Did this thread diffuse from ASSEMBLER-LIST to IBM-MAIN because the diffuser deemed it of more general interest? If so, more exposition should have been quoted for the benefit of the expanded readership.
In a recent note, john gilmore said: > Date: Tue, 28 Nov 2006 13:12:00 +0000 > > This second scheme has excellent, flat performance which is independent of n > and any oirdering of tests for n < 2^8 + 1 = 257 (or for n < 2^16 + 1 = > 65537 if a TROO instructrion and a larger table are used). > > There is thus little excuse for the use of thickets of nested if-then-elses. > Here, as elsewhere, an explicit, reusable table-driven scheme is better > than an ad hoc procedural one. > The table-driven scheme scales poorly to extremely sparse label sets. A compiler could consider both and choose the more efficient one based on the tradeoff between time and space. Many tests can be performed for the cost of fetching a one-page table. -- gil -- StorageTek INFORMATION made POWERFUL ---------------------------------------------------------------------- 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
