Tom Marchant writes:

I seriously doubt that TRTO is faster than 50 compares on a modern
superscalar processor.  I hope you mean faster by a factor of 10x,
not by a factor of 10k.  The latter is absurd, the former unlikely.

in response to my

. . . Maintaining/replacing ad hoc code of this sort is a nightmarish job.
Replacing all of it with a single reentrant assembly-language subroutine
(and, usually, several different 64-kibibyte tables because different
'lines' may do business in different states) makes this job tractable; and
it speeds up the execution of these applications enormously, often by a
factor of more than 10k (sic).

Now 'sic' literally means 'thus' in Latin, and its standard use in English and other European languages is to mark text that might otherwise be thought to be corrupt as correct and intended.

I did not of course assert that one TRTO is faster than 50 compares, although this is in fact the case; what I said was that certain American insurance applications that made repeated, heavy use of state-code driven if-then-else-if constructs were 10k faster after the replacement of these sets of compares by (a single) TRTO-based numeric evaluation of one of the two-character state-code values 'AK', . . . 'WY'. These programs were also much smaller.

Mr. Marchant has a controversial style that is at once naif, uninformed and rudely combative. It dismembers straw men from an assumed posture of unique virtue. Disagreement, even strong, vigorously expressed disagreement, are staples here. His modus operandi is not, need not and should not be.

Moreover, he is comnpulsively loquacious about too many topics. Justice Holmes said that five generations of cretins were enough; and five all but vacuous posts in the 'moving LOADxx' thread already this morning are too many.

John Gilmore
Ashland, MA 01721-1817
USA

_________________________________________________________________
MSN Shopping has everything on your holiday list. Get expert picks by style, age, and price. Try it! http://shopping.msn.com/content/shp/?ctId=8000,ptnrid=176,ptnrdata=200601&tcode=wlmtagline

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html

Reply via email to