[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

decimal arithmetic (was: Clause 10 ...)



Two MINOR points:

... decimal is more accurate if no rounding is required.
But of course the I/O time dominates the arithmetic time.

Exceptions:  large numbers of micro-payments, where "I/O" may
be (after buffering and caching) to in-memory database files.
Example:  that (in)famous telco benchmark.

[effective precision] ... That hump is what made hexadecimal floating
point so problematic, especially in single precision - it was much like
losing a couple of bits of precision

The (bad) effects of truncating arithmetic overwhelmed the effects of
variable effective precision (21-24 bits vs 24; 53-56 bits vs 53).

Michel.
Sent: 2007-08-02 14:41:01 UTC

754 | revision | FAQ | references | list archive