search.noResults

search.searching

saml.title
dataCollection.invalidEmail
note.createNoteMessage

search.noResults

search.searching

orderForm.title

orderForm.productCode
orderForm.description
orderForm.quantity
orderForm.itemPrice
orderForm.price
orderForm.totalPrice
orderForm.deliveryDetails.billingAddress
orderForm.deliveryDetails.deliveryAddress
orderForm.noItems
HIGH PERFORMANCE COMPUTING


Edge productivity


RESEARCHERS HAVE CREATED A SOFTWARE FRAMEWORK TO INCREASE PRODUCTIVITY OF DEVELOPERS WORKING WITH EDGE COMPUTING, IN A SERIES OF EUROPEAN- FUNDED PROJECTS


Researchers from Barcelona Supercomputing Center (BSC) are working on three projects to make


edge computing easier to use and more performant and productive for developers. The work initially focused on the


development of a novel big-data software architecture to enable the development, deployment and execution of advanced analytics services through edge-to-cloud computation, featuring AI, data-in-motion and data-at-rest analytics methods for the efficient processing of vast amounts of geographically distributed data. This framework is being applied to


different use cases through the European projects Class, Elastic and Ampere, co- ordinated by Eduardo Quiñones, senior researcher at the BSC. While the projects do share similar goals


in increasing productivity for developers using edge computing resources,


www.scientific-computing.com | @scwmagazine


Class and Elastic focus on distributed environments. More specifically, scenarios that can be found in smart cities where there is not a single computing unit. Instead, the computing capabilities of the city are distributed across multiple computing elements. These elements may be located at the edge-side, in cabinets at street level, tram stops or bus stops and the near edge/fog nodes and in the cloud. The Ampere project focuses on


developing a framework for computing resources that are needed in cyber- physical systems (CPS) in which the number of computing elements does not have a wide distribution. These systems are all concentrated in a more close environment, for example, in a single connected vehicle.


Promoting productivity ‘For Class and Elastic we are developing a software architecture which will facilitate the development, deployment and execution of what we call complex data analytics workflows into this distributed environment – what we call the compute continuum,’ said Quiñones. ‘You could have data analytics at the edge-level that are responsible for collecting data from a series of cameras. The result of this analytics is then sent to a near-edge computing node responsible


for collecting the results of other edge devices and combining this to augment the information that is available. This information can be uploaded to the cloud providing offline analysis, or it can be sent back again to the edge to implement real- time analysis.’ The added complexity for these


projects is how to implement this efficient distribution of analytics workflows in a way that clears the complexity of the underlying resources from the developer. ‘How can we facilitate designers to describe this complex data analytics workflow without knowing the complexities of the compute continuum underneath?’ notes Quiñones. ‘In the case of Ampere, we are trying


to follow similar objectives, but, in this case, we are addressing heterogeneous computing platforms,’ Quiñones added. ‘Platforms comprised of multicore processors and GPUs, multicore processors and FPGAs and many-core fabrics connected to a host device. Here the objective is how we can describe our systems in such a way that we are hiding the complexities of these platforms, while still exploiting all of the computing capabilities,’ added Quiñones.


“This technology means authorities and city infrastructure can ‘predict trajectories of all objects the camera and connected car can see’. You can detect collisions before they happen”


Winter 2021 Scientific Computing World 13


g


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34