We present the design and first performance and usability evaluation of GeMTC, a novel execution model and runtime system that enables accelerators to be programmed with many concurrent and independent tasks of potentially short or variable duration. With GeMTC, a broad class of such "many-task" applications can leverage the increasing number of accelerated… (More)
—Current software and hardware limitations prevent ManyTask Computing (MTC) from leveraging hardware accelerators (NVIDIA GPUs, Intel MIC) boasting Many-Core Computing architectures. Some broad application classes that t the MTC paradigm are work ows, MapReduce, highthroughput computing, and a subset of high-performance computing. MTC emphasizes using many… (More)
—Many-Task Computing (MTC) aims to bridge the gap between HPC and HTC. MTC emphasizes running many computational tasks over a short period of time, where tasks can be either dependent or independent of one another. MTC has been well supported on Clouds, Grids, and Supercomputers on traditional computing architectures, but the abundance of hybrid large-scale… (More)
This work aims to enable efficient dynamic memory management on NVIDIA GPUs by u?lizing a sub-‐allocator between CUDA and the programmer. This work enables Many-‐Task Compu?ng applica?ons, which need to dynamically allocate parameters for each task, to run efficiently on GPUs. • Improve the worst case ?me complexity of malloc and free opera?ons. Currently… (More)
—This work aims to enable Swift to efficiently use accelerators (such as NVIDIA GPUs) to further accelerate a wide range of applications. This work presents preliminary results in the costs associated with managing and launching concurrent kernels on NVIDIA Kepler GPUs. We expect our results to be applicable to several XSEDE resources, such as Forge,… (More)
—Accelerator technologies are now quite common in Supercomputers, Clusters, Grids, and personal desktops. This work aims to provide an overview of the current technologies that are available today, and examine future accelerator technologies. This work examines the 3 major competitors in the Accelerator market including: NVIDIA, Intel, and AMD.
Current software and hardware limitations prevent Many-Task Computing (MTC) workloads from leveraging hardware accelerators boasting Many-Core Computing architectures. This work aims to address the programmability gap between MTC and accelerators, through the innovative CUDA middleware GeMTC. By working at the warp level, GeMTC enables heterogeneous task… (More)
—This work analyzes the performance increases gained from enabling Swift applications to utilize the GPU through the GeMTC Framework. By identifying computationally intensive portions of Swift applications, we can easily turn these code blocks into GeMTC microkernels. Users can then call these microkernels throughout the lifetime of their Swift application.… (More)
aims to address the programmability gap between MTC and accelerators, through an innovative middleware that enables MIMD programmability of SIMD architectures. This work will enable a broader class of applications to leverage the growing number of accelerated high-end computing systems.
—Effective use of parallel and distributed computing in science depends upon multiple interdependent entities and activities that form an ecosystem. Active engagement between application users and technology catalysts is a crucial activity that forms an integral part of this ecosystem. Technology catalysts play a crucial role benefiting communities beyond a… (More)