Your browser does not support JavaScript!

Αρχική    Ένα Ολοκληρωμένο Κύκλωμα Μεταγωγής ATM: Αρχιτεκτονική και Προκαταρκτική Σχεδίαση  

Αποτελέσματα - Λεπτομέρειες

Προσθήκη στο καλάθι
[Προσθήκη στο καλάθι]
Κωδικός Πόρου uch.csd.msc//1991sidiropoylos
Τίτλος Ένα Ολοκληρωμένο Κύκλωμα Μεταγωγής ATM: Αρχιτεκτονική και Προκαταρκτική Σχεδίαση
Άλλος τίτλος Fast Packet Switches for Asynchronous Transfer Mode
Συγγραφέας Σιδηρόπουλος, Στέφανος
Συντελεστής Μ. Κατεβαίνης
Περίληψη Mode will form the basis for implementing the Broadband Intergated Services Digital Network of the future. This thesis presents the preliminary design of a switch chip that implements a novel flow control algorithm guaranteeing fair allocation and full utilization of bandwidth even under congestion. The switch chip is intented to be connected via its bit parallel links to other identical switch chip and various interface chips, so that it can be used as a building block for constructing composite switches of arbitrary topology and size. Its architecture is scalable up to a dozen multi-Gbit/sec links, although the first prototype is designed for only 4 bidirectional links of 400 Mbits/sec per link in each direction. The organization of the on-chip buffer memory, along with the back-pressure flow control, and the weighted round robin cell scheduling mechanisms that the chip implements in hardware, provide the network manager a set of powerful tools for tuning traffic flows and guaranteeing service performance. Full bandwidth utilization is achieved by providing dedicated buffer space to some classes of Virtual Circuits, and communication latency is reduced by using ``virtual cut-through''. The realization of this architecture under a 1\ \(*mm CMOS technology has been studied. We present the circuits that are crucial for the size and speed of the chip. These crucial circuits are primarily: the buffer memory, the input-output buffers, the content addressable memory used to select the next VC to be serviced, the circuit that schedules the accesses to the shared buffer, and the chip routing tables. We present the design and layout of the above circuits and the results of their simulation as a proof of the fact that the proposed architecture is realizable using the modern VLSI technology.
Θέμα Αρχιτεκτονική Υπολογιστών και Ψηφιακά Συστήματα
Ημερομηνία έκδοσης 1991-09-01
Ημερομηνία διάθεσης 1997-06-2
Συλλογή   Σχολή/Τμήμα--Σχολή Θετικών και Τεχνολογικών Επιστημών--Τμήμα Επιστήμης Υπολογιστών--Μεταπτυχιακές εργασίες ειδίκευσης
  Τύπος Εργασίας--Μεταπτυχιακές εργασίες ειδίκευσης
Εμφανίσεις 486

Ψηφιακά τεκμήρια
No preview available

Κατέβασμα Εγγράφου
Προβολή Εγγράφου
Εμφανίσεις : 7

No preview available

Προβολή Εγγράφου