MOTOROLA MPC2605ZP66R

MOTOROLA
Order this document
by MPC2605/D
SEMICONDUCTOR TECHNICAL DATA
MPC2605
Product Preview
Integrated Secondary Cache
for PowerPC Microprocessors
The MPC2605 is a single chip, 256KB integrated look–aside cache with
copy–back capability designed for PowerPC applications (MPC603 and
MPC604). Using 0.38 µm technology along with standard cell logic technology,
the MPC2605 integrates data, tag, host interface, and least recently used (LRU)
memory with a cache controller to provide a 256KB, 512KB, or 1 MB Level 2
cache with one, two, or four chips on a 64–bit PowerPC bus.
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
ZP PACKAGE
PBGA
CASE 1138–01
Single Chip L2 Cache for PowerPC
66 MHz Zero Wait State Performance (2–1–1–1 Burst)
Four–Way Set Associative Cache Design
32K x 72 Data Memory Array
8K x 18 Tag Array
Address Parity Support
LRU Cache Control Logic
Copy–Back or Write–Through Modes of Operation
Copy–Back Buffer for Improved Performance
Single 3.3 V Power Supply
5 V Tolerant I/O
One, Two, or Four Chip Cache Solution (256KB, 512KB, or 1MB)
Single Clock Operation
Compliant with IEEE Standard 1149.1 Test Access Port (JTAG)
Supports up to Four Processors in a Shared Cache Configuration
High Board Density 25 mm 241 PBGA Package
BLOCK DIAGRAM
COPY–BACK
BUFFER
CONTROL
RD/WR
60X BUS
INTERFACE
A27, A28
DH0 – DH31
DL0 – DL31
DP0 – DP7
8K x 72 x 4
DATA RAM
CONTROLLER
AND
BUS INTERFACE
A0 – A31
WAY SELECT
RD/WR
2K x 18 x 4
TAG RAM
COMPARE
2K x 8 LRU
The PowerPC name is a trademark of IBM Corp., used under license therefrom.
This document contains information on a product under development. Motorola reserves the right to change or discontinue this product without notice.
REV 6
2/26/98

Motorola, Inc. 1998
MOTOROLA
MPC2605
1
PIN ASSIGNMENT
1
A
B
C
D
2
3
4
ABB
L2 BG
DH20
5
CFG4
FDN
CPU3
BR
CPU3 VDD
DBG
L2 BR
CPU2
BR
CPU2
DBG
7
8
DH19 DH17 DH31 DH29
DH23 DH21
L2
MISS INH
CPU3
BG
6
DP2
DH18 DH16 DH30
DH22
9
10
11
12
13
14
15
16
17
18
DL29
DL30
DH27
DH26 DL16
DL19
DL22
DP6
DL25
DL27
DH28
DH25 DL17
DL20
DL23
DL24
DL26
DL28
DL31
DP7
APE
DL18
DL21
VSS
VSS
VDD
VDD
AP3
AP2
AP1
VDD
VSS
VSS
AP0
L2
FLUSH
L2 CI
CFG3 APEN
GBL
TSIZ1 TSIZ0
TSIZ2
VSS
DP3
VSS DH24
VSS
VSS
VDD
VDD
E
TA
F
G
H
J
K
L
19
L2 DBG CPU2
BG
CPU L2 CLAIM NC
DBG
VSS
VSS
VSS
A13
VSS VSS
VSS
A14
A15
A16
VSS
VSS VSS
VDD
A19
A18
A17
ARTRY
CI
AACK
VSS
TEA
CPU
BR
WT
VSS
VDD
VDD
VDD
HRESET DBB PWRDN VDD
VDD
VDD
TT1
TT0
TBST
VDD
VDD
VSS
VSS
VSS
VDD
VDD
A20
A22
A21
TT4
TT2
TS
VDD
VSS
VSS
VSS
VDD
VDD
VDD
A25
A24
A23
CPU BG CLK
VSS
VSS
VSS
VDD
VDD
VDD
VSS
A28
A27
A26
VSS
A31
A30
A29
M
TT3
N
P
R
T
U
V
SRESET
L2 L2 UPDATE VSS
TAG CLR INH
TDI
TCK
TMS
A10
A11
A12
TDO
TRST
NC
A7
A8
A9
CPU4
BG
CPU4
DBG
VDD
VDD
A5
A6
CPU4
BR
CFG0
VDD
VDD
VSS
CFG2
CFG1
DH7
DH5
DP0
DH6
DH4
W
VSS
VSS
VDD
VDD
VDD
VSS
VSS
VSS
VSS
DH14
DH10
DL1
DL4
VSS
VSS
VSS
VDD
DP5
VDD
A3
A4
DH3
DH1
DP1
DH13
DH11
DL0
DL3
DL6
DP4
DL10
DL12
DL14
DL15
A1
A2
DH2
DH0
DH15
DH12
DH9
DH8
DL2
DL5
DL7
DL8
DL9
DL11
DL13
A0
TOP VIEW (X–RAY VIEW)
MPC2605
2
MOTOROLA
PIN DESCRIPTIONS
Pin Locations
Pin Name
Type
Description
19G, 17H – 19H, 17J – 19J,
17K – 19K, 17L – 19L,
17M – 19M, 17N – 19N,
17P – 19P, 17R – 19R,
18T, 19T, 18U, 19U,
A0 – A31
I/O
Address inputs from processor. Can also be outputs for processor snoop
addresses. A0 is the MSB. A31 is the LSB.
3G
AACK
I/O
Address acknowledge input/output.
2A
ABB
I/O
Used as an input to qualify bus grants. Driven as an output during address tenure
initiated by the MPC2605.
AP0 – AP3
I/O
Address parity.
19B
APE
O
Address parity error. When an address parity error is detected, APE will be driven
low one clock cycle after the assertion of TS then High–Z following clock cycle.
18E
APEN
I
Address parity enable. When tied low, enables address parity bits and the
address parity error bit.
1G
ARTRY
I/O
Address retry status I/O. Generated when a read or write snoop to a dirty
processor cache line has occurred.
2U
2V
1V
17E
2B
CFG0
CFG1
CFG2
CFG3
CFG4
I
18V, 19V, 18W *
17C – 19C, 17D
*
Configuration inputs. These must be tied to either VDD or VSS.
CFG0
CFG1
CFG2
0
0
0
256KB
0
1
0
512KB; A26 = 0
0
1
1
512KB; A26 = 1
1
0
0
1MB; A25 – A26 = 00
1
0
1
1MB; A25 – A26 = 01
1
1
0
1MB; A25 – A26 = 10
1
1
1
1MB; A25 – A26 = 11
CFG3
Snoop Data Tenure Selector
0
Supports snoop data tenure
1
Does not support snoop data tenure
CFG4
AACK Driver Enable
0
Disable AACK driver
1
Enable AACK driver
2G
CI
I/O
Cache inhibit I/O.
3M
CLK
I
Clock input. This must be the same as the processor clock input.
2M
CPU BG
I
CPU bus grant input.
3E
CPU2 BG
I
MPC2605 logically ORs this signal with CPU BG. Used in multiprocessor
configuration as the second CPU BG.
1B
CPU3 BG
I
MPC2605 logically ORs this signal with CPU BG. Used in multiprocessor
configuration as the third CPU BG.
1T
CPU4 BG
I
MPC2605 logically ORs this signal with CPU BG. Used in multiprocessor
configuration as the fourth CPU BG.
2H
CPU BR
I
CPU bus request input.
2D
CPU2 BR
I
MPC2605 logically ORs this signal with CPU BR. Used in multiprocessor
configuration as the second CPU BR.
2C
CPU3 BR
I
MPC2605 logically ORs this signal with CPU BR. Used in multiprocessor
configuration as the third CPU BR.
1U
CPU4 BR
I
MPC2605 logically ORs this signal with CPU BR. Used in multiprocessor
configuration as the fourth CPU BR.
1F
CPU DBG
I
CPU data bus grant input from arbiter.
3D
CPU2 DBG
I
MPC2605 logically ORs this signal with CPU DBG. Used in multiprocessor
configuration as the second CPU DBG.
3C
CPU3 DBG
I
MPC2605 logically ORs this signal with CPU DBG. Used in multiprocessor
configuration as the third CPU DBG.
* See pin diagram (page 2) for specific pin assignment of these bus signals.
MOTOROLA
MPC2605
3
Pin Locations
Pin Name
Type
Description
2T
CPU4 DBG
I
MPC2605 logically ORs this signal with CPU DBG. Used in multiprocessor
configuration as the fourth CPU DBG.
11A – 13A, 15A – 18A,
11B – 17B, 11C, 12C, 10U,
11U, 10V – 12V, 14V – 17V,
DL0 – DL31
I/O
Data bus low input and output. DL0 is the MSB. DL31 is the LSB.
DH0 – DH31
I/O
Data bus high input and output. DH0 is the MSB. DH31 is the LSB.
2J
DBB
I/O
Data bus busy. Used as input when processor is master, driven as an output after
a qualified L2 DBG when MPC2605 is the bus master. Note: To operate in Fast
L2 mode, this pin must be tied high.
14A, 18B, 5C, 8C,
DP0 – DP7
I/O
Data bus parity input and output.
1C
FDN
I/O
Flush done I/O used for communication between other MPC2605 devices. Must
be tied together between all MPC2605 parts along with a pullup resistor.
19E
GBL
O
Global transaction. Always negated when MPC2604 is bus master.
1J
HRESET
I
Hard reset input from processor bus. This is an asynchronous input that must be
low for at least 16 clock cycles to ensure the MPC2605 is properly reset. For
proper initialization, TRST must be asserted before HRESET is asserted.
3A
L2 BG
I
Bus grant input from arbiter.
11W – 17W *
4A – 10A, 4B – 10B, 6C,
10C, 8U, 9U, 3V – 6V,
8V, 9V, 3W –10W *
16U, 7V, 13V, 2W *
1D
L2 BR
I/O
19D
L2 CI
I
Bus request I/O. Normally used as an output.
Secondary cache inhibit sampled, after assertion of TS. Assertion prevents
linefill.
2F
L2 CLAIM
O
L2 cache claim output. Used to claim the bus for processor initiated memory
operations that hit the L2 cache. L2 CLAIM goes true (low) before the rising edge
of CLK following TS true. Because this output is not always driven, a pullup
resistor may be necessary to ensure proper system functioning.
2E
L2 DBG
I
Data bus grant input. Comes from system arbiter, used to start data tenure for
bus operations where MPC2605 is the bus master.
18D
L2 FLUSH
I
Causes cache to write back dirty lines and clears all tag valid bits.
3B
L2 MISS INH
I
Prevents line fills on misses when asserted.
2N
L2 TAG CLR
I
Invalidates all tags and holds cache in a reset condition.
3N
L2 UPDATE
INH
I
Cache disable. When asserted, the MPC2605 will not respond to signals on the
local bus and internal states do not change.
3J
PWRDN
I
Provides low power mode. Prevents address and data transitions into the RAM
array. MPC2605 becomes active 4 µs after deassertion. Clock must be externally
disabled.
1N
SRESET
I
Soft reset input from processor bus.
1E
TA
I/O
Transfer acknowledge status I/O from processor bus.
3K
TBST
I/O
Transfer burst status I/O from processor bus. Used to distinguish between
burstable and non–burstable memory operations.
2P
TCK
I
Test clock input for IEEE 1149.1 boundary scan (JTAG).
1P
TDI
I
Test data input for IEEE 1149.1 boundary scan (JTAG).
1R
TDO
O
Test data output for IEEE 1149.1 boundary scan (JTAG).
1H
TEA
I
Transfer error acknowledge status input from processor bus.
3P
TMS
I
Test mode select for IEEE 1149.1 boundary scan (JTAG).
* See pin diagram (page 2) for specific pin assignment of these bus signals.
MPC2605
4
MOTOROLA
Pin Locations
Pin Name
Type
Description
2R
TRST
I
Test reset input for IEEE 1149.1 boundary scan (JTAG). If JTAG will not be used,
TRST should be tied low.
3L
TS
I/O
Transfer start I/O from processor bus (can also come from any bus master on the
processor bus). Signals the start of either a processor or bus master cycle.
17F – 19F *
TSIZ0 – TSIZ2
I/O
Transfer size I/O from processor bus.
1K, 2K, 1L, 2L, 1M *
TT0 – TT4
I/O
Transfer type I/O from processor bus.
3H
WT
I/O
Write through status input from processor bus. When tied to ground, the
MPC2605 will operate in write–through mode only (no copy–back).
4C, 15C, 16C, 9D – 11D,
8H – 10H, 4J, 8J, 9J, 16J, 4K,
8K, 12K, 16K, 4L, 11L, 12L,
16L, 10M – 12M, 3T, 9T – 11T,
17T, 3U, 4U, 15U, 17U
VDD
Supply
Power supply: 3.3 V ± 5%.
7C, 9C, 13C, 14C, 7D, 8D,
12D, 13D, 4G, 16G – 18G,
4H, 11H, 12H, 16H, 10J – 12J,
9K – 11K, 8L – 10L, 4M,
8M, 9M, 16M, 4N, 16N, 7T, 8T,
12T, 13T, 5U – 7U, 12U – 14U
VSS
Supply
Ground.
3F, 3R
NC
—
No connection: There is no connection to the chip.
* See pin diagram (page 2) for specific pin assignment of these bus signals.
ABSOLUTE MAXIMUM RATINGS (See Note 1)
Rating
Power Supply Voltage
Symbol
Value
Unit
VDD
– 0.5 to + 4.6
V
Voltage Relative to VSS
Vin, Vout
– 0.5 to VDD + 0.5
V
Output Current (per I/O)
Iout
± 20
mA
Power Dissipation (Note 2)
PD
—
W
Temperature Under Bias
Tbias
– 10 to + 85
°C
Operating Temperature
TJ
0 to + 125
°C
Tstg
– 55 to + 125
°C
Storage Temperature
This device contains circuitry to protect the
inputs against damage due to high static voltages or electric fields; however, it is advised
that normal precautions be taken to avoid
application of any voltage higher than maximum rated voltages to this high–impedance
circuit.
This BiCMOS memory circuit has been designed to meet the dc and ac specifications
shown in the tables, after thermal equilibrium
has been established.
NOTES:
1. Permanent device damage may occur if ABSOLUTE MAXIMUM RATINGS are
exceeded. Functional operation should be restricted to RECOMMENDED OPERATING CONDITIONS. Exposure to higher than recommended voltages for extended
periods of time could affect device reliability.
2. Power dissipation capability is dependent upon package characteristics and use
environment. See Package Thermal Characteristics.
MOTOROLA
MPC2605
5
DC OPERATING CONDITIONS AND CHARACTERISTICS
(TJ = 20 to + 110°C, Unless Otherwise Noted)
RECOMMENDED OPERATING CONDITIONS (Voltages Referenced to VSS = 0 V)
Parameter
Symbol
Min
Typ
Max
Unit
Supply Voltage (Operating Voltage Range)
VDD
3.135
3.3
3.465
V
Input High Voltage
VIH
2.0
—
5.5
V
Input Low Voltage
VIL
– 0.5*
—
0.8
V
Symbol
Min
Max
Unit
Input Leakage Current (All Inputs, Vin = 0 to VDD)
Ilkg(I)
—
± 1.0
µA
Output Leakage Current (High–Z State, Vout = 0 to VDD)
Ilkg(O)
—
± 1.0
µA
ICCA
—
720
mA
IQ
—
195
mA
Output Low Voltage (IOL = + 8.0 mA)
VOL
—
0.4
V
Output High Voltage (IOH = – 4.0 mA)
VOH
2.4
—
V
Symbol
Typ
Max
Unit
Cin
4
6
pF
Output Capacitance
Cout
6
8
pF
Input/Output Capacitance
CI/O
8
10
pF
Symbol
Max
Unit
Thermal Resistance Junction to Ambient (Still Air, Test Board with Two Internal Planes)
RθJA
26.5
°C/W
Thermal Resistance Junction to Ambient (200 lfpm, Test Board with Two Internal Planes)
RθJA
23.2
°C/W
Thermal Resistance Junction to Board (Bottom)
RθJB
15.9
°C/W
Thermal Resistance Junction to Case (Top)
RθJC
6.6
°C/W
* VIL(min) = – 2.0 V ac (pulse width ≤ 20 ns).
DC CHARACTERISTICS
Parameter
AC Supply Current (Iout = 0 mA, All inputs = VIL or VIH, VIL = 0 V, and VIH ≥ 3.0 V,
Cycle Time = 15 ns, max value assumes a constant burst read hit, with 100% bus utilization,
and 100% hit rate)
AC Quiescent Current (Iout = 0 mA, All inputs = VIL or VIH, VIL = 0 V and VIH ≥ 3.0 V,
Cycle Time = 15 ns, All Other Inputs DC)
CAPACITANCE (f = 1.0 MHz, dV = 3.0 V, TA = 25°C, Periodically Sampled Rather Than 100% Tested)
Parameter
Input Capacitance
PACKAGE THERMAL CHARACTERISTICS
Rating
MPC2605
6
MOTOROLA
AC OPERATING CONDITIONS AND CHARACTERISTICS
(TJ = 20 to + 110°C, Unless Otherwise Noted)
Input Timing Measurement Reference Level . . . . . . . . . . . . . . . 1.5 V
Input Pulse Levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0 to 3.0 V
Input Rise/Fall Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 ns
Output Measurement Timing Level . . . . . . . . . . . . . . . . . . . . . . . 1.5 V
Output Load . . . . . . . . . . . . . . . . . . . . . . . . 50 Ω Termination to 1.5 V
AC CLOCK SPECIFICATIONS
Timing
Reference
P
Parameter
Frequency of Operation
Clock Cycle Time
, Clock Rise and Fall Time
MPC2605–66
Min
Max
U i
Unit
—
66.67
MHz
15
—
ns
1.0
2.0
ns
Clock Duty Cycle Measured at 1.5 V
40
60
%
Clock Short–Term Jitter (Cycle to Cycle)
—
± 150
ps
N
Notes
1, 2
1
NOTES:
1. This parameter is sampled and not 100% tested.
2. Rise and fall times for the clock input are measured from 0.4 to 2.4 V.
CLOCK INPUT TIMING DIAGRAM
VM
VIH
VIL
VM = Midpoint Voltage (1.5 V)
MOTOROLA
MPC2605
7
AC SPECIFICATIONS
Timing
Reference
P
Parameter
Clock Cycle Time
Input Setup Time
Clock to Input Invalid (Input Hold)
Clock to Output Driven
Clock to Output Valid
Clock to Output Invalid
Clock to Output High–Z
PWRDN Disable to Recovery
MPC2605–66
Min
Max
U i
Unit
N
Notes
15
—
ns
4.5
—
ns
1
2
—
ns
1
2
9
ns
2
2
9
ns
2
—
ns
2
2
12
ns
2
—
4
µs
2
NOTES:
1. All input specifications are measured from the TTL level (0.8 or 2.0 V) of the signal in question to the 1.4 V level of the rising edge of the input
clock. Both input and output timings are measured at the pin.
2. This parameter is sampled and not 100% tested.
CLK
INPUTS
OUTPUTS
MPC2605
8
MOTOROLA
MPC2605 RESPONSE TO 60X TRANSFER ATTRIBUTES
TT0 – TT4
TBST
CI
WT
Tag Status
MPC2605 Response
X1X10
0
1
X
Miss
X1X10
0
1
X
Hit
X1010
1
0
X
Hit Clean
Paradox — Invalidate the line (processor n–cacheable read hit
clean line)
X1010
1
0
X
Hit Dirty
Paradox — ARTRY, L2 BR, then write back data, invalidate the line
(processor n–cacheable read hit dirty line)
00110
0
1
X
Miss
Line–fill except right after a snoop hit to processor (processor write
miss)
00110
0
1
1
Hit
00X10
X
1
0
Hit Clean
Cache update (processor write through WT hit clean)
00110
0
1
0
Hit Dirty
Cache update, clear dirty bit
00010
1
1
0
Hit Dirty
Paradox — ARTRY, L2 BR, write back data, keep valid, clear dirty
bit
X0010
1
0
X
Hit Clean
Paradox — Invalidate the line (processor n–cacheable write hit
clean line)
X0010
1
0
X
Hit Dirty
Paradox — ARTRY, L2 BR, then write back data, invalidate the line
(processor n–cacheable SB write hit dirty line)
00100
X
X
X
Hit Clean
Invalidate tag (flush block address–only)
00100
X
X
X
Hit Dirty
ARTRY, L2 BR, write back data, invalidate tag (flush block
address–only)
00000
X
X
X
Hit Clean
No action (clean block address–only)
00000
X
X
X
Hit Dirty
ARTRY, L2 BR, write back data, reset dirty bit (clean block
address–only)
01100
X
X
X
Hit
Line–fill (processor read miss)
L2 CLAIM, AACK, TA (processor read hit)
L2 CLAIM, AACK, TA except after a snoop hit to processor
(processor write hit)
Notes
1, 2, 3
4
1, 3, 5, 6
5, 6
Invalidate tag (kill block address–only)
NOTES:
1. If a line fill is going to replace a dirty line and the cast out buffer (COB) is full, the line fill will be cancelled. (Unless the line fill is a write which
hits in the COB. In this case, the line fill will occur.)
2. If a burst read misses the cache but hits the COB, the MPC2605 will supply the data from the COB, but not perform a line fill.
3. If ARTRY is asserted during a line fill to replace a dirty line, the line fill will be cancelled, the to–be–replaced line will recover its old tag (valid,
dirty, tag field), and the COB goes back to an invalid condition, even if the line fill is a burst write to the line in the COB.
4. If ARTRY is asserted during a read hit, the MPC2605 will abort the process.
5. If a processor burst write occurs right after a snoop write that was a cache hit, the MPC2605 will invalidate the line. If the snoop was a cache
miss, the MPC2605 will not perform a write allocate.
6. If a processor burst write occurs right after a snoop read that was a cache hit, the MPC2605 will update the cache and clear the dirty bit.
If the snoop was a cache miss, the MPC2605 will perform a write allocate.
MPC2605 RESPONSE TO CHIPSET TRANSFER ATTRIBUTES
TT0 – TT4
Tag Status
00100
X0010
X1110
Hit Clean
Invalidate line
MPC2605 Response
00100
X0010
X1110
Hit Dirty
ARTRY and L2 BR write back data, invalidate line (see Note)
00000
X1010
Hit Clean
No action
00000
X1010
Hit Dirty
ARTRY and L2 BR, write back data, reset dirty bit (see Note)
0110X
00110
Hit
Invalidate (kill block)
NOTE: In all snoop push cases, BR is sampled the cycle after the ARTRY window. If BR is asserted in this cycle, L2 BR will be immediately negated
and an assertion of L2 BG will be ignored.
TRANSFER ATTRIBUTES GENERATED FOR L2 COPYBACK
TT0 – TT4
TBST
CI
WT
00010
0
1
1
MOTOROLA
MPC2605
9
FUNCTIONAL OPERATION
SYSTEM USAGE AND REQUIREMENTS
CFG0 – CFG2
The MPC2605 is a high–performance look–aside cache
for PowerPC systems. A look–aside cache is defined as a
cache that resides on the same bus as the processor, the
memory controller, the DMA bridge, and the arbiter. The advantage of a look–aside cache is that, when the processor
makes a memory request, the cache adds no delay to the
memory controller’s response time in the event that the request cannot be satisfied by the cache. However, there are
certain system requirements that must be met before a look–
aside cache can be used.
These three configuration pins are used to implement the
different cache sizes supported by the MPC2605.
Comprehension of L2 CLAIM
Because the memory controller sees every memory request that is issued by the processor, there must be a mechanism for the cache to inform the memory controller that it
has detected a cache hit and that it will satisfy the processor ’s request. The MPC2605 has a signal called
L2 CLAIM that is asserted whenever a cache hit is detected.
Any memory controller with which the MPC2605 is to be
used must have the ability to monitor this signal.
Pipeline Depth
The 60X bus allows pipelining of transactions such that a
new transaction can be initiated before a previous transaction has fully completed. The level of pipelining that exists on
the bus is defined by how many new data transactions have
been initiated while the original transaction is still being processed. By this definition the MPC2605 can only work in a
one level deep pipeline. In the presence of transactions for
which it has asserted L2 CLAIM, the MPC2605 can control
the level of pipelining by delaying its assertion of AACK.
However, for transactions that it cannot control, the
MPC2605 is dependent upon the memory controller to control pipeline depth. Thus, another system requirement for the
use of the MPC2605 is the use of a memory controller that
only allows one level deep of pipelining on the 60X bus.
Bus Mastering
Bus mastering is a requirement only for systems which
seek to use the MPC2605 as a copyback, as opposed to a
write–through, cache. The requirement is that the system arbiter must have the ability to allow the MPC2605 to become
a bus master. Specifically, the system arbiter must be able to
recognize assertions of L2 BR and must have the ability to
assert L2 BG and L2 DBG.
These are the only requirements above and beyond what
should already exist in a PowerPC system. All other necessary control signals are signals that are required for the processor to communicate with the memory controller, the DMA
bridge, and the arbiter.
CONFIGURATION PINS
The MPC2605 has five configuration pins: CFG0, CFG1,
CFG2, CFG3, and CFG4.
MPC2605
10
256KB: For a single chip implementation, CFG0, CFG1,
and CFG2 should all be tied low.
512KB: This two chip configuration requires both parts to
have CFG0 tied low and CFG1 tied high. CFG2 is
used as a chip select when it matches the value of
A26. Therefore, one device must have CFG2 tied
low and the other device must have CFG2 tied
high.
1MB: The four chip configuration requires all four devices to have CFG0 tied high. The CFG1, CFG2
vector becomes the chip select when it matches
the A25, A26 vector. Therefore, each of the four
parts must have a unique value of the CFG[1:2]
vector.
CFG3
Many core logic chipsets are designed such that the DMA
bridge and the memory controller are resident in the same
device. In such systems there is internal communication between these two functional units. Bus transactions generated
by the DMA bridge are solely for the purpose of keeping the
system coherent. They are not explicit requests from
memory that have data tenures associated with them. However, some chipsets are designed with the memory controller
and the DMA bridge partitioned into different devices. In systems such as these, transactions generated by the DMA
bridge are true memory requests that have data tenures
associated with them. These are called snoop data tenures.
Because these two types of systems are fundamentally different, the MPC2605 must know in which type of system it is
resident in order to respond properly to the different types of
transactions. For systems that do not have snoop data tenures, CFG3 must be tied high. For systems that do use
snoop data tenures, CFG3 must be tied low.
CFG4
When the MPC2605 asserts L2 CLAIM to signal to the
memory controller that a cache hit has been detected, it is
taking control of the address and data tenures of the transaction (see 60X Bus Operation and Memory Coherence).
This means that the MPC2605 will assert AACK to end the
address tenure, and it will assert TA as needed for the data
tenure. If the data bus is idle when a processor request is initiated, the MPC2605 will assert AACK the cycle after TS was
asserted. If the data bus is busy when the request is made,
the MPC2605 will wait until the outstanding data tenure has
completed before asserting AACK. By holding off on the
assertion of AACK, the MPC2605 enforces the policy of, at
most, two outstanding data transactions at any one time. Tying CFG4 low prevents the MPC2605 from asserting AACK
to end transactions for which it has asserted L2 CLAIM. In
systems that tie CFG4 low it is necessary for the memory
controller to assert AACK for all transactions. This allows the
DMA bridge to initiate snoop transactions (as defined later)
even when there are two outstanding data transactions. If
this type of system is implemented, the arbiter must ensure
MOTOROLA
that the processor’s bus grant is negated once there are two
outstanding data transactions. It is expected that most systems will tie CFG4 high.
RESET/INITIALIZATION
To ensure proper initialization and system functionality, the
HRESET pin of the MPC2605 should be connected to the
same signal that is used to reset the processor. The TRST
signal must be negated before HRESET is negated. When
HRESET is negated, the MPC2605 commences an internal
initialization sequence to clear all of the valid bits in the
cache. The sequence takes approximately 4000 clock
cycles. During this time the MPC2605 will not participate in
any bus transaction that occurs. All transactions are, however, monitored so that, regardless of when the initialization sequence completes, the MPC2605 is prepared to take action
on the next transaction initiated by the processor.
At some point after this 4000 cycle sequence, the
MPC2605 will detect its first cache hit. At this time the system
will experience its first assertion of L2 CLAIM. If the memory
controller must be configured via software to comprehend
assertions of L2 CLAIM, this configuration operation must
have completed by this time. For systems that cannot guarantee that this requirement is met, it is necessary to disable
the MPC2605 until such time as this configuration can be
guaranteed. Disabling the MPC2605 can be accomplished
by asserting L2 UPDATE INH sometime during reset and negating it when it is deemed safe for caching to commence.
60X BUS OPERATION
All transactions have what is called an address tenure. An
address tenure is a set number of bus cycles during which
the address bus and its associated control signals are being
used for the transaction at hand. In general, there are two
types of transactions. Those that only have address tenures,
called address–only transactions. And those that require the
use of the data bus and therefore will have a data tenure.
These transactions are called data transactions. This section
describes how address and data tenures are defined as
viewed by the MPC2605.
Address Tenures
Address tenures on the 60X bus are fairly well defined.
They start with an assertion of TS by a device that has been
granted the bus by the system arbiter. This device is called
the bus master for this transaction. At the same time that TS
is asserted, the bus master also drives the address and all
other relevant control signals that define the transaction. TS
is only asserted for one cycle but all other signals are held
valid by the bus master until some other device asserts
AACK. The device that asserts AACK becomes the slave to
this transaction. Typically, the slave is the memory controller,
although for transactions that are cache hits the MPC2605
becomes the slave by driving L2 CLAIM.
Transactions can be aborted by any device on the bus by
asserting ARTRY. ARTRY may be asserted at any time after
TS is asserted, but must be held through the cycle after
AACK is asserted. This cycle is referred to as the ARTRY
window, since it is the cycle in which all devices sample
ARTRY to determine if the address tenure has completed
successfully.
MOTOROLA
If an address tenure is not aborted by an assertion of
ARTRY, then the next bus master is free to assert TS, the
cycle after the ARTRY window to start a new address tenure.
If ARTRY is asserted in the ARTRY window, all devices that
are not asserting ARTRY must negate their bus request in
the following cycle. This next cycle is called the BR window.
The purpose of this protocol is to give immediate bus mastership to the device that asserted ARTRY with the expectation
that that device will take this opportunity to clean up whatever circumstances caused it to assert ARTRY. Typically, this
involves writing data back to memory to maintain coherence
in the system.
Data Tenures
Data tenures are more complicated to define than address
tenures. They require two conditions to start: an assertion of
TS that initiates a data transaction and a qualified assertion
of the bus master’s data bus grant. For a data bus grant to be
considered qualified, no device on the bus may be asserting
DBB in the cycle that the data bus grant is asserted.
Data transactions come in two types: single–beat transactions and burst transactions. The type is determined by the
state of TBST during the address tenure of the transaction. If
the bus master asserts TBST, the transaction is a burst transaction and will require four assertions of TA in order to complete normally. If TBST is negated during the address tenure,
the transaction only requires one assertion of TA, thus the
name single–beat.
Which device drives the data bus during a data transaction
depends upon whether the transaction is a read or a write.
For a read transaction, the slave device drives the data bus.
For a write transaction, the master drives the data bus. In all
data transactions, the slave device asserts TA to indicate
that either valid data is present on the bus, in the case of a
read; or that it is reading data off the data bus, in the case of
a write. The master device asserts DBB the cycle after it has
been granted the data bus and keeps it asserted until the
data tenure has completed.
A data tenure can be aborted in two different ways. The
address tenure for the transaction can be aborted by an
assertion of ARTRY. Or, the slave device may assert TEA to
indicate that some error condition has been detected. Either
event will prematurely terminate the data tenure.
Data Streaming
For the majority of data transactions there must be a wait
state between the completion of one data tenure and the
start of the next. This turnaround cycle avoids the contention
on the data bus that would occur if one device starts driving
data before another device has had a chance to turn off its
data bus drivers. When a cache read hit is pipelined on top of
another cache read hit, there is no need for this turnaround
cycle since the same device will be driving the data bus for
both data tenures. The 60X bus has the ability to remove this
unnecessary wait state and allow back–to–back cache read
hits to stream together. This ability is only enabled if the system is put into Fast L2 mode. Note that not all PowerPC processors support Fast L2 mode.
One of the requirements for taking advantage of this data
streaming capability is that the system arbiter must be
sophisticated enough to identify situations in which streaming may occur. Upon recognizing these situations, it must assert the processor’s data bus grant in the cycle coincident
MPC2605
11
with the fourth assertion of TA of the first cache read, so that
the data tenure for the second cache read may commence in
the next cycle.
Because it only recognizes qualified assertions of
CPU DBG, the MPC2605 must not be aware of the processor’s assertions of DBB. This means that the DBB pin of the
MPC2605 must be tied to a pullup resistor rather than connected to the system DBB to which all other devices are
connected. This forces the system arbiter to a level of sophistication such that it only supplies qualified data bus
grants and thus the DBB signal is unnecessary to the whole
system.
Note: In a multi–chip configuration each MPC2605 device
acts as an independent cache. Zero wait state data streaming can only occur if the back to back read hits occur in a given device. If the second read hit is not in the device as the
first read hit, a wait state will occur between the two data tenures (2–1–1–1–2–1–1–1 timing).
Data Bus Parking
The MPC2605 has the ability to respond to a processor
read or write hit starting in the cycle after the processor has
asserted TS. This is referred to as a 2–1–1–1 response.
However, even though the MPC2605 has this ability, it is dependent upon the system to allow this quick of a response to
occur. As discussed above, a data tenure cannot start until
the master has been given a qualified bus grant. In order for
the data tenure to start the cycle after TS is asserted, the
data bus must be granted in the cycle coincident with the
assertion of TS. At bus speeds of 66 MHz it is extremely difficult for an arbiter to detect an assertion of TS and itself assert CPU DBG in the same cycle. In order to realistically
allow this situation to occur, CPU DBG must be asserted independent of the processor’s assertion of TS.
Data bus parking is a system feature whereby the processor always has a qualified data bus grant when the data bus
is idle. It is also a requirement for systems which seek to take
advantage of the 2–1–1–1 response time capabilities of the
MPC2605. This feature is typically present in arbiters that
have the level of sophistication necessary to support data
streaming. But it is also a feature of systems that do not even
have a data bus arbiter. In these systems the data bus grant
of every device in the system is tied to ground. The assertion
of DBB by the current data bus master effectively removes
the qualified data bus grant of all devices in the system, including its own. Note that in systems that have no data bus
arbiter that it is impossible to take advantage of data streaming.
There is another caveat associated with data bus parking.
Care must be taken when using data bus parking along with
Fast L2 mode. In normal bus mode when the processor
reads data off the bus, it will wait one cycle before passing
the data on to internal functional units. The purpose of this
one cycle waiting period is to check for an assertion of
DRTRY, which invalidates the data that has been already
read. One of the advantages of running the processor in Fast
L2 mode is that this internal processor wait state is removed.
A problem will arise, however, if the processor is given
data the cycle after TS is asserted, as is possible with the
MPC2605, and the transaction is aborted by some other device asserting ARTRY. Because the processor will not sam-
MPC2605
12
ple ARTRY until two cycles after the assertion of TS, the data
read off the bus will have already been forwarded to the
internal functional units. Thus, incorrect results may occur in
the system.
To avoid this situation in a system that seeks to run Fast L2
mode with the data bus parked, there must be a guarantee
that ARTRY will never be asserted for cache read hits. This
is a further requirement to be imposed upon the DMA bridge
and the memory controller. If this guarantee cannot be made,
the data bus cannot be parked when running in Fast L2
mode.
Processor Reads
When the processor issues a read transaction, the
MPC2605 does a tag lookup to determine if this data is in
the cache. If there is a cache hit and CI is not asserted, the
MPC2605 will assert L2 CLAIM and supply the data to the
processor when the data tenure starts.
If the processor issues a cache–inhibited read (CI asserted) and the MPC2605 detects a cache hit to a non–dirty,
or clean, cache line, the line will be marked invalid. If the
cache–inhibited read hits a dirty line, the MPC2605 will assert ARTRY and write the dirty line back to memory.
If the read misses in the cache, the MPC2605 will perform
a linefill only if it is a burst read and it is not marked cache–inhibited. During a linefill, the MPC2605 stores the data present on the bus as it is supplied by the memory controller.
Processor Writes
The conditions for asserting L2 CLAIM for processor writes
are almost the same as for processor reads. There must be a
cache hit and CI must not be asserted. In addition, however,
WT must not be asserted. Single beat writes that are marked
either write–through or cache–inhibited that hit in the cache
cause the MPC2605 to assert ARTRY and write the dirty line
back to memory.
Transaction Pipelining
As explained in Pipeline Depth, the MPC2605 can only
handle one level of pipelining on the bus. Since the assertion
of L2 CLAIM gives it the ability to assert AACK, the
MPC2605 has the ability to control this pipeline depth for
transactions that are cache hits by delaying its assertion of
AACK.
Pipelined cache hits are transactions that hit in the cache
but occur while there is still an outstanding data transaction
on the bus. The timing of the assertion of AACK for a pipelined cache hit is dependent upon the completion of the previous transaction. For explanation purposes, the previous
transaction will be referred to as transaction one. The pipelined cache hit will be referred to as transaction two.
If transaction one is a cache hit, the MPC2605 will be the
slave device for the transaction. Since, for burst operations,
the MPC2605 always asserts TA for four consecutive clock
cycles, the end of the data tenure for transaction one will be
at a deterministic clock cycle. In this case, AACK for transaction two can be asserted coincident with the last assertion of
TA for transaction one. If transaction one is not a cache hit,
the MPC2605 will wait until after the data tenure for transaction one has completed before asserting AACK to complete
the address tenure of transaction two.
MOTOROLA
MEMORY COHERENCE
When a processor brings data into its on–chip cache and
modifies it, a situation has arisen in which the main memory
now contains irrelevant, or stale, data. Given that most systems support some form of DMA there must exist a means by
which the processor is forced to write this modified, or dirty,
data back to main memory. The DMA bridge is responsible
for generating bus transactions to ensure that main memory
locations accessed by DMA operations do not contain stale
data. These transactions, called snoops, come in three different categories, each of which will be discussed below.
Snoops cause the processor and the MPC2605 to check
to see if they have dirty copies of the memory location specified in the snoop transaction. If either device does have a
dirty copy it will assert ARTRY and make use of the opportunity presented in the BR window to write this data back to
main memory.
Situations can arise where a cache line is dirty in both the
processor’s L1 cache and in the MPC2605. In cases such as
these, snoop transactions should cause the processor to
write its data back to memory since it is by definition more
recent than the data in the MPC2605. Since ARTRY is a
shared signal and it cannot be determined which devices are
driving it, the MPC2605 samples CPU BR in the BR window
to determine if the snoop hit a dirty line in the L1 cache. If
CPU BR is asserted during this window, the MPC2605 will
defer to the processor.
Snoop Reads
A snoop read causes dirty data to be written back to
memory but allows both the L1 and L2 to keep a valid copy.
In cases where the snoop hits a dirty cache line in the processor, the MPC2605 will update its contents as the processor writes the data back to main memory.
Snoop reads can be implemented in two ways. One is that
the DMA bridge can issue a clean transaction (TT[0:4] =
00000). The other is that the DMA bridge can do a read
transaction (TT[0:4] = x1010). If the DMA bridge does a read
transaction, the MPC2605 determines that it is a snoop read
rather than a processor read by the state of CPU BG the
cycle before TS was asserted. If the processor was not
granted the bus then the transaction had to have been issued by the DMA bridge and is therefore a snoop read.
Snoop Writes
Snoop writes also cause dirty data to be written back to
main memory. The difference from a snoop read is that the
cache line must then be invalidated in both the processor’s
cache and in the L2 cache. When the processor writes data
back to memory in response to a snoop write, the MPC2605
will not cache the data as it appears on the bus. If a valid
copy resides in the cache, the MPC2605 will invalidate it.
Again there are multiple transactions that can be used by
the DMA bridge to implement a snoop write. It can issue a
flush transaction (TT[0:4] = 00100), a read with intent to
modify (TT[0:4] = x1110), or a write with flush (TT[0:4]
= 00010). As with snoop reads, the MPC2605 distinguishes
between processor issued data transactions and snoop
transactions by the state of CPU BG in the cycle previous to
the assertion of TS.
MOTOROLA
Snoop Kills
Kills are snoops that cause cache entries to be immediately invalidated, regardless of whether they are dirty. This
saves time if the DMA operation is going to modify all the
data in the cache line. To implement a snoop kill the DMA
bridge can issue a kill transaction (TT[0:4] = 01100) or a write
with kill (TT[0:4] = 00110).
TWO/FOUR CHIP IMPLEMENTATION
Multiple Castouts
Because each MPC2605 has its own castout buffer
(COB), it is possible for situations to arise in which more than
one device needs to do a copyback operation. Under normal
circumstances each device will enter castout conditions at
different times. In these cases, when a device determines
that it needs to do a castout, the L2 BR signal is first
sampled. If L2 BR is already asserted then it is clear that
another device is also in a castout situation. The late device
will wait until L2 BR is negated before continuing in its attempt to perform its castout.
Because of the BR window protocol associated with assertions of ARTRY, it is possible for a situation to arise where
device two is waiting for device one to do its castout before
asserting L2 BR. If there is an assertion of ARTRY by a device other than device one, device one is required to negate
L2 BR in the BR window. In order to prevent device two from
interpreting device one’s negation of L2 BR as an indication
that device one has completed its castout, a simple arbitration mechanism is used. All devices have a simple two–bit
counter that is synchronized such that all counters always
have the same value. For the purposes of performing a castout operation, a given pair can only assert L2 BR if the counter is equal to its value of CFG[1:2]. This simple mechanism
prevents more than one device from asserting L2 BR in the
same cycle and therefore not being cognizant of the another
device’s need to perform a castout.
Snoop Hit Before Castout
The other situation that can cause problems with a shared
bus request occurs when a snoop hits a dirty line in one of
the MPC2605 devices. If device one has a cache line in its
COB, it will assert L2 BR so that it may perform a castout
operation. If a snoop hits a dirty line in device two, it will assert both ARTRY and L2 BR so that it can write the snoop
data back to main memory. When device one detects that
ARTRY has been asserted, it needs to be made aware that
device two needs to request the bus. Otherwise, at the same
time that device two is asserting L2 BR, device one will attempt to conform to the BR window protocol and negate
L2 BR. This situation is avoided by device one sampling FDN
when it detects that ARTRY has been asserted. If FDN is
asserted at the same time as ARTRY is asserted, device one
will recognize that device two is asserting ARTRY. device
one will then high–Z L2 BR so that there will not be contention when device two is asserting L2 BR.
MULTIPROCESSING
The MPC2605 can be used as a common cache for up to
four processors. For each processor there is a bus request,
bus grant, and data bus grant signal pin on the MPC2605.
Each of these pins needs to be connected to the respective
processor’s arbitration signals in the system.
MPC2605
13
The MPC2605 treats multiple processors as one processor. Thus, the same restrictions on pipelining depth are true
with regard to how many processor transactions can be outstanding at any one time. There can only be one data
transaction from ANY processor pipelined on top of a current
data transaction that was issued by ANY processor.
The data tenures for all processors must be performed in
the same order as the address tenures on a system–wide
basis. If processor one makes a request and then processor
two makes a request, processor one’s data tenure must precede processor two’s data tenure. Note that this is not a 60X
bus restriction, but rather a restriction necessary for proper
operation of the MPC2605.
The MPC2605 keeps coherent with the L1 caches of multiple processors as defined by the MESI (modified–exclusive–
shared–invalid) protocol without actually implementing the
protocol. This is possible for two reasons. Since the
MPC2605 is a look–aside cache, all transactions are monitored by all devices on the bus. Also, the MPC2605 cannot,
on its own, modify data. Thus, if one processor requests exclusive access to a cache line, it is not necessary for the
MPC2605 to invalidate its copy of the data, as would be required under the MESI protocol. If a second processor requests the same data, the transaction will cause the first
processor to assert ARTRY. This will prevent the MPC2605
from supplying stale data to the second processor.
As discussed in Data Bus Parking, care must be taken
when parking the data bus in Fast L2 mode. By the nature of
MP systems running under the MESI protocol there will be
assertions of ARTRY to abort cache read hits. Thus, in an
MP system, the data bus cannot be parked to any processor
if the system is to be run in Fast L2 mode.
PWRDN
An assertion of PWRDN will cause the MPC2605 to go into
a low–power sleep state. This state is entered after PWRDN
is synchronized and both the address and data buses are
idle. All data is retained while in the sleep state.
The behavior of the MPC2605 upon negation of PWRDN
is dependent upon the state of WT at the rising edge of
HRESET. If WT is asserted at reset, the MPC2605 will
invalidate all cache entries when PWRDN is negated. If WT
is negated at reset, the MPC2605 will leave all cache entries
as they were prior to the assertion of PWRDN. However, in
this situation, the system designer must insure that no bus
activity occur within two microseconds of the negation of
PWRDN.
Note: While in the sleep state the MPC2605 does not disable its internal clock network. The low power state current
stated in this specification assumes that the system clock is
not toggling.
synchronize them internally. This process takes eight clock
cycles. Thus, to guarantee recognition by the MPC2605,
assertions of any one of these signals must last a minimum
of eight clock cycles.
L2 FLUSH
When L2 FLUSH is asserted, the MPC2605 initiates an internal sequence that steps through every cache line present.
Valid lines that are clean are immediately marked invalid.
Valid lines that are dirty must be written back to main
memory.
To keep memory up to date, the MPC2605 must still monitor all transactions on the bus. Any transaction that is not a
processor burst write will cause the MPC2605 to assert
ARTRY. Burst writes cause the MPC2605 to do a lookup on
the affected address and mark the line invalid if it is present.
Because the MPC2605 must still monitor all transactions,
it cannot use the tag RAM for the flush sequence unless
there is a guarantee that no new transaction will be initiated
on the bus. The only way to ensure that no new transactions
will occur is for the MPC2605 to be granted the bus. Thus,
upon entering the sequence initiated by the assertion of
L2 FLUSH, the MPC2605 will assert L2 BR. As soon as
L2 BG is asserted, the MPC2605 can start stepping through
the tag RAM entries.
L2 FLUSH need not be held asserted for the flush sequence to complete. Once started the sequence will run to
completion unless overridden by an assertion of HRESET.
L2 MISS INH
When L2 MISS INH is asserted, the MPC2605 will not load
any new data into the cache. The data already present will
remain valid and the MPC2605 will respond to cache hits.
This condition only lasts as long as L2 MISS INH is asserted.
When L2 MISS INH is negated, the MPC2605 will start to
bring new data into the cache when there are cache misses.
L2 TAG CLR
When L2 TAG CLR is asserted, the MPC2605 will invalidate all entries in the cache. This internal sequence is the
same as the one initiated by an assertion of HRESET. During
this sequence, the MPC2605 will not participate in any bus
transaction. However, it will keep track of all bus transactions
so that when the sequence is finished, the MPC2605 can immediately participate in the next bus transaction.
As is the case with assertions of L2 FLUSH, an assertion
of L2 TAG CLR need not be held for the duration of the sequence. Once asserted the sequence will run to completion
regardless of the state of L2 TAG CLR.
L2 UPDATE INH
ASYNCHRONOUS SIGNALS
The MPC2605 supports four asynchronous control signals. These signals were originally defined in the PowerPC
reference platform (PReP) specification. Because these signals are defined to be asynchronous, the MPC2605 must
MPC2605
14
When L2 UPDATE INH is asserted, the MPC2605 is
disabled from responding to cacheable transactions. Bus
transactions continue to be monitored so that as soon as
L2 UPDATE INH is negated, the MPC204GA can participate
in the next transaction.
MOTOROLA
READ HIT/WRITE HIT
Figure 1 shows a read hit from an idle bus state. The
MPC2605 asserts L2 CLAIM the cycle after TS to inform the
memory controller that there is a cache hit and the cache will
control the rest of the transaction. L2 CLAIM is held through
the cycle after AACK is asserted. Since there are no active
data tenures from previous transactions, the MPC2605
asserts AACK the cycle after TS is asserted. Note there must
1
be a qualified assertion of CPU DBG in the same cycle as the
assertion of TS for the MPC2605 to respond with TA in the
next cycle. CPU DBG does not affect the timing of L2 CLAIM
or AACK.
The write hit timing is virtually the same. The only difference is the processor drives the data instead of the
MPC2605.
2
3
4
5
A1
A2
A3
A4
6
CLK
CPU BG
TS
A
A0 – A31
TBST
L2 CLAIM
AACK
CPU DBG
DBB
TA
DH0 – DH31,
DL0 – DL31
LEGEND
Signal driven to the MPC2605
Signal driven by the MPC2605
High–Z
Figure 1. Burst Read (or Write) Hit
MOTOROLA
MPC2605
15
MULTIPLE READ/WRITE HITS (NORMAL BUS MODE)
Figure 2 is an illustration of MPC2605 pipeline depth limit
with multiple read hits. The MPC2605 supports only one
level of address pipelining for data transfer. Therefore, it
must hold off on its assertion of AACK for a pipelined TS until
1
2
3
the data tenure for the first TS is done. The MPC2605
asserts AACK at the same time as the fourth TA for data
tenures that it controls.
4
5
6
7
8
9
10
B3
B4
11
CLK
CPU BG
TS
A0 – A31
A
B
C
TBST
L2 CLAIM
AACK
CPU DBG
DBB
TA
DH0 – DH31,
DL0 – DL31
A1
A2
A3
A4
B1
B2
LEGEND
Signal driven to the MPC2605
Signal driven by the MPC2605
High–Z
Figure 2. Multiple Burst Read (or Write) Hits
MPC2605
16
MOTOROLA
READ MISS (NORMAL BUS MODE)
Figure 3 is an illustration of MPC2605 pipeline depth with a
read miss followed by a read hit.
For illustration purposes the read miss is shown as a
3–1–1–1 response from memory. AACK for the second
access is not driven true until the cycle after the fourth TA of
1
2
3
4
the read miss. This is because the MPC2605 is not in control
of TA for the first access and must, therefore, wait until the
first access’ data tenure is complete before it can drive AACK
true for the read hit.
5
6
7
8
9
10
11
B1
B2
B3
B4
12
CLK
CPU BG
TS
A0 – A31
A
B
TBST
L2 CLAIM
AACK
CPU DBG
DBB
TA
DH0 – DH31,
DL0 – DL31
A1
A2
A3
A4
LEGEND
Signal driven to the MPC2605
Signal driven by the MPC2605
High–Z
Figure 3. Read Miss Followed by a Burst Read Hit for MPC603/604
MOTOROLA
MPC2605
17
MULTIPLE READ HITS (FAST L2 MODE)
Back to back pipelined burst read hits for the MPC604 in
Fast L2 mode, also called data streaming mode, are shown
in Figure 4. Note that CPU DBG is negated except for the
cycles coincident with the fourth TA of each data tenure. This
1
2
3
4
is a requirement for data streaming. Note also that DBB is
not shown. For proper operation in Fast L2 mode the DBB
pin of the MPC2605 must be tied to a pull–up resistor.
5
6
7
8
9
10
11
12
CLK
CPU BG
TS
A
A0 – A31
B
C
D
TBST
L2 CLAIM
AACK
CPU DBG
TA
DH0 – DH31,
DL0 – DL31
A1
A2
A3
A4
B1
B2
B3
B4
C1
C2
C3
LEGEND
Signal driven to the MPC2605
Signal driven by the MPC2605
High–Z
Figure 4. Multiple Burst Read Hits in Fast L2 Mode
MPC2605
18
MOTOROLA
WRITE THROUGH BURST WRITE HIT
Figure 5 shows the fastest possible burst write hit to a
write–through mode L2 cache line, read miss or write miss
processing that replaces a clean line. For these operations
MPC2605 will not assert any signals on the 60X bus. A
cache line is considered write through if WT is asserted by
1
2
the processor when it asserts TS.
The speed at which a write–through operation completes
is solely dependent on the memory controller. The timing
shown here assumes that the memory controller has a write
buffer that can accept data this quickly.
3
4
5
A2
A3
A4
6
7
8
CLK
CPU BG
TS
A
A0 – A31
TBST
WT
L2 CLAIM
AACK
CPU DBG
DBB
TA
DH0 – DH31,
DL0 – DL31
A1
LEGEND
Signal driven to the MPC2605
Signal driven by the MPC2605
High–Z
Figure 5. Fastest Possible Write Through Burst Write Hit for MPC603/604
MOTOROLA
MPC2605
19
READ/WRITE MISS
Figure 6 is an illustration of a processor read or write miss
that causes the MPC2605 to replace a dirty line. L2 BR is
asserted two clocks after TS. The dirty data to be replaced is
moved into the internal cast out buffer (COB) at the same
time the new data is written into the cache. Note that the
1
2
3
4
copyback operation occurs after the processor request is
satisfied. In addition, no delay is added to the processor
transaction. It proceeds as fast as the memory controller will
allow.
5
6
7
8
9
10
11
B1
B2
B3
B4
12
CLK
CPU BG
L2 BR
L2 BG
TS
A
A0 – A31
B
TBST
L2 CLAIM
AACK
CPU DBG
L2 DBG
DBB
TA
DH0 – DH31,
DL0 – DL31
A1
A2
A3
A4
LEGEND
Signal driven to the MPC2605
Signal driven by the MPC2605
High–Z
Figure 6. Read or Write Miss Followed by Castout
MPC2605
20
MOTOROLA
READ/WRITE SNOOP HIT (DIRTY L2 LINE)
Figure 7 is an illustration of a read or write snoop to a
cache line that is dirty in the L2, but is not dirty in the
processor’s cache. When a snoop hits a dirty line, the
MPC2605 will assert ARTRY through the cycle following the
assertion of AACK. This cycle is called the ARTRY window.
Note that the MPC2605 also asserts L2 BR at the same time
it asserts ARTRY. Because the snoop could also have hit a
dirty line in the processor’s cache, the MPC2605 samples
1
2
3
4
the processor’s BR signal the cycle following the ARTRY
window. This cycle is called the BR window. If the
processor’s BR signal is not asserted, the MPC2605 will start
sampling L2 BG, the cycle after the BR window.
Note that the MPC2605 cannot do a 2–1–1–1 copy back
burst. The earliest that it can handle the first assertion of TA
is two cycles after its assertion of TS.
5
6
7
8
9
10
11
12
A3
A4
CLK
CPU BR
CPU BG
L2 BR
L2 BG
TS
A0 – A31
A
A
A
L2 CLAIM
AACK
ARTRY
CPU DBG
L2 DBG
DBB
TA
DH0 – DH31,
DL0 – DL31
A1
A2
LEGEND
Signal driven to the MPC2605
Signal driven by the MPC2605
High–Z
Figure 7. Read or Write Snoop Hit to Dirty L2 Cache Line and Clean Processor Cache Line
MOTOROLA
MPC2605
21
READ/WRITE SNOOP HIT (DIRTY L2 AND PROCESSOR LINE)
An illustration of PowerPC read or write snoop hit to a dirty
L2 cache line is shown in Figure 8. The processor has a dirty
copy of the cache line. In this case, both the processor and
the MPC2605 assert ARTRY. This situation is detected by
sampling CPU BR in the BR window, as described in the
previous example. If CPU BR is asserted in the BR window,
1
2
3
4
the MPC2605 will negate L2 BR. It will also ignore assertions
of L2 BG. This allows the processor to write back its dirty
cache line, at which time the MPC2605 will either update or
invalidate its copy depending on whether it is a snoop read or
snoop write.
5
6
7
8
9
10
11
12
13
14
CLK
CPU BR
CPU BG
L2 BR
L2 BG
TS
A0 – A31
A
A
A
L2 CLAIM
AACK
ARTRY
CPU DBG
L2 DBG
DBB
TA
DH0 – DH31,
DL0 – DL31
A1
A2
A3
A4
LEGEND
Signal driven to the MPC2605
Signal driven by the MPC2605
High–Z
Figure 8. Read or Write Snoop Hit to Dirty L2 Cache Line and Dirty Processor Cache Line
MPC2605
22
MOTOROLA
READ HIT/WRITE HIT (WITHOUT CPU DBG PARKED)
Most of the previous examples have assumed CPU DBG
is asserted in the same cycle that the processor asserts TS.
This implies CPU DBG is parked. In some systems it may not
be desirable or possible to park CPU DBG. Figure 9 shows
1
2
the response for a read hit from the MPC2605 is gated by the
assertion of CPU DBG. The fastest response possible in a
system that does not park CPU DBG is 3–1–1–1.
3
4
5
6
7
CLK
CPU BG
TS
A
A0 – A31
TBST
L2 CLAIM
AACK
CPU DBG
DBB
TA
DH0 – DH31,
DL0 – DL31
A1
A2
A3
A4
LEGEND
Signal driven to the MPC2605
Signal driven by the MPC2605
High–Z
Figure 9. Burst Read (or Write) Hit Without CPU DBG Parked
MOTOROLA
MPC2605
23
JTAG
AC OPERATING CONDITIONS AND CHARACTERISTICS
FOR THE TEST ACCESS PORT (IEEE 1149.1)
(TJ = 20 to + 110°C, Unless Otherwise Noted)
Input Timing Measurement Reference Level . . . . . . . . . . . . . . . 1.5 V
Input Pulse Levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0 to 3.0 V
Input Rise/Fall Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 ns
Output Measurement Timing Level . . . . . . . . . . . . . . . . . . . . . . . 1.5 V
Output Load . . . . . . . . . . . . . . . . . . . . . . . . 50 Ω Termination to 1.5 V
TAP CONTROLLER TIMING
MPC2605–66
P
Parameter
S b l
Symbol
Min
Max
U i
Unit
tCK
30
—
ns
Clock High Time
tCKH
12
—
ns
1
Clock Low Time
tCKL
12
—
ns
1
tA
5
9
ns
Clock Low to Output High–Z
tCKZ
0
9
ns
2
Clock Low to Output Active
tCKX
0
9
ns
3, 4
Cycle Time
Clock Low to Output Valid
N
Notes
Setup Times:
TMS
TDI
ts
tsd
2
—
ns
1
Hold Times:
TMS
TDI
th
thd
2
—
ns
1
NOTES:
1. This parameter is sampled and not 100% tested.
2. TDO will High–Z from a clock low edge depending on the current state of the TAP state machine.
3. TDO is active only in the SHIFT–IR and SHIFT–DR state of the TAP state machine.
4. Transition is measured ± 500 mV from steady–state voltage. This parameter is sampled and not 100% tested.
tCKH
tCK
tCKL
TCK TEST
CLOCK
tS
tH
TMS TEST
MODE SELECT
tSD
tHD
TDI TEST
DATA IN
tA
tCKZ
tCKX
TDO TEST
DATA OUT
Figure 10. TAP Controller Timing
MPC2605
24
MOTOROLA
TEST ACCESS PORT DESCRIPTION
INSTRUCTION SET
A five pin IEEE Standard 1149.1 Test Port (JTAG) is included on this device. When the TAP (Test Access Port) controller is in the SHIFT–IR state, the instruction register is
placed between TDI and TDO. In this state, the desired
instruction would be serially loaded through the TDI input.
TRST resets the TAP controller to the test–logic reset state.
The TAP instruction set for this device are as follows.
STANDARD INSTRUCTIONS
Instruction
Code
(Binary)
BYPASS
1111*
Bypass instruction
SAMPLE/PRELOAD
0010
Sample and/or preload
instruction
Description
EXTEST
0000
Extest instruction
HIGHZ
1001
High–Z all output pins while
bypass register is between
TDI and TDO
CLAMP
1100
Clamp output pins while
bypass register is between
TDI and TDO
* Default state at power–up.
pins. The EXTEST instruction would then be loaded. During
EXTEST, the boundary–scan register is placed between TDI
and TDO in the SHIFT–DR state of the TAP controller. Once
the EXTEST instruction is loaded, the TAP controller would
then be moved to the run–test/idle state. In this state, one
cycle of TCK would cause the preloaded data on the output
pins to be driven while the values on the input pins would be
sampled. Note the TCK, not the clock pin (CLK), is used as
the clock input while CLK is only sampled during EXTEST.
After one clock cycle of TCK, the TAP controller would then
be moved to the SHIFT–DR state where the sampled values
would be shifted out of TDO (and new values would be
shifted in TDI). These values would normally be compared to
expected values to test for board connectivity.
CLAMP TAP INSTRUCTION
The CLAMP instruction is provided to allow the state of the
signals driven from the output pins to be determined from the
boundary–scan register while the bypass register is selected
as the serial path between TDI and TDO. The signals driven
from the output pins will not change while the CLAMP
instruction is selected. EXTEST could also be used for this
purpose, but CLAMP shortens the board scan path by inserting only the bypass register between TDI and TDO. To use
CLAMP, the SAMPLE/PRELOAD instruction would be used
first to scan in the values that will be driven on the output pins
when the CLAMP instruction is active.
SAMPLE/PRELOAD TAP INSTRUCTION
HIGHZ TAP INSTRUCTION
The SAMPLE/PRELOAD TAP instruction is used to allow
scanning of the boundary–scan register without causing interference to the normal operation of the chip logic. The
169–bit boundary–scan register contains bits for all device
signal and clock pins and associated control signals. This
register is accessible when the SAMPLE/PRELOAD TAP
instruction is loaded into the TAP instruction register in the
SHIFT–IR state. When the TAP controller is then moved to
the SHIFT–DR state, the boundary–scan register is placed
between TDI and TDO. This scan register can then be used
prior to the EXTEST instruction to preload the output pins
with desired values so that these pins will drive the desired
state when the EXTEST instruction is loaded. As data is written into TDI, data also streams out TDO which can be used
to pre–sample the inputs and outputs.
SAMPLE/PRELOAD would also be used prior to the
CLAMP instruction to preload the values on the output pins
that will be driven out when the CLAMP instruction is loaded.
The HIGH–Z instruction is provided to allow all the outputs
to be placed in an inactive drive state (high–Z). During the
HIGH–Z instruction the bypass register is connected between TDI and TDO.
EXTEST TAP INSTRUCTION
The EXTEST instruction is intended to be used in conjunction with the SAMPLE/PRELOAD instruction to assist in
testing board level connectivity. Normally, the SAMPLE/
PRELOAD instruction would be used to preload all output
MOTOROLA
BYPASS TAP INSTRUCTION
The BYPASS instruction is the default instruction loaded at
power up. This instruction will place a single shift register
between TDI and TDO during the SHIFT–DR state of the
TAP controller. This allows the board level scan path to be
shortened to facilitate testing of other devices in the scan
path.
DISABLING THE TEST ACCESS PORT AND
BOUNDARY SCAN
It is possible to use this device without utilizing the four
pins used for the test access port. To circuit disable the
device, TCK must be tied to VSS to preclude mid level inputs.
TRST should be tied to VSS to ensure proper HRESET operation. Although TDI and TMS are designed in such a way
that an undriven input will produce a response equivalent to
the application of a logic 1, it is still advisable to tie these
inputs to VDD through a 1K resistor. TDO should remain
unconnected.
MPC2605
25
1
TEST–LOGIC
RESET
0
0
RUN–TEST/
IDLE
1
SELECT DR–SCAN
SELECT IR–SCAN
1
0
1
0
1
CAPTURE–DR
CAPTURE–IR
0
0
SHIFT–DR
SHIFT–IR
0
1
1
1
EXIT1–IR
0
0
PAUSE–IR
0
PAUSE 1–DR
1
0
1
0
EXIT2–DR
EXIT2–IR
1
1
UPDATE–DR
1
0
1
EXIT1–DR
0
1
UPDATE–IR
0
1
0
NOTE: The value adjacent to each state transition represents the signal present at TMS at the rising edge of TCK.
Figure 11. TAP Controller State Diagram
MPC2605
26
MOTOROLA
BOUNDARY SCAN ORDER
BIT NUMBER
The order of the boundary scan chain. Bit 0 is the closest
to TDO.
Bit
Number
Bit/Pin Name
Bit/Pin
Type
Output
Enable
23
DH28
I/O
DOE
BIT/PIN NAME
24
DH29
I/O
DOE
The name of the physical pin. For an output enable cell,
this is the name of the corresponding output enable.
25
DH30
I/O
DOE
26
DH31
I/O
DOE
27
DP3
I/O
DOE
28
DH16
I/O
DOE
29
DH17
I/O
DOE
30
DH18
I/O
DOE
31
DH19
I/O
DOE
32
DH20
I/O
DOE
33
DH21
I/O
DOE
34
DH22
I/O
DOE
OUTPUT ENABLE
35
DH23
I/O
DOE
The name of the output enable cell that determines if the
cell is enabled or in the high–Z state. If the pin type is input or
output enable, this entry will be empty.
36
DP2
I/O
DOE
37
L2 BG
Input
38
L2 MISS INH
Input
39
ABB
I/O
40
CPU3 DBG
Input
41
CPU3 BG
Input
BIT/PIN TYPE
Input — Input only pin.
I/O — Bi–directional pin that can be put into high–Z state.
Output — Output only pin.
Output Enable — Boundary scan cell to hold the output
enable state of other I/O pads. Output enable does not
correspond to a physical pin. To set an I/O to an input, the
output enable must have a 1. To set an I/O to an output, the
output enable must have a 0. Note that these internal output
enables are active low.
Reserved – This signal is reserved and must always be a 1.
Bit
Number
Bit/Pin Name
Bit/Pin
Type
0
Reserved
Reserved
1
DL16
I/O
Output
Enable
DOE
42
CPU3 BR
Input
CPU2 DBG
Input
ABBOE
2
DL17
I/O
DOE
43
3
DL18
I/O
DOE
44
CPU2 BG
Input
4
DL19
I/O
DOE
45
CPU2 BR
Input
5
DL20
I/O
DOE
46
FDN
I/O
6
DL21
I/O
DOE
47
L2 DBG
Input
7
DL22
I/O
DOE
48
L2 BR
I/O
L2BROE
8
DL23
I/O
DOE
49
TA
I/O
TAOE
9
DP6
I/O
DOE
50
L2 CLAIM
Output
L2CLAIMOE
CPU DBG
Input
FDNOE
10
DL24
I/O
DOE
51
11
DL25
I/O
DOE
52
AACK
I/O
AACKOE
12
DL26
I/O
DOE
53
CI
I/O
AOE
13
DL27
I/O
DOE
54
ARTRY
I/O
ARTRYOE
14
DL28
I/O
DOE
55
WT
I/O
AOE
15
DL29
I/O
DOE
56
CPU BR
Input
16
DL30
I/O
DOE
57
TEA
Input
17
DL31
I/O
DOE
58
PWRDN
Input
DBB
I/O
18
DP7
I/O
DOE
59
19
DH24
I/O
DOE
60
HRESET
Input
20
DH25
I/O
DOE
61
TBST
I/O
AOE
21
DH26
I/O
DOE
62
TT0
I/O
AOE
22
DH27
I/O
DOE
63
TS
I/O
AOE
MOTOROLA
DBBOE
MPC2605
27
Bit
Number
Bit/Pin Name
Bit/Pin
Type
Output
Enable
Bit
Number
Bit/Pin Name
Bit/Pin
Type
Output
Enable
64
TT1
I/O
AOE
107
DL10
I/O
DOE
65
TT2
I/O
AOE
108
DL11
I/O
DOE
66
TT4
I/O
AOE
109
DL12
I/O
DOE
67
TT3
I/O
AOE
110
DL13
I/O
DOE
68
CPU BG
Input
111
DL14
I/O
DOE
69
SRESET
Input
112
DL15
I/O
DOE
70
L2 TAG CLR
Input
113
DP5
I/O
DOE
71
L2 UPDATE INH
Input
114
A0
I/O
AOE
72
CPU4 BG
Input
115
A1
I/O
AOE
73
CPU4 DBG
Input
116
A2
I/O
AOE
74
CPU4 BR
Input
117
A3
I/O
AOE
75
CFG0
Input
118
A4
I/O
AOE
76
CFG2
Input
119
A5
I/O
AOE
77
CFG1
Input
120
A6
I/O
AOE
79
DH8
I/O
DOE
121
A7
I/O
AOE
79
DH9
I/O
DOE
122
A8
I/O
AOE
80
DH10
I/O
DOE
123
A9
I/O
AOE
81
DH11
I/O
DOE
124
A10
I/O
AOE
82
DH12
I/O
DOE
125
A11
I/O
AOE
83
DH13
I/O
DOE
126
A12
I/O
AOE
84
DH14
I/O
DOE
127
A31
I/O
AOE
85
DH15
I/O
DOE
128
A30
I/O
AOE
86
DP1
I/O
DOE
129
A29
I/O
AOE
87
DH0
I/O
DOE
130
A28
I/O
AOE
88
DH1
I/O
DOE
131
A27
I/O
AOE
89
DH2
I/O
DOE
132
A26
I/O
AOE
90
Dh3
I/O
DOE
133
A25
I/O
AOE
91
DH4
I/O
DOE
134
A24
I/O
AOE
92
DH5
I/O
DOE
135
A23
I/O
AOE
93
DH6
I/O
DOE
136
A22
I/O
AOE
94
DH7
I/O
DOE
137
A21
I/O
AOE
95
DP0
I/O
DOE
138
A20
I/O
AOE
96
DL0
I/O
DOE
139
A19
I/O
AOE
97
DL1
I/O
DOE
140
A18
I/O
AOE
98
DL2
I/O
DOE
141
A17
I/O
AOE
99
DL3
I/O
DOE
142
A16
I/O
AOE
100
DL4
I/O
DOE
143
A15
I/O
AOE
101
DL5
I/O
DOE
144
A14
I/O
AOE
102
DL6
I/O
DOE
145
A13
I/O
AOE
103
DL7
I/O
DOE
146
TSIZ2
I/O
AOE
104
DP4
I/O
DOE
147
TSIZ0
I/O
AOE
105
DL8
I/O
DOE
148
TSIZ1
I/O
AOE
106
DL9
I/O
DOE
149
GBL
Output
AOE
MPC2605
28
MOTOROLA
Bit
Number
Bit/Pin Name
Bit/Pin
Type
Output
Enable
Bit
Number
Bit/Pin Name
150
CFG3
Input
151
L2 CI
Input
152
L2 FLUSH
Input
153
AP0
I/O
AOE
154
AP1
I/O
AOE
155
AP2
I/O
AOE
156
AP3
I/O
AOE
157
APE
Output
APEOE
158
TAOE
159
160
Bit/Pin
Type
161
FDNOE
Output
Enable
162
DBBOE
Output
Enable
163
DOE
Output
Enable
164
ARTRYOE
Output
Enable
165
APEOE
Output
Enable
Output
Enable
166
ABBOE
Output
Enable
L2CLAIMOE
Output
Enable
167
AACKOE
Output
Enable
L2BROE
Output
Enable
168
AOE
Output
Enable
Output
Enable
ORDERING INFORMATION
(Order by Full Part Number)
MPC
2605
ZP
66
X
Motorola PowerPC Prefix
Blank = Trays, R = Tape and Reel
Part Number
Speed (66 = 66 MHz)
Package (ZP = PBGA)
Full Part Number — MPC2605ZP66
MPC2605ZP66R
MOTOROLA
MPC2605
29
PACKAGE DIMENSIONS
ZP PACKAGE
PBGA
CASE 1138–01
4X
0.20
D
A
C
0.20 C
0.25 C
0.35 C
NOTES:
1. DIMENSIONING AND TOLERANCING PER ASME
Y14.5M, 1994.
2. DIMENSIONS IN MILLIMETERS.
3. DIMENSION b IS THE SOLDER BALL DIAMETER
MEASURED PARALLEL TO DATUM C.
E
E2
DIM
A
A1
A2
A3
b
D
D1
D2
e
E
E1
E2
D2
B
TOP VIEW
D1
18X
W
V
U
T
R
P
N
M
L
K
J
H
G
F
E
D
C
B
A
e
MILLIMETERS
MIN
MAX
–––
2.05
0.50
0.70
0.95
1.35
0.70
0.90
0.60
0.90
25.00 BSC
22.86 BSC
22.40
22.60
1.27 BSC
25.00 BSC
22.86 BSC
22.40
22.60
A2
A3
A1
E1
A
SIDE VIEW
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
BOTTOM VIEW
241
b
0.03
M
C A B
0.15
M
C
Motorola reserves the right to make changes without further notice to any products herein. Motorola makes no warranty, representation or guarantee regarding
the suitability of its products for any particular purpose, nor does Motorola assume any liability arising out of the application or use of any product or circuit, and
specifically disclaims any and all liability, including without limitation consequential or incidental damages. “Typical” parameters which may be provided in Motorola
data sheets and/or specifications can and do vary in different applications and actual performance may vary over time. All operating parameters, including “Typicals”
must be validated for each customer application by customer’s technical experts. Motorola does not convey any license under its patent rights nor the rights of
others. Motorola products are not designed, intended, or authorized for use as components in systems intended for surgical implant into the body, or other
applications intended to support or sustain life, or for any other application in which the failure of the Motorola product could create a situation where personal injury
or death may occur. Should Buyer purchase or use Motorola products for any such unintended or unauthorized application, Buyer shall indemnify and hold Motorola
and its officers, employees, subsidiaries, affiliates, and distributors harmless against all claims, costs, damages, and expenses, and reasonable attorney fees
arising out of, directly or indirectly, any claim of personal injury or death associated with such unintended or unauthorized use, even if such claim alleges that
Motorola was negligent regarding the design or manufacture of the part. Motorola and
are registered trademarks of Motorola, Inc. Motorola, Inc. is an Equal
Opportunity/Affirmative Action Employer.
Mfax is a trademark of Motorola, Inc.
How to reach us:
USA / EUROPE / Locations Not Listed: Motorola Literature Distribution;
P.O. Box 5405, Denver, Colorado, 80217. 1-303-675-2140 or 1-800-441-2447
JAPAN: Nippon Motorola Ltd.: SPD, Strategic Planning Office, 141,
4-32-1 Nishi-Gotanda, Shagawa-ku, Tokyo, Japan. 03-5487-8488
Mfax : [email protected] – TOUCHTONE 1-602-244-6609
ASIA/PACIFIC: Motorola Semiconductors H.K. Ltd.; 8B Tai Ping Industrial Park,
Motorola Fax Back System
– US & Canada ONLY 1-800-774-1848 51 Ting Kok Road, Tai Po, N.T., Hong Kong. 852-26629298
– http://sps.motorola.com /mfax /
HOME PAGE : http://motorola.com/sps /
CUSTOMER FOCUS CENTER: 1-800-521-6274
MPC2605
30
◊
MPC2605/D
MOTOROLA