Search for: All records

While High Performance Computing systems are increasingly based on heterogeneous cores, their eectiveness depends on how well the scheduler can allocate workloads onto appropriate computing devices and how communication and computation can be overlapped. With dierent types of resources integrated into one system, the complexity of the scheduler correspondingly increases. Moreover, for applications with varying problem sizes on dierent heterogeneous resources, the optimal scheduling approach may vary accordingly. We thus present PDAWL, an event-driven pro le-based Iterative Dynamic Adaptive Work-Load balance scheduling approach to dynamically and adaptively adjust workload to eciently utilize heterogeneous resources. It combines online scheduling (DAWL), which can adaptively adjust workload based on available real time heterogeneous resources, with an oine machine learning (pro lebased estimation model) which can build a device-speci c communication computation estimation model. Our scheduling approach is tested on control-regular applications, Stencil kernel (based on a Jacobi Algorithm) and Sparse Matrix-Vector Multiplication (SpMV) in an event-driven runtime system. Experimental results show that PDAWL is either on-par or far outperforms whichever yields the best results (CPU or GPU). 

An x-ray Fresnel diffractive radiography platform was designed for use at the National Ignition Facility. It will enable measurements of micron-scale changes in the density gradients across an interface between isochorically heated warm dense matter materials, the evolution of which is driven primarily through thermal conductivity and mutual diffusion. We use 4.75 keV Ti K-shell x-ray emission to heat a 1000 μm diameter plastic cylinder, with a central 30 μm diameter channel filled with liquid D2, up to 8 eV. This leads to a cylindrical implosion of the liquid D2 column, compressing it to ∼2.3 g/cm3. After pressure equilibration, the location of the D2/plastic interface remains steady for several nanoseconds, which enables us to track density gradient changes across the material interface with high precision. For radiography, we use Cu He-α x rays at 8.3 keV. Using a slit aperture of only 1 μm width increases the spatial coherence of the source, giving rise to significant diffraction features in the radiography signal, in addition to the refraction enhancement, which further increases its sensitivity to density scale length changes at the D2/plastic interface.