Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Method for dynamic timeslot allocation for packet relay

IP.com Disclosure Number: IPCOM000008928D
Publication Date: 2002-Jul-24
Document File: 7 page(s) / 126K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a method for dynamic timeslot allocation for packet relay. Benefits include improved performance, improved functionality, and improved design flexibility.

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 33% of the total text.

Method for dynamic timeslot allocation for packet relay

Disclosed is a method for dynamic timeslot allocation for packet relay. Benefits include improved performance, improved functionality, and improved design flexibility.

Background

              Packet length specified in the packet header of a packet relay implementation is variable. The crossbar must relay from and to any combination of lx and 4x ports. The variable packet length, port configurability, and the requirement for high performance increase packet relay design complexity.

              A relay must be cut-through capable. Two virtual channels of a port share a datapath through the crossbar. A single crossbar datapath has enough bandwidth to relay at a 4x link rate. The result is an opportunity for 1x source ports to share datapath bandwidth and make good use of the excess bandwidth available through the crossbar. Together, these facts push the design toward a packet relay scheme that dynamically allocates crossbar bandwidth on a word-by-word basis.

General description

              The disclosed method is dynamic timeslot allocation for packet relay for a cut-through capable switch with multiple VLs, variable packet length, and configurable ports for either 1x or 4x links. The switch utilizes both transmit and receive RAM on either side of the crossbar.

              The method includes a logic implementation that achieves optimal crossbar bandwidth utilization for both lx and 4x packets by dynamically allocating crossbar bandwidth on a word-by-word basis. The elements of the disclosed method are:

·        Receive RAM resource arbitration unit

·        Transmit RAM resource arbitration unit

·        Global timeslot counter

·        Timeslot allocation unit

·        Request logic and link rate lookup

·        Grant generation and distribution

Advantages

·        The disclosed method provides advantages, including:

·        Improved performance due to the improved allocation of crossbar bandwidth

·        Improved performance due to improved throughput

·        Improved functionality due to the dynamic granting of timeslots and to increase and throttle transfer activity

·        Improved design flexibility

Detailed description

              The disclosed method dynamically allocates timeslots for packet relay and uses RAM on either side of the crossbar. The economical use of RAM dictates that multiple virtual channels share RAM. The switch utilizes both transmit and receive RAM on either side of the crossbar (see Figure 1). Each RAM and each crossbar datapath accommodates the 4x link speed that is four times the speed required to relay data from a lx port.

              An important part of the disclosed method is that the crossbar bandwidth is partitioned into time slices (timeslots). Each timeslot is an opportunity to relay one word through each of the crossbar data-paths. Because the lx link rate is one-fourth the bandwidth available in the crossbar it is useful to conceptualize a grouping of four timeslots and name them timeslot 0, timeslot 1, timeslot 2, timeslot 3. A 4x packet uses all four timeslots (0-3) to relay at the 4x link...