Amazon / GPU / The times / Wordware Publishing Inc. / IEEE Computer Society Press / MultiCore Systems / NVIDIA / Microsoft / Intel / /
Country
United States / / /
Facility
arch Tesla / /
IndustryTerm
heterogeneous devices / image processing pipelines / text mining workloads / front-end programming tools / important parallel applications / concurrent applications / hardware/software boundary / heterogeneous processors / simplified data processing / computer vision algorithms / heterogeneous systems / streaming applications / programmable devices / combinatorial graph bisection algorithm / deep belief network / cell broadband engine processor / make systems / deep belief neural network / real-time constraints / Optical flow algorithms / image-processing pipeline / considerable systems / energy function / neural network / image processing / graphics processors / Liquid metal / brute force search / bisection algorithm / context-dependent deep networks / cluster-as-service systems / opencl devices / computing / Compute-bound algorithms / distributed computing clusters / attractive solution / gpus as compute devices / /
OperatingSystem
Windows Server 2008 / Microsoft Windows / /
Person
A. Prokopec / V / Graph Clone / Adam Eversole / M. Isard / V / Maestro / Christopher J. Rossbach / H. Chafi / V / /
Position
high fidelity model for the partitioner / RT / starss programming model for platforms with multiple gpus / General / model / greedy scheduler / representative / DA FIFO Hand Heur Opt RT / scheduler / programming model for the cell BE architecture / programmer / na¨ıve scheduler / /
Scheduling Dataflow Execution Across Multiple Accelerators Jon Currey, Adam Eversole, and Christopher J. Rossbach Microsoft Research ABSTRACT