Search

Terrain topography controls insolation variations at catchment scale. This effects are known to be important in cold and mountainous regions due to high diurnal and seasonal variability in incoming radiation. However, meteorological data in such areas lacks accuracy due to sparse station network and coarse re-analysis grids. Simulation tools that model hydrologic processes at local scales require ways to overcome the lack of accuracy in the observational data, particular at high elevations, so downscaling to cell level is done carefully. With an introduction of irregular triangular networks into distributed hydrologic modelling framework Shyft, steps are taken to account for hillslope-scale terrain structures within scope of operational hydrology. This new functionality allows translation of radiation measurements or re-analysis data onto inclined surfaces improving the predictive power of the model. Based on the Shyft and Rasputin toolbox we show importance of topographic details such as slope and aspect on predicting snowmelt and rainfall-runoff simulations in snow-covered mountainous region of Himalaya. We conduct the series of experi- ments for catchments in Narayani area of central Nepal in two steps. First, we demonstrate sensitivity of streamflow simulation to mesh size and shape. The results show that there is an upper limit after which further mesh refinement is not useful for simulations. Second, we incorporate "on the fly" correction of incoming solar radiation depending on surface inclination. The experiments with both coarse and fine tin meshes demonstrate that snow-water equivalent and potential evapotranspiration are directly affected by variations in insolation, with less snow and more evaporation on south-facing slopes. Finally, we perform 10-years of hydrological simulation of Budhi-Gandaki catchment with several model configuration incorporating both pro- posed features (tins and radiation correction), where we reveal better correspondence of simulated and observed discharge forgrid-based solution, which is contradictory to our previous study at Marsyangdi-2 catchment.This is an unfinished study: the Budhi-Gandaki catchment hydrology has to be analyzed carefully. One of the possible reasons of the unexpected results is a complex precipitation pattern in the catchment, which was not caught by the modelproperly.

Brain-inspired event-based processors have attracted considerable attention for edge deployment because of their ability to efficiently process Convolutional Neural Networks (CNNs) by exploiting sparsity. On such processors, one critical feature is that the speed and energy consumption of CNN inference are approximately proportional to the number of non-zero values in the activation maps. Thus, to achieve top performance, an efficient training algorithm is required to largely suppress the activations in CNNs. We propose a novel training method, called Adaptive-Regularization Training Schedule (ARTS), which dramatically decreases the non-zero activations in a model by adaptively altering the regularization coefficient through training. We evaluate our method across an extensive range of computer vision applications, including image classification, object recognition, depth estimation, and semantic segmentation. The results show that our technique can achieve 1.41 × to 6.00 × more activation suppression on top of ReLU activation across various networks and applications, and outperforms the state-of-the-art methods in terms of training time, activation suppression gains, and accuracy. A case study for a commercially-available event-based processor, Neuronflow, shows that the activation suppression achieved by ARTS effectively reduces CNN inference latency by up to 8.4 × and energy consumption by up to 14.1 ×.

Machine learning, and specifically Deep Neural Networks (DNNs) impact all parts of daily life. Although DNNs can be large and compute intensive, requiring processing on big servers (like in the cloud), we see a move of DNNs into loT-edge based systems, adding intelligence to these systems. These systems are often energy constrained and too small for satisfying the huge DNN computation and memory demands. DNN model quantization may come to the rescue. Instead of using 32-bit floating point numbers, much smaller formats can be used, down to 1-bit binary numbers. Although this largely may solve the compute and memory problems, it comes with a huge price, model accuracy reduction. This problem spawned a lot of research into model repair methods, especially for binary neural networks. Heavy quantization triggers a lot of debate; we even see some movements of going back to higher precision using brainfloats. This paper therefore evaluates the trade-off between energy reduction through extreme quantization versus accuracy loss. This evaluation is based on ResNet-I8 with the ImageNet dataset, mapped to a fully programmable architecture with special support for 8-bit and 1-bit deep learning, the BrainTTA. We show that, after applying repair methods, the use of extremely quantized DNNs makes sense. They have superior energy efficiency compared to DNNs based on 8-bit precision of weights and data, while only having a slightly lower accuracy. There is still an accuracy gap, requiring further research, but results are promising. A side effect of the much lower energy requirements of BNNs is that external DRAM becomes more dominant. This certainly requires further attention.

Current Scanning Electron Microscopy (SEM) acquisition techniques are far too slow to capture large volumes in a feasible time. One solution is to use low-dose and sparse imaging. By computationally denoising and inpainting an image with acceptable quality can be approximated. This approach, however, requires significant compute resources. Therefore, this paper proposes CELR, a framework, that hides the computationally expensive workload of reconstructing low-dose sparse SEM images, such that (delayed) live reconstruction is possible. Live reconstruction is possible by using Convolutional Neural Networks (CNNs) that approximate a classical reconstruction algorithm like GOAL. The reconstruction by CNNs is done locally, while recurring training of CNNs is done in the cloud. Moreover, training labels are generated by GOAL in the cloud. Next to the framework, this paper evaluates and optimizes the CNN reconstruction throughput by employing Nvidia's TensorRT. This paper also touches upon open research questions about on-the-fly CNN training. The combination of CELR and TensorRT enables large volume acquisitions with a dwell-time of 1μs and 10% pixel coverage to be reconstructed on a single GPU.

Exposed Datapath Architectures (EDPAs) with aggressively pruned data-path connectivity, where not all function units in the design have connections to a centralized register file, are promising solutions for energy-efficient computation. A direct bypassing of data between function units without temporary copies to the register file is a prime optimization for programming such architectures. However, traditional compiler frameworks, such as LLVM, assume function-units connect to register-files and allocate all live variables in register-files. This leads to schedule inefficiencies in terms of instruction-level parallelism and reg-ister accesses in the EDPAs. To address these inefficiencies, we propose Prebypass; a new optimization pass for EDPA compiler backends. Experimental results on an EDPA class of architecture, Transport- Triggered Architecture, show that Prebypass improves the runtime, register reads, and register writes up to 16%, 26 %, and 37 % respectively, when the datapath is extremely pruned. Evaluation in a 28-nm FDSOI technology reveals that Prebypass improves the core-level Energy by 17.5 % over the current heuristic scheduler. acceptedVersion