If you're working on a network, WordSmith will be s-l-o-w if it has to read and write results across the network. It's much faster to do your work locally on a C:\ or D:\ drive and then copy any useful results over to network storage later if required.


and generally


To make a word-list on 4.2 million words used to take about 20 minutes on a 1993 vintage 486-33 with 8Mb of RAM. The sorting procedure at the end of the processing took about 30 seconds. A 200Mz Pentium with 64MB of RAM handled over 1.7 million words per minute. On a 100Mz Pentium with 32Mb of RAM this whole process took about 3 and a half minutes, working at over a million words a minute.

When concordancing, tests on the same Pentium 100, using one 55MB text file of 9.3 million words, and a quad-speed CD-ROM drive, showed

search-word        source        speed        

quickly        CD-ROM        6 million words per minute        

quickly        hard disk        12 million wpm        

the        CD-ROM        900,000 wpm        

the        hard disk        1 million wpm        

thez        CD-ROM        6 million wpm        

thez        hard disk        16 million wpm        


Tests using a set of text files ranging from 20K down to 4K, using quickly as the search-word, gave speeds of 2 million wpm rising with the longer files to 4 million wpm. Making a word list on the same set of files gave an average speed of 800,000 wpm. On the 55MB text file the speed was around 1.35 million wpm.

These data suggest that factors which slow concordancing down are, in order, word rarity (the was much slower than quickly or the non-existent thez), text file size (very small files of only 500 words or so (3K) will be processed about three times as slowly as big ones) and disk speed (the outdated quad speed CD-ROM being roughly half the speed of the 12ms hard disk). When Concord finds a word it has to store the concordance line and collocates and show it (so that you can decide to suspend any further processing if you don't like the results or have enough already). This is a major factor slowing down the processing. Second, reading a file calls on the computer's file management system, which is quite slow in loading it, in comparison with Concord actually searching through it. Third, disk speeds are quite varied, floppy disks being much the worst for speed.

If processing seems excessively slow, close down as many programs as possible and run WordSmith Tools again. Or install more RAM. Get advice about setting Windows to run efficiently (virtual memory, disk caches, etc.) Use a large fast hard drive.

You can run other software while the programs are computing, but they will take up a lot of the processor's time. Shoot-em-up games may run too jerkily, but printing a document at the same time should be fine.

Click the Permalink button if you want to copy a link to this page.