
Data Sheet
July 2000
DSP16210 Digital Signal Processor
Lucent Technologies Inc.
DRAFT COPY
71
Software Architecture
(continued)
Instruction Set Quick Reference
(continued)
Table 32. Instruction Set Summary
Instruction
Flags
szlme
Cycles
Out
Words
In
Multiply/Accumulate (MAC) Group
F1
F1
F1
F1
F1
F1
F1
yh=aTh
F1
if CON
F1E
F1E
y
h,l
=
aTE
h,l
F1E
aTE
h,l
=
y
h,l
F1E
y=aE_Ph
F1E
aE_Ph=y
F1E
F1E
F1E
F1E
F1E
F1E
F1E
F1E
F1E
y
h
=
*r0
F1E
F1E
F1E
F1E
F1E
y=aE_Ph
F1E
y
h
=aTE
h
F1E
aTE
h
=y
h
F1E
F1E
F1E
F1E
F1E
F1E
F1E
X
C
is one cycle if XAAU contention occurs and zero cycles otherwise. XAAU contention occurs frequently for these instruc-
tion types and can only be avoided by use of the cache.
For this transfer, the postincrement options
*rME
and
*rME––
are not available
for double-word loads.
§ The – (40-bit subtraction) operation is encoded as
aDE=aSE+IM16
with the IM16 value negated.
For conditional branch instructions, the execution time is two cycles if the branch is not taken.
The instruction performs the same function whether or not
near
(optional) is included.
§§ Not including the N instructions.
Y
x
h,l
=Y
y
h,l
=Y
a
h,l
=Y
Y=y
h,l
Y=aT
h,l
szlm–
szlm–
szlm–
szlm–
szlm–
szlm–
szlm–
szlm–
szlme
szlme
szlme
szlme
szlme
szlme
szlme
szlme
szlme
szlme
szlme
szlme
szlme
szlme
szlme
szlme
szlme
szlme
szlme
szlme
szlme
szlme
szlme
szlme
szlme
szlme
szlme
szlme
1
1
1
xh=X
xh=X
1+X
C
yh=Y
1
1
2
x
h,l
=YE
y
h,l
=YE
aTE
h,l
=YE
aE_Ph=YE
YE=x
h,l
YE=y
h,l
YE=aTE
h,l
YE=aE_Ph
r0=rNE+jhb
YE
x
h,l
=
XE
aTE
h,l
=XE
aE_Ph=XE
x
h
=
XE
x
h
=
XE
x
h
=
XE
a4
h
=XE
x
h
=XE
x
h
=
XE
a4_5h=XE
x
h
=XE
x
h
=
XE
a4
h
=
XE
1+X
C
y
h
=
YE
y
h,l
=YE
YE=y
h,l
y
h
=YE
YE=a6_7h
YE=a6
h
YE=a6
h
D