Browse Prior Art Database

Distributed Memory Mapping

IP.com Disclosure Number: IPCOM000014788D
Original Publication Date: 2000-Oct-01
Included in the Prior Art Database: 2003-Jun-20
Document File: 2 page(s) / 56K

Publishing Venue

IBM

Abstract

Disclosed is a distributed memory mapping scheme applicable to multiprocessor systems. The objective of the scheme is to achieve fast and efficient mapping operations in a multiprocessor environment in which memory contained in separate memory nodes is mapped in a common address space. The scheme is illustrated in figure 1. This figure shows a diagram of a multiprocessor system that contains two types of elements, processor nodes and memory nodes. 1 Figure 1: Distributed memory mapping scheme.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 53% of the total text.

Page 1 of 2

Distributed Memory Mapping

Disclosed is a distributed memory mapping scheme applicable to multiprocessor systems. The objective of the scheme is to achieve fast and efficient mapping operations in a multiprocessor environment in which memory contained in separate memory nodes is mapped in a common address space.

The scheme is illustrated in figure 1. This figure shows a diagram of a multiprocessor system that contains two types of elements, processor nodes and memory nodes.

1

[This page contains 1 picture or other non-text object]

Page 2 of 2

Figure 1: Distributed memory mapping scheme.

A mapping operation is now performed in the following way. If a CPU contained in a processor node, wants to perform an access to a memory bank that is located in a memory node, it will generate the address of the memory location to be accessed. Two segments denoted as X and Y are extracted from this address and are used to index a lookup table LUT1 in order to obtain a bank number of the memory bank that contains the memory location that is to be accessed. Based on this bank number an access request that includes the original address or part of it, is routed over the interconnect to the memory node that corresponds to that bank number. This part of the mapping operation can be performed in parallel in all processor nodes.

Upon reception of an access request, the address contained in that request is used to determine the internal bank address of the memory location within the memor...