Intel Itanium 9150M, Manual

Page: 95 / 420

3:86 Volume 3: Instruction Reference
fms
fms — Floating-point Multiply Subtract
Format: (
qp
) fms.
pc
.
sf f
1
=
f
3
,
f
4
,
f
2
F1
Description:
The product of FR
f
3
and FR
f
4
is computed to infinite precision and then FR
f
2
is
subtracted from this product, again in infinite precision. The resulting value is then
rounded to the precision indicated by
pc
(and possibly FPSR. sf . pc and FPSR. sf . wre )
using the rounding mode specified by FPSR. sf . rc . The rounded result is placed in FR
f
1
.
If any of FR
f
3
, FR
f
4
, or FR
f
2
is a NaTVal, a NaTVal is placed in FR
f
1
instead of the
computed result.
If
f
2
is f0, an IEEE multiply operation is performed instead of a multiply and subtract.
See “fmpy — Floating-point Multiply” on page 3:85.
The mnemonic values for the opcode’s
pc
are given in Table 2-22 on page 3:56 . The
mnemonic values for
sf
are given in Table 2-23 on page 3:56 . For the encodings and
interpretation of the status field’s pc , wre , and rc , refer to Table 5-5 and Table 5-6 on
page 1:90 .
Operation: if (PR[ qp ]) {
fp_check_target_register( f
1
);
if (tmp_isrcode = fp_reg_disabled( f
1
, f
2
, f
3
, f
4
))
disabled_fp_register_fault(tmp_isrcode, 0);
if (fp_is_natval(FR[ f
2
]) || fp_is_natval(FR[ f
3
]) ||
fp_is_natval(FR[ f
4
])) {
FR[ f
1
] = NATVAL;
fp_update_psr( f
1
);
} else {
tmp_default_result = fms_fnma_exception_fault_check( f
2
, f
3
, f
4
,
pc, sf, &tmp_fp_env);
if (fp_raise_fault(tmp_fp_env))
fp_exception_fault(fp_decode_fault(tmp_fp_env));
if (fp_is_nan_or_inf(tmp_default_result)) {
FR[ f
1
] = tmp_default_result;
} else {
tmp_res = fp_mul(fp_reg_read(FR[ f
3
]), fp_reg_read(FR[ f
4
]));
tmp_fr2 = fp_reg_read(FR[ f
2
]);
tmp_fr2.sign = !tmp_fr2.sign;
if (
f
2
!= 0)
tmp_res = fp_add(tmp_res, tmp_fr2, tmp_fp_env);
FR[ f
1
] = fp_ieee_round(tmp_res, &tmp_fp_env);
}
fp_update_fpsr( sf , tmp_fp_env);
fp_update_psr( f
1
);
if (fp_raise_traps(tmp_fp_env))
fp_exception_trap(fp_decode_trap(tmp_fp_env));
}
}
FP Exceptions:
Invalid Operation (V) Underflow (U)
Denormal/Unnormal Operand (D) Overflow (O)
Software Assist (SWA) fault Inexact (I)
Software Assist (SWA) trap