coordinated with warp throttling to dynamically control the active number of warps. We also propose a simple predictor to dynamically estimate the optimal number of active warps that can take full advantage of the cache space and on-chip resources. Experimental results show that cache efficiency is significantly improved and on-chip resources are better utilized for cachesensitive benchmarks. This results in a harmonic mean IPC improvement of 74% and 17% (maximum 661% and 44% IPC improvement), compared to the baseline GPU architecture and optimal static warp throttling, respectively.

No abstract

No abstract

With the rapid development of household photovoltaics and electric vehicles, demand-side energy management has become an important means to release the burden of power grid during the load peaking period. In order to ensure the user comfort and reduce the cost of electricity, a multi-time scale home energy management method is proposed based on mixed integer programming algorithm. Firstly, on the basis of time-of-use electricity price, household photovoltaic, electric vehicles, storage batteries and HVAC are taken into consideration. And then, short time scale model of HVAC is adopted, which increases the rationality of modeling while discretizing. The simulation results verify the superiority of multi-time scale and the optimization effect of the proposed method, which can reduce the cost of electricity for users, smooth the load curve and improve the utilization efficiency of renewable energy.

No abstract