scholarly journals Usage of GPUs in ALICE Online and Offline processing during LHC Run 3

2021 ◽  
Vol 251 ◽  
pp. 04026
Author(s):  
David Rohr
Keyword(s):  
Second Phase ◽  
Full System ◽  
Raw Data ◽  
Huge Data ◽  
Processing Demands ◽  
Gpu Processing ◽  
Speed Up ◽  
Asynchronous Processing ◽  
Offline Processing ◽  
Gpu Architecture

ALICE will significantly increase its Pb–Pb data taking rate from the 1 kHz of triggered readout in Run 2 to 50 kHz of continuous readout for LHC Run 3. Updated tracking detectors are installed for Run 3 and a new twophase computing strategy is employed. In the first synchronous phase during the data taking, the raw data is compressed for storage to an on-site disk buffer and the required data for the detector calibration is collected. In the second asynchronous phase the compressed raw data is reprocessed using the final calibration to produce the final reconstruction output. Traditional CPUs are unable to cope with the huge data rate and processing demands of the synchronous phase, therefore ALICE employs GPUs to speed up the processing. Since the online computing farm performs a part of the asynchronous processing when there is no beam in the LHC, ALICE plans to use the GPUs also for this second phase. This paper gives an overview of the GPU processing in the synchronous phase, the full system test to validate the reference GPU architecture, and the prospects for the GPU usage in the asynchronous phase.

scholarly journals Deep Learning Strategies for ProtoDUNE Raw Data Denoising

2022 ◽  
Vol 6 (1) ◽  
Author(s):  
Marco Rossi ◽  
Sofia Vallecorsa
Keyword(s):  
Neural Networks ◽  
Deep Learning ◽  
Learning Strategies ◽  
Simulation Data ◽  
Raw Data ◽  
Digital Detector ◽  
Speed Up ◽  
Neural Network Hardware ◽  
Graph Neural Networks ◽  
High Level

AbstractIn this work, we investigate different machine learning-based strategies for denoising raw simulation data from the ProtoDUNE experiment. The ProtoDUNE detector is hosted by CERN and it aims to test and calibrate the technologies for DUNE, a forthcoming experiment in neutrino physics. The reconstruction workchain consists of converting digital detector signals into physical high-level quantities. We address the first step in reconstruction, namely raw data denoising, leveraging deep learning algorithms. We design two architectures based on graph neural networks, aiming to enhance the receptive field of basic convolutional neural networks. We benchmark this approach against traditional algorithms implemented by the DUNE collaboration. We test the capabilities of graph neural network hardware accelerator setups to speed up training and inference processes.

scholarly journals Speed up of Fresnel transforms for digital holography using pre-computed chirp and GPU processing

2009 ◽  
Author(s):  
Nitesh Pandey ◽  
Damien P. Kelly ◽  
Thomas J. Naughton ◽  
Bryan M. Hennelly
Keyword(s):  
Digital Holography ◽  
Gpu Processing ◽  
Speed Up

scholarly journals Local Alignment Tool Based on Hadoop Framework and GPU Architecture

2014 ◽  
Vol 2014 ◽  
pp. 1-7 ◽  
Author(s):  
Che-Lun Hung ◽  
Guan-Jie Hua
Keyword(s):  
High Availability ◽  
Local Alignment ◽  
Sequencing Technologies ◽  
Computational Performance ◽  
Huge Data ◽  
Alignment Tool ◽  
Gpu Architectures ◽  
Gpu Architecture ◽  
Generation Sequencing ◽  
Hadoop Framework

With the rapid growth of next generation sequencing technologies, such as Slex, more and more data have been discovered and published. To analyze such huge data the computational performance is an important issue. Recently, many tools, such as SOAP, have been implemented on Hadoop and GPU parallel computing architectures. BLASTP is an important tool, implemented on GPU architectures, for biologists to compare protein sequences. To deal with the big biology data, it is hard to rely on single GPU. Therefore, we implement a distributed BLASTP by combining Hadoop and multi-GPUs. The experimental results present that the proposed method can improve the performance of BLASTP on single GPU, and also it can achieve high availability and fault tolerance.

Initial Attitude Acquisition in-orbit Status of ALSAT-2A Remote Sensing Satellite

2020 ◽  
Vol 19 ◽  
Keyword(s):  
Angular Momentum ◽  
Attitude Determination ◽  
Second Phase ◽  
Knowing That ◽  
Third Phase ◽  
Solar Local Time ◽  
Speed Up ◽  
Initial Attitude ◽  
Synchronous Orbit ◽  
Small Rotation

- In this paper we present the in orbit performance during the initial attitude acquisition, immediately after separation from the final stage of the launcher, until the satellite converges toward a sun pointing. On 12th july 2010 ALSAT-2A microsatellite was launched into a 670 km sun synchronous orbit, with a solar local time at an ascending node of 22h15. In the initial acquisition mode or even in the safe mode, only sun sensors and magnetometer are used for attitude determination. Knowing that, the satellite once is separated from the launcher it starts tumbling. So, as to detumble the satellite, a strategy consisting of three phases is set. The first phase consists on reducing the velocity using only the magnetorquers, until reaching the threshold angular momentum of 0.05 Nms, once this is done, the second phase is automatically enabled in such a way, the four reaction wheels are ON and speed up to get an angular momentum of -0.15 Nms along the satellite x-axis. At the end of this phase the satellite is completely detumbled and a third phase is carried out, in which the satellite minus X axis is pointed toward sun with a small rotation.

Scalable Parallel Motion Estimation on Muti-GPU System

2013 ◽  
Vol 347-350 ◽  
pp. 3708-3714 ◽  
Author(s):  
Dong Chen ◽  
Hua You Su ◽  
Wen Mei ◽  
Li Xuan Wang ◽  
Chun Yuan Zhang
Keyword(s):  
Parallel Computing ◽  
Motion Estimation ◽  
Parallel Algorithm ◽  
Estimation Algorithm ◽  
Video Encoding ◽  
Communication Model ◽  
Linear Speedup ◽  
Parallel Motion ◽  
Speed Up ◽  
Gpu Architecture

With NVIDIA’s parallel computing architecture CUDA, using GPU to speed up compute-intensive applications has become a research focus in recent years. In this paper, we proposed a scalable method for multi-GPU system to accelerate motion estimation algorithm, which is the most time consuming process in video encoding. Based on the analysis of data dependency and multi-GPU architecture, a parallel computing model and a communication model are designed. We tested our parallel algorithm and analyzed the performance with 10 standard video sequences in different resolutions using 4 NVIDIA GTX460 GPUs, and calculated the overall speedup. Our results show that a speedup of 36.1 times using 1 GPU and more than 120 times for 4 GPUs on 1920x1080 sequences. Further, our parallel algorithm demonstrated the potential of nearly linear speedup according to the number of GPUs in the system.

scholarly journals Rapid glacier sliding, reverse ice motion and subglacial water pressure during an autumn rainstorm

2009 ◽  
Vol 50 (52) ◽  
pp. 101-108 ◽  
Author(s):  
T.J. Fudge ◽  
J.T. Harper ◽  
N.F. Humphrey ◽  
W.T. Pfeffer
Keyword(s):  
Water Pressure ◽  
Drainage System ◽  
Second Phase ◽  
Length Scales ◽  
System Evolution ◽  
Rapid Motion ◽  
Motion Event ◽  
Basal Cavity ◽  
Velocity Relationship ◽  
Speed Up

AbstractMeasurements of basal water pressure from 15 boreholes located at both local (tens of meters) and regional (kilometers) length scales were used to elucidate the pressure/sliding relationship during an autumn rapid motion event on Bench Glacier, Alaska, USA. The 8 day event had two distinct phases, each with a ten-fold speed-up with respect to winter velocity. The water pressure in all 15 boreholes varied synchronously during the speed-up. The first phase of rapid sliding began after a peak in basal water pressure and continued while the pressure was elevated and stable, or decreasing. The second phase of rapid sliding occurred when the basal water pressure was low but increasing, and terminated before the pressure peaked. Pressure and velocity do not appear unrelated, but the pressure/sliding relationship was not consistently linked to increasing, decreasing or a critical water pressure. The pressure variations and sliding accelerations are a response to a warm rainstorm, although equally large input events occurred in weeks prior with no apparent response. Drainage system evolution therefore appears to play a key role in both the acceleration and the pressure/velocity relationship. Basal cavity dynamics are likely responsible for three episodes of reverse (up-valley) motion observed after enhanced sliding.

First stage of the French vertical flow constructed wetland system: experiments with the reduction of surface area and number of units

2014 ◽  
Vol 5 (1) ◽  
pp. 50-55 ◽  
Author(s):  
E. S. Manjate ◽  
L. C. O. Lana ◽  
D. C. Moraes ◽  
G. R. Vasconcellos ◽  
G. R. M. Maciel ◽  
...  
Keyword(s):  
Constructed Wetland ◽  
Oxygen Demand ◽  
Feeding Strategies ◽  
Attractive Alternative ◽  
Second Phase ◽  
Vertical Flow ◽  
Full System ◽  
Mean Values ◽  
Vertical Flow Constructed Wetland ◽  
Removal Efficiencies

Small vertical-flow constructed wetland units comprising the first stage of the French system were studied in Brazil for the treatment of raw sewage. Planted and unplanted units and different feeding strategies were tested. In the first phase, hourly batches of a daily flow of 13 m3 d−1 were applied over three alternating units, resulting in an average hydraulic loading rate (HLR) on the full system of 0.15 m3 m−2 d−1. A second phase, aimed at reducing land requirements, kept the same daily flow and batch frequency, but used two alternating units, resulting in a HLR on the full system of 0.22 m3 m−2 d−1. Removal efficiencies were very good when the system operated with three units, with mean values of 82% for biochemical oxygen demand (BOD), 81% for chemical oxygen demand (COD), 85% for total suspended solids (TSS) and 59% for NH4+-N. With two units, the equivalent values were 74% for BOD, 59% for COD, 67% for TSS and 51% for NH4+-N. There were significant differences in the median removal efficiencies of COD and TSS. No significant differences were found between planted and unplanted units for most constituents. In both phases, the overall good performance and the simplicity of the system make this treatment process a very attractive alternative for developing countries.

scholarly journals Ingest pipeline for ASKAP

2020 ◽  
Vol 245 ◽  
pp. 01038
Author(s):  
Maxim Voronkov
Keyword(s):  
Real Time ◽  
Lessons Learned ◽  
Small Cluster ◽  
Raw Data ◽  
Square Kilometre Array ◽  
Scientific Results ◽  
Offline Processing ◽  
New Generation ◽  
Real Time Application

The Australian Square Kilometre Array Pathfinder (ASKAP) is a new generation 36-antenna 36-beam interferometer capable of producing about 2.2 Gb/s of raw data. The data are streamed from the observatory directly to the dedicated small cluster at the Pawsey Supercomputing centre. The ingest pipeline is a distributed real time software which runs on this cluster and prepares the data for further (offline) processing by imaging and calibration pipelines. In addition to its main functionality, it turned out to be a valuable tool for various commissioning experiments and allowed us to run an interim system to achieve the first scientific results much earlier than otherwise possible. I will review the architecture of the ingest pipeline, its role in the overall ASKAP design as well as the lessons learned by developing a real-time application in the HPC environment.

scholarly journals GGU Open File Series: a review of reports related to Greenland mineral exploration

1993 ◽  
Vol 159 ◽  
pp. 22-24
Author(s):  
H.K Schønwandt
Keyword(s):  
Mineral Exploration ◽  
Geological Survey ◽  
Raw Data ◽  
Open File ◽  
Analytical Results ◽  
New Information ◽  
Data Field ◽  
Speed Up ◽  
File Series

In 1989 the Geological Survey of Greenland (GGU) increased its publication range by the introduction of an Open File Series, the aim of which was to speed up the presentation of new information to industry (Ghisler, 1990). This unedited series presents primarily raw data (field and analytical results) allowing exploration companies to use the information in assessment and planning. Although the series was formalised in 1989, a handful of reports of specific relevance to mineral exploration, and released by GGU up to three years previously, were numbered as the first six titles of the series. At the end of 1992, 40 reports made up the series.

Sign in / Sign up

Export Citation Format

Share Document