The MPC7410 is the second implementation of the fourth

generation (G4) microprocessors from Freescale. The MPC7410

implements the full PowerPC 32-bit architecture and is targeted at

both computing and embedded systems applications.

Some comments on the MPC7410 with respect to the MPC750:

? The MPC7410 adds an implementation of the new

AltiVec? technology instruction set.

? The MPC7410 includes significant improvements in

memory subsystem (MSS) bandwidth and offers an

optional, high-bandwidth MPX bus interface.

? The MPC7410 adds full hardware-based multiprocessing

capability, including a five-state cache coherency protocol

(four MESI states plus a fifth state for shared

intervention).

? The MPC7410 is implemented in a next generation process technology for core frequency improvement.

? The MPC7410 floating-point unit has been improved to make latency equal for double- and single-precision

operations involving multiplication.

? The completion queue has been extended to eight slots.

? There are no other significant changes to scalar pipelines, decode/dispatch/completion mechanisms, or the

branch unit. The MPC750 four-stage pipeline model is unchanged (fetch, decode/dispatch, execute,

complete/writeback).

Some comments on the MPC7410 with respect to the MPC7400:

? The MPC7410 adds configurable direct-mapped SRAM capability to the L2 cache interface.

? The MPC7410 adds 32-bit interface support to the L2 cache interface. The MPC7410 implements a 19th L2

address pin (L2ASPARE on the MPC7400) in order to support additional address range.

? The MPC7410 removes support for 3.3-V I/O on the L2 cache interface.

Features

This section summarizes features of the MPC7410 implementation of the PowerPC architecture. Major features of

the MPC7410 are as follows:

? Branch processing unit

— Four instructions fetched per clock

— One branch processed per cycle (plus resolving two speculations)

— Up to one speculative stream in execution, one additional speculative stream in fetch

— 512-entry branch history table (BHT) for dynamic prediction

— 64-entry, four-way set-associative branch target instruction cache (BTIC) for eliminating branch delay

slots

? Dispatch unit

— Full hardware detection of dependencies (resolved in the execution units)

— Dispatch two instructions to eight independent units (system, branch, load/store, fixed-point unit 1,

fixed-point unit 2, floating-point, AltiVec permute, AltiVec ALU)

— Serialization control (predispatch, postdispatch, execution serialization)