Intel ITANIUM ARCHITECTURE - SOFTWARE DEVELOPERS MANUAL VOLUME 3 REV 2.3 Manual page 116

fpma — Floating-point Parallel Multiply Add
( ) fpma.
Format: qp sf f
The pair of products of the pairs of single precision values in the significand fields of FR
Description:
and FR
f
3
values in the significand field of FR
precision. The resulting values are then rounded to single precision using the rounding
mode specified by FPSR.sf.rc. The pair of rounded results are stored in the significand
field of FR
(0x1003E) and the sign field of FR
If any of FR
results.
Note: If
(See "fpmpy — Floating-point Parallel Multiply" on page
operand, is not a packed pair of 1.0 values, it is just the register file format's
1.0 value.
The mnemonic values for sf are given in
The encodings and interpretation for the status field's rc are given in
page 1:90.
Volume 3: Instruction Reference
= , ,
f f f
1 3 4 2
are computed to infinite precision and then the pair of single precision
f
4
. The exponent field of FR
f
1
, FR , or FR is a NaTVal, FR
f f f
3 4 2
is f0 in the fpma instruction, just the IEEE multiply operation is performed.
f
2
is added to these products, again in infinite
f
2
is set to the biased exponent for 2.0
f
1
is set to positive (0).
f
1
is set to NaTVal instead of the computed
f
1
Table 2-23 on page
fpma
63
3:115.) FR f1, as an
3:56.
Table 5-6 on
3:107
F1