2,157 research outputs found
Elastic-DF: Scaling Performance of DNN Inference in FPGA Clouds through Automatic Partitioning
No full textCustomized compute acceleration in the datacenter is key to the wider roll-out of applications based on deep neural network (DNN) inference. In this article, we investigate how to maximize the performance and scalability of field-programmable gate array (FPGA)-based pipeline dataflow DNN inference accelerators (DFAs) automatically on computing infrastructures consisting of multi-die, network-connected FPGAs. We present Elastic-DF, a novel resource partitioning tool and associated FPGA runtime infrastructure that integrates with the DNN compiler FINN. Elastic-DF allocates FPGA resources to DNN layers and layers to individual FPGA dies to maximize the total performance of the multi-FPGA system. In the resulting Elastic-DF mapping, the accelerator may be instantiated multiple times, and each instance may be segmented across multiple FPGAs transparently, whereby the segments communicate peer-to-peer through 100 Gbps Ethernet FPGA infrastructure, without host involvement. When applied to ResNet-50, Elastic-DF provides a 44% latency decrease on Alveo U280. For MobileNetV1 on Alveo U200 and U280, Elastic-DF enables a 78% throughput increase, eliminating the performance difference between these cards and the larger Alveo U250. Elastic-DF also increases operating frequency in all our experiments, on average by over 20%. Elastic-DF therefore increases performance portability between different sizes of FPGA and increases the critical throughput per cost metric of datacenter inference. Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Computer Engineerin
Hydrodynamical turbulence by fractal fourier decimation
No full textWe present a systematic numerical investigation of high-resolution 3D isotropic and homogeneous turbulence resolved on a decimated set of Fourier modes. Fractal decimation acts to decrease the effective dimensionality of the flow by allowing triadic interactions only in a set of Fourier modes N(k) proportional to k^DF for large k. While keeping the symmetries of the original 3D Navier-Stokes equations unchanged, a dramatic change in small-scale statistics is detected at decreasing the fractal dimension DF . Already at fractal dimension DF = 2.8, a global self-similar behaviour is observed in the inertial range of scales, the consequence of such transition are the restoration of the scaling symmetry and vorticity distribution that becomes close to Gaussian. We relate the results to the different roles of local vs non-local interactions in the energy transfer range
'Response by the author, Daniel F. Vukovich.'
No full textResponse by the author (Vukovich) to a review of Illiberal China (my 2019 monograph
DF AND HF LASER SPECTRA
No full textW. B. Roh and K. Narahari Rao, J. Mol. Spectrosc. 49, 317 (1974).""Author Institution: Department of Physics, The Ohio State University; Department of Electrical Engineering, The Ohio State UniversityA pulsed laser was used as a source to study DF and HF infrared laser spectra with a grating spectrometer. The rotational structure observed for the 1-0, 2-1, and 3-2 bands of DF, and the 1-0 and 2-1 bands of HF has been measured by following a procedure similar to that used for the CO laser Molecular constants derived by combining these laser measurements with other available data will be presented
OPTOACOUSTIC MEASUREMENT OF DF LASER ABSORPTION BY METHANE
No full textAuthor Institution:An optoacoustic system has been constructed for measurement of absorption of DF laser radiation. Absorption by methane in Argon and Nitrogen has been measured at 15 DF laser lines in the 3.6 – 4.0 region. Construction and calibration of the spectrophone will be discussed
Comparison of system architecture and converter topology for a solar powered electric vehicle charging station
No full textAccepted Author ManuscriptOld - EWI-ESE-DC&S DC systems & StoragePhotovoltaic Materials and Device
Simulation of a surface-transverse wave (STW) biosensor for DF-1 cells
No full textA 250 MHz Surface-Transverse Wave (STW) resonator is employed as a sensor element for the
detection of DF-1 cells. STW belongs to the shear-horizontal acoustic plate modes (SH-APM) waves’
family where it has attracted plenty considerable interest. STWs are horizontally polarized shear waves
which are generated and detected by the interdigital transducers (IDTs) similar to surface-acoustic wave
(SAW) resonators [1]. Detection of chemical and biological agents in aqueous solutions is a difficult
problem, especially when the detection technique has to be sensitive, power-efficient and very handy.
Acoustic plate mode is a mode of vibration where particle motion is parallel to the surface. This makes it
possible to produce a sensitive sensor capable of operating in fluids [2]. This paper presents the biosensor
prototype utilizing STW resonator
Period changes of two contact binaries: DF Hya and WZ And
No full text33rd International Physics Congress of the Turkish-Physical-Society (TPS) -- SEP 06-10, 2017 -- Bodrum, TURKEYOrbital period variations of two contact binaries DF Hya and WZ And are analyzed with the least-squares method by using all available minima times. It is shown that the period variations of these systems are due mainly to the LightTime Effect (LITE) due originates from gravitational influence of a third body. New LITE elements such as, orbital periods and minimum masses of possibility third bodies are given.Turkish Phys So
Image processing and analysis methods in quantitative endothelial cell biology
No full textThis thesis details the development of computerised image processing and analysis pipelines for quantitative evaluation of microscope image data acquired in endothelial vascular biology experimentation. The overarching objective of this work was to advance our understanding of the cell biology of cardiovascular processes; principally involving haemostasis, thrombosis, and inflammation. Bioinformatics techniques are increasingly necessary to extract and evaluate information from biological experimentation. In cell biology advances in microscopy and the increased acquisition of large scale digital image data sets have created a need for automated image processing and data analysis. The development, testing, and evaluation of three computerised workflows for analysis of microscopy images investigating cardiovascular cell biology are described here. The first image analysis pipeline extracts morphometric features from high-throughput experiments imaging endothelial cells and organelles. Segmentation of endothelial cells and their organelles followed by extraction of morphometric features provides a rich quantitative data set to investigate haemostatic mechanisms. A second image processing workflow was applied to platelet images obtained from super-resolution microscopy, and used in a proof-of-principle study of a new platelet dense-granule deficiency diagnostic method. The method was able to efficiently differentiate between healthy volunteers and three patients with Hermansky-Pudlak syndrome. This was achieved by segmenting and counting the number of CD63-positive structures per platelet, allowing for the differentiation of patients from control volunteers with 99\% confidence. The final workflow described is a video analysis method that quantifies interactions of leukocytes with an endothelial monolayer. Phase contrast microscopy videos were analysed with a Haar-like features object detection and custom tracking method to quantify the dynamic interaction of rolling leukocytes. This technique provides much more information than a manual evaluation and was found to give a tracking accuracy of 92\%. These three methodologies provide a toolkit to further biological understanding of multiple facets of cardiovascular behaviour
Delivery of bioactive molecules to the mitochondrial genome using a membrane-fusing, liposome-based carrier, DF-MITO-Porter.
No full textMitochondrial dysfunction has been implicated in a variety of human diseases. It is now well accepted that mutations and defects in the mitochondrial genome form the basis of these diseases. Therefore, mitochondrial gene therapy and diagnosis would be expected to have great medical benefits. To achieve such a strategy, it will be necessary to deliver therapeutic agents into mitochondria in living cells. We report here on an approach to accomplish this via the use of a Dual Function (DF)-MITO-Porter, aimed at the mitochondrial genome, so-called mitochondrial DNA (mtDNA). The DF-MITO-Porter, a nano carrier for mitochondrial delivery, has the ability to penetrate the endosomal and mitochondrial membranes via step-wise membrane fusion. We first constructed a DF-MITO-Porter encapsulating DNase I protein as a bioactive cargo. It was expected that mtDNA would be digested, when the DNase I was delivered to the mitochondria. We observed the intracellular trafficking of the carriers, and then measured mitochondrial activity and mtDNA-levels after the delivery of DNase I by the DF-MITO-Porter. The findings confirm that the DF-MITO-Porter effectively delivered the DNase I into the mitochondria, and provides a demonstration of its potential use in therapies that are selective for the mitochondrial genome
- …
