TNETX3150 datasheet(41/78 Pages) TI

41 / 78 page

TNETX15AE

ADDRESS-LOOKUP DEVICE

SPWS041A – AUGUST 1997 – REVISED OCTOBER 1997

POST OFFICE BOX 655303

• DALLAS, TEXAS 75265

PRINCIPLES OF OPERATION

a graphical example of the lookup algorithm (continued)

Continuing this example, we can add 0xB1 and 0xB2 with the same number of tables allocated. This is called

the best-case scenario since the maximum amount of addresses can be stored in the minimum amount of

memory. Now add 0x22 (00.10.00.10b) to a lookup table containing only 0xB2. Figure 8 shows the results.

Figure 8. Example of Lookup-Number Algorithm (Adding 0x22)

Adding 0x22 requires allocating three additional tables. Seven tables are now required to hold two addresses.

Compared to numbers that differ in their least-significant bits, numbers that differ in their most-significant bits

require more tables. Continuing the example, adding 0x2A requires an additional three tables, as would 0xE2.

This is the worst-case scenario of storing addresses with this technique.

The TNETX15AE is designed to handle the worst-case address distribution. The worst-case address

distribution requires a separate thread per address. A purely random distribution creates multiple threads at the

top levels. The most-significant bits of a network card address are the vendor bits. However (in most networks)

there are only a couple of vendors used. These cards do not have a purely random address distribution, and

they all share a common set of bits that identifies the vendor. This configuration requires fewer pointers for the

same number of addresses. In such a network, the tables look more like those in Figure 9.

There is a need to allocate for worst case, but since the worst case is not likely to happen in a real system, the

opportunity exists to include more addresses than indicated by the worst-case rating. The actual number of

addresses supported depends on the nature of the nodes in the network. TNETX15AE devices in networks with

nodes from one or a small number of manufacturers can recognize more addresses than those in a purely

random-address network.

This algorithm has the primary advantage that the lookup time is independent of the number of addresses stored

in the lookup table. Whether the number of addresses present is one or a half million, the lookup time depends

on the number of table levels required to match the address.

TNETX3150 数据表(PDF) 41 Page - Texas Instruments