Significance X-ray fluorescence microscopy provides unparalleled sensitivity for measuring the distribution of trace elements in many-micrometer-thick specimens, whereas ptychography offers a path to imaging weakly fluorescing biological ultrastructure at beyond-focusing-optic resolution. We demonstrate here first time, our knowledge, combination and frozen-hydrated specimens cryogenic temperatures, with excellent structural chemical preservation. This combined approach will have significant...

Ptychography is an imaging method whereby a coherent beam scanned across object, and image obtained by iterative phasing of the set diffraction patterns. It able to be used extended objects at resolution limited scattering strength object detector geometry, rather than optics-imposed limit. As technical advances allow larger fields imaged, computational challenges arise for reconstructing correspondingly data volumes, yet same time there also need deliver reconstructed images immediately so...

Ptychography is a coherent diffraction imaging (CDI) method for extended objects in which patterns are acquired sequentially from overlapping illumination spots. The object's complex transmission function can be reconstructed those at spatial resolution limited only by the scattering strength of object and detector geometry. Most experiments to date have positioned spots on sample using move-settle-measure sequence move settle steps take longer complete than measure step. We describe here...

Unmatched computation and storage performance in new HPC systems have led to a plethora of I/O optimizations ranging from application-side collective network disk-level request scheduling on the file system side. As we deal with ever larger machines, interference produced by multiple applications accessing shared parallel concurrent manner becomes major problem. Interference often breaks single-application optimizations, dramatically degrading application and, as result, lowering machine...

As we move toward the exascale era, performance variability in HPC systems remains a challenge. I/O interference, major cause of this variability, is becoming more important every day with growing number concurrent applications that share larger machines. Earlier research efforts on mitigating interference focus single potential (e.g., network). Yet root causes can be diverse. In work, conduct an extensive experimental campaign to explore various storage systems. We use microbenchmarks...

The increasing gap between the computation performance of post-petascale machines and their I/O subsystem has motivated many optimizations including prefetching, caching, scheduling techniques. In order to further improve these techniques, modeling predicting spatial temporal patterns HPC applications as they run became crucial. this paper we present Omnisc'IO, an approach that builds a grammar-based model behavior uses it predict when future operations will occur, where how much data be...

Dedicated cluster parallel computers (DCPCs) are emerging as low-cost high performance environments for many important applications in science and engineering. A significant class of that perform well on a DCPC coarse-grain involve large amounts file I/O. Current research systems distributed is providing mechanism adapting these to the environment. We present Parallel Virtual File System (PVFS), system provides disk striping across multiple nodes computer partitioning among tasks program....

On current large-scale HPC platforms the data path from compute nodes to final storage passes through several networks interconnecting a distributed hierarchy of serving as nodes, I/O and file system servers. Although applications compete for resources at various levels, software offers no mechanisms globally coordinating flow attaining optimal resource usage reacting overload or interference. In this paper we describe CLARISSE, middleware designed enhance data-staging coordination control...

In-system solid state storage is expected to be an important component of the I/O subsystem on first exascale platforms, as it has potential reduce DRAM requirements, increase system reliability, and smooth loads.

X-ray fluorescence offers unparalleled sensitivity for imaging the nanoscale distribution of trace elements in micrometer thick samples, while x-ray ptychography an approach to image light element containing structures at a resolution beyond that lens used. These methods can be used combination, and continuous scan mode rapid data acquisition when using multiple probe reconstruction methods. We discuss here opportunities limitations making use additional information provided by improve...

The increasing gap between the computation performance of post-petascale machines and their I/O subsystem has motivated many optimizations including prefetching, caching, scheduling. In order to further improve these techniques, modeling predicting spatial temporal patterns HPC applications as they run become crucial. this paper we present Omnisc'IO, an approach that builds a grammar-based model behavior uses it predict when future operations will occur, where how much data be accessed. To...

As high-performance computing increases in popularity and performance, the demand for similarly capable input output systems rises. Parallel I/O takes advantage of many data server machines to provide linearly scaling performance parallel applications that access storage over system area network. The demands placed on network by a are considerably different than those imposed message-passing algorithms or data-center operations; and, there popular varied networks use modern machines. These...

In situ analysis and visualization have grown increasingly popular for enabling direct access to data from high-performance computing (HPC) simulations. As a simulation progresses interesting physical phenomena emerge, however, the produced may become complex, users need dynamically change type scale of in tasks being carried out consequently adapt amount resources allocated such tasks. To date, none production frameworks offer an elasticity feature, good reason: assumption that number...

Distributed data storage services tailored to specific applications have grown popular in the high-performance computing (HPC) community as a way address I/O and challenges. These offer variety of interfaces, semantics, representations. They also expose many tuning parameters, making it difficult for their users find best configuration given workload platform. To this issue, we develop novel variational-autoencoder-guided asynchronous Bayesian optimization method tune HPC service parameters....

Two-tiered direct network topologies such as Dragonflies have been proposed for future post-petascale and exascale machines, since they provide a high-radix, low-diameter, fast interconnection network. Such call redesigningMPI collective communication algorithms in order to attain the best performance. Yet increasingly more applications share machine, it is not clear how these topology-aware will react interference with concurrent jobs accessing same In this paper, we study three broadcast...

Beowulf workstations have become a popular choice for high-end computing in number of application domains. One the key building blocks parallel applications on is message passing library. While there are library implementations available use workstations, as yet none been specifically tailored to this new, unique architecture. Thus it important evaluate existing packages order determine how these perform environment. This paper examines set four libraries focusing their features,...

In recent years, more and applications in cloud have needed to process large-scale on-line data sets that evolve over time as entries are added or modified. Several programming frameworks, such Percolator Oolong, proposed for incremental processing can achieve efficient updates with an event-driven abstraction. However, these frameworks inherently asynchronous, leaving the heavy burden of managing synchronization developers. Such a limitation significantly restricts their usability. this...

In 2003, the High End Computing Revitalization Task Force designated file systems and I/O as an area in need of national focus. The purpose Interagency Working Group (HECIWG) is to coordinate government spending on File Systems (FSIO) R&D by all agencies that are involved Computing. HECIWG tasked a smaller advisory group list, categorize, prioritize HEC needs. 2005, leaders FSIO from academia, industry collaborated list areas research FSIO. This led very successful University Research...