STR910FAZ32H6T STMICROELECTRONICS [STMicroelectronics], STR910FAZ32H6T Datasheet - Page 12

• Manufacturer Part Number: STR910FAZ32H6T
• Manufacturer: STMICROELECTRONICS [STMicroelectronics]

Functional overview
3
3.1
3.2
3.3
3.4
3.4.1
12/99
Functional overview
System-in-a-Package (SiP)
The STR91xFA is a SiP device, comprised of two stacked die. One die is the ARM966E-S
CPU with peripheral interfaces and analog functions, and the other die is the burst Flash.
The two die are connected to each other by a custom high-speed 32-bit burst memory
interface and a serial JTAG test/programming interface.
Package choice
STR91xFA devices are available in 128-pin (14 x 14 mm) 80-pin (12 x 12 mm) LQFP and
LFBGA144 (10 x 10 mm) packages. Refer to
available peripherals for each of the package choices.
ARM966E-S CPU core
The ARM966E-S core inherently has separate instruction and data memory interfaces
(Harvard architecture), allowing the CPU to simultaneously fetch an instruction, and read or
write a data item through two Tightly-Coupled Memory (TCM) interfaces as shown in
Figure
reduction in cycle count per instruction. In addition to this, a 5-stage pipeline is used to
increase the amount of operational parallelism, giving the most performance out of each
clock cycle.
Ten DSP-enhanced instruction extensions are supported by this core, including single-cycle
execution of 32x16 Multiply-Accumulate, saturating addition/subtraction, and count leading-
zeros.
The ARM966E-S core is binary compatible with 32-bit ARM7 code and 16-bit Thumb
Burst Flash memory interface
A Burst Flash memory interface
(I-TCM) path of the ARM966E-S core. Also in this path is an 8-instruction Pre-Fetch Queue
(PFQ) and a 15-entry Branch Cache (BC), enabling the ARM966E-S core to perform up to
96 MIPS while executing code directly from Flash memory. This architecture provides high
performance levels without a costly instruction SRAM, instruction cache, or external
SDRAM. Eliminating the instruction cache also means interrupt latency is reduced and code
execution becomes more deterministic.
Pre-Fetch Queue (PFQ)
As the CPU core accesses sequential instructions through the I-TCM, the PFQ always looks
ahead and will pre-fetch instructions, taking advantage any idle bus cycles due to variable
length instructions. The PFQ will fetch 32-bits at a time from the Burst Flash memory at a
rate of up to 96 MHz.
1. The result is streamlined CPU Load and Store operations and a significant
(Figure
1) has been integrated into the Instruction TCM
STR91xFAx32 STR91xFAx42 STR91xFAx44
Table 2
and
Table 3 on page 11
for a list of
®
code.