CrystalGPU: Transparent and Efficient Utilization of GPU Power
Abdullah Gharaibeh, Samer Al-Kiswany, Matei Ripeanu
Abstract
Open-access reader
Abdullah Gharaibeh, Samer Al-Kiswany, Matei Ripeanu
Abstract
Open-access reader
General-purpose computing on graphics processing units (GPGPU) has recently gained considerable attention in various domains such as bioinformatics, databases and distributed computing. GPGPU is based on using the GPU as a co-processor accelerator to offload computationally-intensive tasks from the CPU. This study starts from the observation that a number of GPU features (such as overlapping communication and computation, short lived buffer reuse, and harnessing multi-GPU systems) can be abstracted and reused across different GPGPU applications. This paper describes CrystalGPU, a modular framework that transparently enables applications to exploit a number of GPU optimizations. Our evaluation shows that CrystalGPU enables up to 16x speedup gains on synthetic benchmarks, while introducing negligible latency overhead.
OpenAlex reports 7 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
General-purpose computing on graphics processing units (GPGPU) has recently gained considerable attention in various domains such as bioinformatics, databases and distributed computing. GPGPU is based on using the GPU as a co-processor accelerator to offload computationally-intensive tasks from the CPU. This study starts from the observation that a number of GPU features (such as overlapping communication and computation, short lived buffer reuse, and harnessing multi-GPU systems) can be abstracted and reused across different GPGPU applications. This paper describes CrystalGPU, a modular framework that transparently enables applications to exploit a number of GPU optimizations. Our evaluation shows that CrystalGPU enables up to 16x speedup gains on synthetic benchmarks, while introducing negligible latency overhead.
Key concepts: Computer science, Parallel computing, General-purpose computing on graphics processing units, CUDA, Power (physics), Computational science, Computer graphics (images), Graphics