LetterWise has been ported to several 8-bit processors, including the Intel 80C51 and the Mitsubishi 740. A port to the 4-bit Winbond W742(C)816 is in progress.

These and similar CPUs are the workhorse processors at the heart of numerous electronic devices, especially low-cost and/or low power consumption devices. These include remote controls, toys, cordless phones, televisions, in-car navigation system, and CD players.

In every case, LetterWise functions have had runtime well below required minimums and memory consumption is miserly. For example:

- On the Mitsubishi 740, the LetterWise engine occupies a mere 651 bytes and uses 24 bytes of RAM.

- LetterWise language databases can occupy as little as a few hundred bytes.

- In one 80C51 implementation, LetterWise and six languages (Dutch, English, French, German, Italian, Spanish) were provided in under 5K of space.

- On the 80C51, the main LetterWise function (called between keystrokes) completes in 7-10 milliseconds.

- LetterWise uses at most 4 stack frames, consuming, at worst, 184 bytes of stack memory.

- LetterWise performance degrades gracefully with reduced sized database. Simplistic dictionary-based methods (such as T9 and eZiText) are forced to discard entire words: to halve the memory requirement for these methods, throw out half your words and make your product even worse. With LetterWise, halving memory typically has relatively little impact on performance.

No competing predictive text-entry product can operate under such intense technical constraints on code size, database size, and processor power. And yet, the value that easy-to-use text entry adds to such products is no less important than the value brought to devices that are by comparison "power-tools", such as cell phones.

With LetterWise, predictive text entry can be made available in formerly unimaginable places.

If you manufacture a device based on a 8-bit (or even 4-bit) processor and you have a few kilobytes of memory to spare, please contact us to discuss your text-entry needs. We may be able to help.