x86 unsupported formats
Chris Hinds H3-292
uunet!wtkatz.sps.mot.com!chinds
Mon May 16 06:35:54 PDT 1994
> The extended format on x86 permits bit patterns that do not fall in
> the "standard" categories (normal, denormal, pseudodenormal, infinity,
> NaN). These are patterns with non-zero biased exponent and zero
> leading bit of the significand. According to the Intel486 Programmers
> Reference Manual they are not supported. The Intel486 processor raises
> the invalid-operation exception when they are encountered as operands.
> I am seeking opinions on the following statement: in the spirit of
> IEEE 754, software should treat these bit patterns as if they were
> NaN.
As I read in 754, P 3.3,
"This standard allows an implementation to encode some values redundantly,
provided that redundancy be transparent to the user in the following
sense: an implementation either shall encode every non-zero value uniquely
or it shall not distinguish redundant encodings of nonzero values."
In the 68000 family (68881, 68882, 68040, and 68060) these encodings are
considered "unnormal" but legitimate. The part converts these values
to normalized or denormalized numbers or zero and continues the operation
transparent to the user.
This was done in hardware on the 881 and 882, but takes an unsupported
data type exception on the 040 and 060 and is handled by the software
envelope without reporting a user visible exception unless explicitly
asked enabled. Neither part will produce an unnormal result from an
arithmetic operation; however, a movein of a single (or double) denormal
with round-to-single (or double) will produce a unnormalized extended
precision value in the register.
chris
****************************************************
* Chris N. Hinds <>< *
* chindsaoakhill.sps.mot.com *
* (512) 891-2753 FAX: (512) 891-8315 *
* *
* Motorola Inc. *
* Microprocessor and Memory Technologies Group *
* Mail Drop OE38 *
* 6501 William Cannon Dr. West *
* Austin, Tx 78735-8598 *
****************************************************
More information about the Numeric-interest
mailing list