9.0 Instruction Set
This section summarizes the Geode GXm processor
instruction set and provides detailed information on the
instruction encodings. The instruction set is broken into
four categories:
4. There are no local bus HOLD requests delaying
processor access to the bus.
5. No exceptions are detected during instruction execu-
tion.
•
Processor Core Instruction Set - listed in Table 9-27 on
page 213
6. If an effective address is calculated, it does not use
two general register components. One register,
scaling and displacement can be used within the
clock count shown. However, if the effective address
calculation uses two general register components,
add one clock to the clock count shown.
•
•
•
FPU Instruction Set - listed in Table 9-29 on page 225
MMX Instruction Set - listed in Table 9-31 on page 230
National Semiconductor Extended MMX Instruction Set
- listed in Table 9-33 on page 235
7. All clock counts assume aligned 32-bit memory/IO
operands.
These tables provide information on the instruction encod-
ing, and the instruction clock counts for each instruction.
The clock count values for these tables are based on the
following assumptions
8. If instructions access a 32-bit operand on odd
addresses, add one clock for read or write and add
two clocks for read and write.
1. All clock counts refer to the microprocessor core
internal clock frequency. For example, clock doubled
9. For non-cached memory accesses, add two clocks
(clock doubled GXm processor cores) or four clocks
(clock tripled GXm processor cores), assuming zero
wait state memory accesses.
GXm processor cores will reference
frequency that is twice the bus frequency.
a
clock
2. The instruction has been prefetched, decoded and is
ready for execution.
10. Locked cycles are not cacheable. Therefore, using the
LOCK prefix with an instruction adds additional clocks
as specified in item 9 above.
3. Bus cycles do not require wait states.
Revision 3.1
201
www.national.com