Κατάλογος Εκδηλώσεων

29
Ιουλ

Παρουσίαση Μεταπτυχιακής Εργασίας κ. Εμμανουήλ Πισσαδάκη, Σχολή ΗΜΜΥ
Κατηγορία: Παρουσίαση Μεταπτυχιακής Εργασίας   ΗΜΜΥ  
ΤοποθεσίαΛ - Κτίριο Επιστημών/ΗΜΜΥ, 145Π-58, Πολυτεχνειούπολη
Ώρα29/07/2019 10:00 - 11:00

Περιγραφή:

Σχολή Ηλεκτρολόγων Μηχανικών και Μηχανικών Υπολογιστών

Πρόγραμμα Μεταπτυχιακών Σπουδών

 

ΠΑΡΟΥΣΙΑΣΗ ΜΕΤΑΠΤΥΧΙΑΚΗΣ ΕΡΓΑΣΙΑΣ

ΕΜΜΑΝΟΥΗΛ ΠΙΣΣΑΔΑΚΗ

 

με θέμα

Design and Implementation of a Framework for Efficient Remote Reconfigurable Accelerator Deployment in Disaggregated Environment

 

 Δευτέρα 29 Ιουλίου 2019, 10 π.μ.

Αίθουσα 145.Π58, Κτίριο Επιστημών, Πολυτεχνειούπολη

 

Εξεταστική Επιτροπή

Καθηγητής Διονύσιος  Πνευματικάτος (επιβλέπων)

Αναπληρωτής Καθηγητής Βασίλειος Σαμολαδάς

 Αναπληρωτής Καθηγητής Ιωάννης Παπαευσταθίου (ΑΠΘ)

 

​​​​​​​Abstract

Cloud computing usage has drastically increased over the years, providing data security and  privacy  which are  prime concerns these days. The scalability of the cloud capacity, as well as the  accessibility of the provided services, constitute relevant factors for the cloud computing evolution.

 Data centers  are  mainly  deployed  as cloud computing resources to deal with  large  storage  and computation requirements. The need for specialized hardware acceleration in this domain is well established and intensified by the insatiable demand for compute power. Hardware accelerators can also provide high energy efficiency for many application domains in comparison with current  architectures based on general purpose processors.   The fixed amount of  the available resources constitutes the major disadvantage of traditional data centers. Resource disaggregation  alleviates this issue while  offering the opportunity to manage  resources more efficiently. In disaggregated computing environments, where all data transfers between remote nodes are realized via packet exchanges over a rack-scale network, reducing communication and synchronization is a prerequisite to the effective employment of remote acceleration. To this end, this thesis presents ReFiRe (Remote Fine-grained Reconfigurable acceleration), a generic deployment framework with native support for partial reconfiguration that allows  considerable reduction of communication needs between a processor and remote accelerators. Custom instructions that encapsulate complex sequences of operations and their respective synchronization requirements   deployed for shifting   control flow and partial reconfiguration decisions to the remote side. Considering the high complexity of the instruction-initialization procedure, a source-to-source transformation framework based on  the ReFiRe infrastructure  was further implemented. Through this framework, these instructions are automatically generated according to  application requirements   transparently to the  user level.

To evaluate  ReFiRe, three  benchmark applications were employed. A 2D-FFT algorithm , a genomics application that detects positive selection in genomes and a  Binarized Neural Network,  demonstrate that offloading computations to remote accelerators using ReFiRe leads to superior aggregate performance on the same specialized hardware platform compared to using dedicated accelerator calls on a per-operation basis.

© Σχολή Ηλεκτρολόγων Μηχανικών & Μηχανικών Υπολογιστών 2014
Πολυτεχνείο Κρήτης