Increasing Cluster Throughput while Reducing Energy Consumption for GPU Workloads Federico Silla Technical University of Valencia

First Page		Document Content
Date: 2015-03-23 04:22:15 Computing GPGPU Computer architecture Computer engineering Video cards Parallel computing Coprocessors Graphics hardware Graphics processing unit RCUDA InfiniBand Nvidia		Increasing Cluster Throughput while Reducing Energy Consumption for GPU Workloads Federico Silla Technical University of Valencia Add to Reading List Source URL: hpcadvisorycouncil.com Download Document from Source Website File Size: 3,75 MB Share Document on Facebook

	rCUDA Frequently Asked Questions What is rCUDA? rCUDA (Remote CUDA) is a middleware that allows CUDA applications to use GPUs located in remote nodes without any modifications to original CUDA applications. DocID: 18Yyz - View Document
	Microsoft PowerPoint - rCUDA_HPC-Council_European_2013_final [Modo de compatibilidad] DocID: 18Xzn - View Document
	Improving the Efficiency of your GPU-accelerated Cluster with rCUDA F. Silla Technical University of Valencia Spain DocID: 18TRv - View Document
	Extending SLURM with support for rCUDA Why extending SLURM to support rCUDA? Previous to this extension, SLURM did not support the virtual GPUs provided by rCUDA or any other remote GPU virtualization framework. That is, DocID: 18GOB - View Document
	rCUDA: Virtualizing GPUs to reduce cost and improve performance Federico Silla Technical University of Valencia Spain DocID: 18Gc1 - View Document