MUAD8K136-83B272C [MUSIC]
Content Addressable SRAM, 4KX272, CMOS, PBGA272, BGA-272;型号: | MUAD8K136-83B272C |
厂家: | MUSIC SEMICONDUCTORS |
描述: | Content Addressable SRAM, 4KX272, CMOS, PBGA272, BGA-272 双倍数据速率 静态存储器 内存集成电路 |
文件: | 总36页 (文件大小:875K) |
中文: | 中文翻译 | 下载: | 下载PDF数据表文档文件 |
Advance Information
MUAD "Harmony" 2M and 1M Ternary CAMs
GENERAL DESCRIPTION
FEATURES
The MUAD "Harmony" Ternary CAM is a fast look-up
table device supporting ternary (0, 1, don’t care) elements
for networking and communication applications. Harmony
is a member of MUSIC Semiconductors RouteCAM
family.
•
•
•
•
16K and 8K x 136-bit full ternary CAMs
Configurable as 8K/4K x 272 or 32K/16K x 68
68-bit interface operates at 13.6Gbit/sec
Sustains 100 million searches per second on a 68-bit
or 136-bit field
The organization of the Harmony 2M part is 16K x 136
bits, and the 1M part is 8K x 136-bit wide, with
double-word and half-word options.
•
50 million searches per second in 272-bit
configuration
•
•
•
Holds multiple word widths within the same device
Synchronous pipelined operation
Harmony is ideally suited for high-speed, high-capacity
functions, including Ethernet and IP address search, data
Up to eight CAMs cascadable without performance
degradation or additional logic
compression,
pattern
recognition,
cache
tags,
high-bandwidth address filtering, and fast routing search
tables. These functions also include privileged, secured, or
encrypted packet-by-packet information utilized in
high-performance Internet equipment such as switches,
firewalls, bridges, and routers.
•
Glueless interface to industry-standard synchronous
SRAMs
•
•
•
Supports IEEE 1149.1 Test Access
1.8 and 3.3V power supply
272-pin BGA package
The flexibility of the Harmony device allows the creation
of multiple search tables within the same device. It
compares, simultaneously, the desired information (data)
against an entire, pre-stored, array of addresses, providing
a performance advantage by reducing search times an
order-of-magnitude over typical binary or tree-based
search algorithms. The Harmony device can be designed
into many applications, but it is particularly well suited to
perform highly intensive search operations.
/TRST
TCLK
TMS
TDI
TDO
CSO[1:0]
/SEN[3:0]
CAM
Cascade
CSI[6:0]
UID[4:0]
MF
/MM
MV
FF
FI[6:0]
CLK
PHASE
/RST
Controller
/ACK
EOT
FO[1:0]
OP[8:0]
OPV
SADR[21:0]
/OE
High-Speed
Interface
SRAM
Interface
/WE
/CE
DQ[67:0]
/ALE
SCLK
Harmony Block Diagram
MUSIC Semiconductors, the MUSIC logo, and the phrase "MUSIC Semiconductors" are
Registered trademarks of MUSIC Semiconductors. MUSIC is a trademark of
MUSIC Semiconductors.
April 4, 2001 Rev. 0
MUAD "Harmony" 2M and 1M Ternary CAMs
April 4, 2001 Rev. 0
CONTENTS
Ball Descriptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2
Functional Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4
Content Addressable Memory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
The I/O Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
SRAM Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
SRAM Addressing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Instruction Bus and DQ Bus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Dual Data Rate Clock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Cascade Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Power Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Test Access Port . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5
Initialization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Depth-Cascading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5
Arbitration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Search (68-bit Configuration with LCAM = 1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Search (68-bit Configuration with LRAM = 1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Depth-Cascading to Generate Full . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6
Harmony Table Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Multiple Search Table Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Multiple Logical Tables of Different Widths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Depth Cascading to Create Larger Logical Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Register Descriptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .8
Comparand Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Global Mask Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Result Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Instruction Register . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Information Register . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Burst Read Address Register . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Burst Write Address Register . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Next Free Address Register . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Configuration Register . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Harmony Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .13
Instruction Codes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .13
Instructions and Instruction Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
READ Instruction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .14
SINGLE READ Instruction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
BURST READ Instruction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
WRITE Instruction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
SINGLE WRITE Instruction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .15
BURST WRITE Instruction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
SEARCH Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Full Word Searches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Half-Word Searches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .17
Double-Word Searches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Write Next Free Address . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
SRAM Addressing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
SRAM READ or WRITE Accesses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
SRAM READ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
SRAM WRITE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .18
Application Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .20
34-Bit Word Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Timing Diagrams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .21
i
Contents
MUAD "Harmony" 1M and 2M Ternary CAMs
Single Location Read Timing Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Write Cycle Timing Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Data and Mask READ (BLEN = 4) for a Group of Cascaded Devices Timing Diagram . . . . . . . . . . . . . . .22
Data and Mask WRITE (BLEN = 4) for a Group of Cascaded Devices Timing Diagram . . . . . . . . . . . . . . 22
68-Bit SEARCH Operation Timing Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
136-Bit SEARCH Operation Timing Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .24
272-Bit SEARCH Operation Timing Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .25
Arbitration for a Group of Cascaded Devices Timing Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
WRITE Timing Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .26
SRAM READ Cycle Timing Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
SRAM WRITE Cycle TIming Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .28
AC Timing Waveforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Electrical Specifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .30
Electrical Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
Capacitance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
Operating Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
Package . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .31
272-Pin BGA Dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Ordering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .32
ii
Ball Descriptions
MUAD "Harmony" 1M and 2M Ternary CAMs
NC
NC
DQ65
DQ57
DQ53
VDD
VDDQ
DQ43
DQ37
DQ31
VDD
DQ27
DQ21
DQ17
VDD
NC
VDDQ
DQ3
NC
NC
A1
A2
A3
A4
A5
A6
A7
A8
A9
A10
A11
A12
A13
A14
A15
A16
A17
NC
A18
A19
NC
A20
NC
VDD
NC
DQ61
NC
DQ49
DQ51
DQ45
VDD
DQ33
DQ29
DQ25
VDDQ
DQ15
DQ13
DQ7
VDD
SADR1
B1
B2
B3
B4
B5
B6
B7
B8
B9
B10
B11
NC
B12
B13
B14
NC
B15
B16
B17
NC
B18
B19
B20
TDO
TDI
NC
DQ63
DQ59
DQ55
VDDQ
NC
DQ39
DQ35
DQ23
DQ19
DQ11
DQ5
SADR0
SADR3
SADR2
C1
C2
C3
C4
C5
C6
C7
C8
C9
C10
C11
C12
NC
C13
C14
C15
C16
C17
C18
NC
C19
C20
UID0
TMS
TCK
GND
DQ67
VDDQ
DQ47
GND
DQ41
VDDQ
VDDQ
GND
VDDQ
DQ9
DQ1
GND
SADR4
VDDQ
D1
D2
D3
D4
D5
D6
D7
D8
D9
D10
D11
D12
D13
D14
D15
D16
D17
D18
D19
NC
D20
UID2
UID3
UID1
/TRST
VDDQ
SADR6
SADR8
E1
E2
E3
E4
E17
E18
E19
E20
VDD
CSI2
CSI0
UID4
SADR5
SADR7
SADR11
VDD
F1
F2
F3
F4
F17
F18
F19
F20
NC
CSI3
VDDQ
CSI1
VDDQ
SADR9
SADR10
SADR13
G1
G2
G3
G4
G17
G18
G19
NC
G20
CSI6
CSI5
CSI4
GND
GND
SADR12
SADR14
H1
H2
H3
H4
H17
H18
H19
H20
BHI0
VDDQ
CSO1
CSO0
GND
GND
GND
GND
SADR15
SADR16
VDDQ
SADR17
J1
J2
J3
J4
J9
J10
J11
J12
J17
J18
J19
J20
VDD
BHI2
VDD
BHI1
GND
GND
GND
GND
VDD
SADR18
SADR19
SADR20
K1
K2
K3
K4
K9
K10
K11
K12
K17
K18
K19
K20
BHO0
BHO1
BHO2
FI0
GND
GND
GND
GND
CLK
SADR21
VDDQ
VDD
L1
L2
L3
L4
L9
L10
L11
L12
L17
L18
/OE
L19
L20
FI1
FI2
VDDQ
FI3
GND
GND
GND
GND
/WE
PHASE
SCLK
M1
FI4
M2
FI6
M3
NC
M4
M9
M10
M11
M12
M17
M18
M19
M20
/CE
GND
GND
/ALE
VDDQ
N1
N2
N3
N4
N17
OP2
N18
OP0
N19
N20
NC
FI5
NC
NC
FO0
OPV
P1
P2
P3
P4
P17
P18
P19
P20
VDD
FO1
VDDQ
NC
OP4
OP3
OP1
VDD
R1
R2
R3
R4
R17
NC
R18
OP6
R19
OP5
R20
NC
NC
VDD
/RST
VDDQ
T1
T2
FF
T3
T4
T17
T18
MF
T19
MV
T20
NC
VDDQ
GND
DQ66
DQ58
DQ54
GND
DQ44
DQ38
DQ30
DQ26
GND
VDDQ
DQ6
DQ0
GND
OP7
U1
U2
U3
U4
U5
U6
U7
U8
U9
U10
U11
U12
U13
U14
U15
U16
U17
U18
U19
/MM
U20
OP8
EOT
/ACK
/SEN3
VDD
/SEN2
DQ56
DQ52
DQ48
VDDQ
DQ36
DQ32
VDDQ
DQ20
DQ14
VDDQ
DQ8
VDDQ
VDD
V1
V2
V3
V4
V5
V6
V7
V8
V9
V10
NC
V11
V12
V13
V14
V15
V16
V17
NC
V18
V19
NC
V20
NC
GND
NC
/SEN1
/SEN0
VDDQ
NC
VDDQ
DQ46
DQ42
DQ34
DQ28
VDD
DQ16
DQ18
DQ12
DQ4
W1
NC
W2
NC
W3
W4
W5
W6
W7
W8
NC
W9
W10
VDD
W11
W12
NC
W13
W14
W15
VDD
W16
NC
W17
W18
DQ2
W19
NC
W20
NC
DQ64
DQ62
DQ60
VDD
DQ50
DQ40
VDDQ
DQ24
DQ22
DQ10
Y1
Y2
Y3
Y4
Y5
Y6
Y7
Y8
Y9
Y10
Y11
Y12
Y13
Y14
Y15
Y16
Y17
Y18
Y19
Y20
Figure 1: Connection Diagram
Rev. 0 April 4, 2001
1
MUAD "Harmony" 1M and 2M Ternary CAMs
Ball Descriptions
BALL DESCRIPTIONS
This section lists and describes the Harmony signals.
Table 1: MUAD Harmony Ball Descriptions
Symbol
Type Description
Pin Number(s)
Clocks and Reset
CLK
I
Master Clock. Harmony samples all the control and data
signals either on the positive edge of CLK, or on the positive
of CLK when PHASE is low.
L17
PHASE
SCLK
/RST
I
O
I
PHASE. This signal runs at half the frequency of CLK and M19
generates an internal clock from CLK.
SRAM Clock. This signal generates clock for the external
SRAM bus. It runs at half the speed of the input CLK.
Reset. Driving /RST low initializes the device to a known
state.
M20
T4
Instruction and DQ Bus
OP[8:0]
I
Instruction Bus. OP[1:0] specify the instruction. OP[8:2]
contain the instruction parameters. The descriptions of
individual instructions explain the details of the parameters.
The encoding of instructions based on the [1:0] fields are:
00:READ
B0:P18, B1:R19, B2:P17, B3:R18, B4:R17, B5:T19, B6:T18,
B7:U20, B8:V20
01:WRITE
10:SEARCH
11:WRITE NEXT FREE ADDRESS
Op Code Valid. OPV qualifies the Instruction bus.
0:No Instruction, 1:Instruction
OPV
I
P19
DQ[67:0]
I/O Address/Data Bus. DQ[67:0] carries the read and write
address and data during register, data, and mask array
operations. It carries the compare data during search
B0:U16, B1:D16, B2:Y18, B3:A18, B4:W18, B5:C16, B6:U15,
B7:B16, B8:V16, B9:D15, B10:Y17, B11:C15, B12:W16,
B13:B15, B14:V14, B15:B14, B16:W14, B17:A14, B18:W15,
operations. It also carries the SRAM address during SRAM B19:C13, B20:V13, B21:A13, B22:Y14, B23:C12, B24:Y13,
accesses.
B25:B12, B26:U12, B27:A12, B28:W12, B29:B11, B30:U11,
B31:A10, B32:V11, B33:B10, B34:W11, B35:C10, B36:V10,
B37:A9, B38:U10, B39:C9, B40:Y9, B41:D9, B42:W9,
B43:A8, B44:U9, B45:B8, B46:W8, B47:D7, B48:V8, B49:B6,
B50:Y7, B51:B7, B52:V7, B53:A5, B54:U7, B55:C6, B56:V6,
B57:A4, B58:U6, B59:C5, B60:Y5, B61:B4, B62:Y4, B63:C4,
B64:Y3, B65:A3, B66:U5, B67:D5
/ACK
T
Read Acknowledge. /ACK indicates that valid data is
available on the DQ bus during register, data word, and word
mask array READ operations, or the data is available on the
SRAM data bus during SRAM READ operations.
V2
EOT
T
T
T
T
I
End of Transfer. EOT indicates the end of burst transfer
during READ or WRITE burst operations.
V1
MF
Match Flag. When asserted, MF indicates that the device is U18
selected in a SEARCH operation.
MV
Match Flag Valid. When asserted, MV qualifies the Match U19
Flag and Multi-Match Flag signals.
/MM
Multi-Match Flag. When asserted, indicates that two or
more matches were found.
V19
/SEN[3:0]
Search Enable. /SEN[3:0] controls which pages within the B0:W4, B1:W3, B2:V5, B3:V3
CAM array participate in SEARCH operations. These pins
have pull-down resistors, so they may be left unconnected.
SRAM Interface
SADR[21:0]
T
SRAM Address. SADR contains address lines to access
B0:C18, B1:B20, B2:C20, B3:C19, B4:D19, B5:F17, B6:E18,
external SRAMs that contains associative data. See Tables B7:F18, B8:E20, B9:G18, B10:G19, B11:F19, B12:H18,
22 and 23 for SRAM bus addressing details.
B13:G20, B14:H20, B15:J17, B16:J18, B17:J20, B18:K18,
B19:K19, B20:K20, B21:L18
N20
/CE
T
T
T
T
SRAM Chip Enable. /CE is the chip enable control for
external SRAMs.
/WE
/OE
/ALE
SRAM Write Enable. /WE is the write enable control for
external SRAMs.
M17
SRAM Output Enable. /OE is the output enable control for M18
external SRAMs.
Address Latch Enable. /ALE is the latch enable control for N18
external SRAMs.
2
April 4, 2001 Rev. 0
Ball Descriptions
MUAD "Harmony" 1M and 2M Ternary CAMs
Table 1: MUAD Harmony Ball Descriptions
Symbol
Type Description
Pin Number(s)
Cascade Interface
CSI[6:0]
I
Cascade In. These pins depth-cascade the device to form a B0:F3, B1:G4, B2:F2, B3:G2, B4:H3, B5:H2, B6:H1
larger table size. One signal of this bus is connected to the
CSO[1] or CSO[0] of each of the upstream devices in a
block. Connect all unused CSI pins to a logic 0. For more
information, see the Depth-Cascading section.
CSO[1:0]
FI[6:0]
O
I
Cascade Out. CSO[1] and CSO[0] are the same logical
signal. CSO[1] or CSO[0] is connected to one input of the
CSI bus of up to four down- stream devices (in a block that
contains up to eight devices).
B0:J4, B1:J3
Full In. Each signal in this bus is connected to FO[0] or
FO[1] of an upstream device to generate the FF signal for
the depth-cascaded block. Connect all unused FI signals to
logic 1.
B0:R2, B1:M1, B3:M4, B4:N1, B5:P1, B6:N2
FO[1:0]
O
Full Out. FO[1] and FO[0] are the same logical signal. One B0:P4, B1:R2
of these two signals must be connected to the FI of up to
four down-stream devices in a depth-cascaded table. Bit[0]
in the CAM array indicates if the entry is full (1) or empty
(0).This signal is asserted if all bits in the CAM array are 1s.
Block Hit In. These pins are used for cascading more than B0:J1, B1:K4, B2:K2
eight devices. They must be tied to GND if cascading eight
or less devices.
BHI[2:0]
BHO[2:0]
FF
I
O
O
Block Hit Out. These pins are used for cascading more that B0:L1, B1:L2, B2:L3
eight devices. Output is NC if cascading eight or less
devices.
Full Flag. When asserted, this signal indicates that the table U2
consisting of all depth-cascaded devices is full.
Device Identification
UID[3:0]
I
Device Identification. The binary-encoded device ID for a B0:D1, B1:E3, B2:E1, B3:E2
depth-cascaded system starts at 0000 and goes up to 0111.
1111 is reserved for a special broadcast address that
selects all cascaded CAMs in the system. On a broadcast
read-only, the device with the LCAM bit set to 1 responds.
UID[4]
I
This device independent bit should be tied to VDD.
F4
Test Access Port Pins
TDI
I
I
Test Data In
Test Clock
C2
D3
C1
D2
E4
TCK
TDO
TMS
/TRST
T
I
Test Data Out
Test Mode Select
Reset
I
Note: The BHI and BHO pins would be used when a group of more than eight devices are cascaded; otherwise, these pins are not
connected or tied to ground.
Rev. 0 April 4, 2001
3
MUAD "Harmony" 1M and 2M Ternary CAMs
Functional Description
FUNCTIONAL DESCRIPTION
The Content Addressable Memory (CAM), High Speed
I/O Interface, Cascade Control, SRAM Interface, Test
Assess Port, and the Instruction and DQ Bus Interface
comprise the Harmony block diagram.
SRAM Addressing
The SRAM address is formed by the information obtained
from the DQ bus and either the lowest match address from
a SEARCH instruction or the address supplied by
Instruction register. The interface timing and control will
select the address from the instruction register by asserting
the applicable READ or WRITE instruction. During a
READ or WRITE instruction to the SRAM, if the
identification (UID) is the global address, then the last
CAM on the SRAM bus of the depth cascaded devices
will drive the SRAM signals (LCAM = 1).
Content Addressable Memory
The CAM section of the Harmony 2M consists of 16,384
136-element ternary words, and the Harmony 1M consists
of 8,192 136-element ternary words, arranged such that
each ternary element contains a data bit and a mask bit.
The combination of data and mask bits determine whether
the ternary element address is a 0, 1, or X (don’t care).
Internally, bit 0 determines if the ternary word contains
valid data; if the bit is set to 0, then the word is available as
it does not contain valid data. This bit is used to determine
the next free address in the device.
Instruction Bus and DQ Bus
OP[8:0] transports the instruction and its associated
parameters. DQ[67:0] is used for data transfer to, and
from, the CAM array. The DQ bus transports the search
data during the SEARCH instruction as well as the
addressing and data during the READ/WRITE operations
of the CAM array, and internal registers. The DQ bus also
carries the address information for SRAM accesses.
The priority encoder generates the address of the word
with the lowest address that satisfies the match criteria
using the searched data words, CAM array words, and the
specified global mask register.
Dual Data Rate Clock
The I/O Interface
The dual data rate clock, configured as cycle A and cycle
B, allows the DQ bus interface to operate at double speed
while maintaining 100 Mhz search rates even though the
I/O width is less than the data width. Hence, only 68 pins,
instead of 136 pins, are required to support 136-bit data
words. Furthermore, Harmony can perform consecutive
searches on 136-bit data words. The phase signal ensures
that these double-speed operations are correctly aligned
with Harmony.
The high-speed input port is a double-data rate, 68-bit,
data bus incorporated with 9-bits to encode instructions,
such as READ and WRITE. The inputs are read on the
rising edge of clock, whereas the phase input is used to
distinguish between the first and second halves of the I/O
cycle. The first half of the I/O cycle transports bits 135:68
and the second half transports bits 67:0.
SRAM Interface
The SRAM interface sections drives the address and
control signals required to access the external SRAM.
Harmony can generate a synchronous output clock
(SCLK) to perform SRAM accesses. Using the SCLK
signal Harmony reduces the amount of required interface
logic by synchronously driving the SRAM address and
control signals. When cascaded, the Harmony device
which contains a match in its Results register will drive
the SRAM bus. However, in the case where a no match
exists, the last Harmony device of the cascade (LRAM =
1) will drive the SRAM bus. Also, when cascaded, this
section also inserts pipeline delays for the SRAM address
and SRAM control for Harmony. The SRAM data bus is
connected to the appropriate host ASIC, therefore SRAM
data does not pass through Harmony.
Cascade Control
The cascade control section drives the cascade output
(CSO) signal when the Harmony devices are depth
cascaded. Up to eight Harmony devices can be
depth-cascaded. Harmony also contains the control logic
to determine if the entry in a single device is full or if the
table consisting of multiple devices is full. In addition, the
cascade control section provides support for multiple
matches. Although the cascade control section does not
drive the validity of matches, the success of matches,
multiple matches, or the required SRAM signals (these
signals are located in the controller section of the block
diagram), it does contain the control logic to enable the
output for these signals.
4
April 4, 2001 Rev. 0
Functional Description
MUAD "Harmony" 1M and 2M Ternary CAMs
Figure 2 shows how up to eight devices can cascade to
form a 128K x 68, 64K x 136, or 32K x 272 bit table and
the interconnection between the devices for
depth-cascading. Additionally, the host ASIC must
program the table size (TLSZ) field to 01. For each search,
if a device determines a local match within the device, it
asserts the CSO[1:0] signals.
Power Management
The power management feature within Harmony reduces
power dissipation by limiting search operations to selected
portions with the CAM. If known beforehand, the desired
data can be isolated and only those portions will need to be
selected for the SEARCH instruction. The input pins
(/SEN[3:0]) independently control four equal sections of
the device. If fewer sections are desired then the designer
can connect multiple inputs together. To disable power
management, set bit 0 of the Configuration register or
connect /SEN[3:0] to GND.
SRAM
DQ[67:0]
6
5
5
5
5
5
4
4
4
4
3
2
1
0
CSI
Harmony
CSO[1]
CSO[0]
6
3
CSI
2
1
0
0
0
Test Access Port
Harmony
CSO[1]
CSO[0]
The Harmony test access port provides an interface for
manufacturing tests and consists of the boundary scan
access port used to support the standard JTAG IEEE
1149.1.
6
3
CSI
2
1
Harmony
CSO[1]
CSO[0]
Initialization
6
3
CSI
2
1
After a hard or soft reset the device register and internal
state machines are place in a known state. However, the
contents of the CAM must still be initialized. The
minimum required initialization will set bit 0 of each
136-bit CAM word to 0 to indicate that the word does not
contain valid data. In addition, the table configuration bits
of the Instruction register must be initialized to indicate
the width of the each of the four addressable sections.
Harmony
CSO[1]
CSO[0]
6
4
3
2
1
0
CSI
Harmony
CSO[0]
3
2
1
0
6
5
4
CSI
CSI
Harmony
CSO[0]
The last CAM in the table bit (LCAM) of the Instruction
register must be set to 1 in the last device of the cascaded
block; the LCAM bit must be set to 0 in all other devices.
In addition, the last CAM on the SRAM bus (LRAM)
must be set to 1 in the last device of the cascaded block;
the LRAM must be set to 0 in all other devices.
3
2
CSI
1
0
6
5
CSI
4
Harmony
CSO[0]
3
2
1
0
6
5
CSI
4
CSI
Harmony
For single Harmony configurations the table size bit, bits 2
and 3, of the Instruction register should be set to 00 to
reduce the latency from five or six to four clock cycles.
Finally, the Mask register must be initialized to values that
depend upon the specific application.
CSO[1] CSO[0]
Figure 2: Depth Cascading of Eight Harmony
Devices
Note: The Mask registers are initialized to all 0s, which will
guarantee a match, when compared with any word, in the device
regardless of the data values contained. Latency also needs to be
initialized.
Arbitration
Four cycles after the SEARCH instruction, each device
drives CSO[1:0] with the match result of the search. At the
next cycle, all downstream devices know the outcome of
the search in all the upstream devices. If any of the
upstream devices has a match, all the subsequent devices
defer driving the SRAM bus. If a search or no match
occurs, Harmony with its LRAM bit set (the last in the
chain) drives the SRAM bus signals. Also, the device with
LCAM set to 1 is the default driver of the MV, MF, and
/MM signals.
Depth-Cascading
The Harmony search engine can depth-cascade up to eight
devices. Harmony performs all the necessary arbitration to
decide which device drives the SRAM bus, thereby
eliminating bus contention. The latency of the searches
increases as the table size increases; however, the search
rate remains constant.
Rev. 0 April 4, 2001
5
MUAD "Harmony" 1M and 2M Ternary CAMs
Functional Description
Search (68-bit Configuration with LCAM = 1)
The device is configured to be the last in the
depth-cascaded table by setting LCAM to 1 in the
Instruction register. The device with LCAM set to 1 drives
the MV, MF, and /MM signals in cycles when none of the
upstream devices drive these signals. Harmony with its
LCAM bit set drives MV, MF, and /MM during a search
with a no match or with non-search instructions.
Harmony Table Configuration
The table configuration (CFG) field of the Instruction
register allows the designer to configure and manage the
internal tables of the Harmony device, using bits 9 through
16. The Harmony (1M) is internally divided into four
pages consisting of 2048 x 136 bits, each of which may be
configured as 4096 x 68 bits, 2048 x 136 bits, or 1024 x
272 bits by setting the following bits:
Search (68-bit Configuration with LRAM = 1)
The device is configured to be the last on the SRAM bus by
setting LRAM to 1 in the Instruction register. In a cycle
where the upstream Harmony does not drive the SRAM
bus, the last device of the SRAM bus (with LRAM = 1)
drives the SRAM control signals (SADR, /CE, /WE, /ALE)
when they are active. When set to 1, the LRAM bit sets the
default driver for the SRAM control signals (SADR, /CE,
/WE, and /ALE).
•
•
•
00:4096 x 68 bits
01:2048 x 136 bits
10:1024 x 272 bits
Table 2: Table Configuration Bits
Bits
Function
[10:9]
These bits configure the address space within the
first quadrant.
[12:11]
[14:13]
[16:15]
These bits configure the address space within the
second quadrant.
Depth-Cascading to Generate Full
These bits configure the address space within the
third quadrant.
Bit 0 of each of the entries, regardless of width, is
designated as a special bit (1 = Full; 0 = Empty). For each
WRITE NEXT FREE ADDRESS or WRITE to the CAM
array, a device asserts FO[1] and FO[0] if it does not have
any empty locations.The Full Flag (FF) is asserted if the
device is full and all FI inputs are high.
These bits configure the address space within the
fourth quadrant.
For the Harmony 2M device, which has a similar
architecture, this same bit configuration will yield double
the amount of possible entries within the device. For
example, setting the bits in the CFG field to 00 would
yield a capacity of 8192 x 68 bits as opposed to 4096 x 68
bits.
Note: The BHI and BHO pins would be used when a group of
more than eight devices are cascaded; otherwise, these pins are
not connected or tied to ground.
Vcc
DQ[67:0]
6
5
5
5
5
5
4
4
4
4
3
2
2
1
1
0
FI
Multiple Search Table Configuration
Harmony
FO[1]
FO[0]
FF
FF
FF
FF
FF
FF
FF
There are a variety of ways to internally configure and
manage multiple search tables, with variable widths,
within the Harmony device. We will show these methods,
by example, each of which may be configured by the
designer. See Figures 4, 5, and 6.
Vcc
Vcc
Vcc
Vcc
Vcc
Vcc
6
3
FI
0
Harmony
FO[1]
FO[0]
6
3
FI
2
1
1
0
0
Harmony
FO[1]
FO[0]
The Harmony device CAM is divided into four quadrants.
Each quadrant may be configured individually to a width
of 68 bits, 136 bits, or 272 bits.
6
3
FI
2
Harmony
FO[1]
FO[0]
The Harmony device is fully capable of performing
one-cycle, successive search operations even when
configured for half-word widths of 68 bits or two cycle
search operations on double-word widths of 272 bits.
6
4
3
2
1
0
FI
FO[0]
Harmony
3
2
1
0
6
5
4
4
FI
FI
Harmony
FO[0]
3
2
FI
1
0
6
5
FI
Harmony
FO[0]
3
2
1
0
6
5
FI
4
FI
Harmony
FF
FO[1]
FO[0]
Figure 3: Full Generation in a Cascaded Table
6
April 4, 2001 Rev. 0
Functional Description
MUAD "Harmony" 1M and 2M Ternary CAMs
capacity. For example, if the CFG bits [16:9] are set to
10000101 then we will have configured the (2M) device to
support four different tables, of widths 2048 x 272 bits,
4096 x 136 bits, 8192 x 68 bits, and 4096 x 136 bits
respectively.
136
136
0
271
0
4
1
5
2
6
3
7
4 K
16380
16381
16382
16383
136
CFG = 10
(272-Bit Configuration)
4 K
Figure 4: 272-Bit (Double-Word) Configuration
Note: The WRITE NEXT FREE ADDRESS operation is not
136
4 K
supported in 272-bit mode.
136
135
0
68
8 K
0
2
4
6
1
3
5
7
272
8 K
2K
CFG = 10000101
Figure 7: Different Width Example
16382
16383
CFG = 01
(136-Bit Configuration)
Depth Cascading to Create Larger Logical Tables
Some high-performance applications require larger tables
than can be provided by one device. For these specific
applications, up to eight Harmony devices can be depth
cascaded to form larger table sizes without a loss of
throughput or any external glue logic.
Figure 5: 136-Bit (Single-Word) Configuration
68
67
0
0
1
2
3
If more than one Harmony is desired, the host ASIC must
set the TLSZ field of the Instruction register to 01 and the
LCAM (bit 7 in the Instruction register) to 1 in the last
Harmony of the depth-cascaded chain. The LCAM bit of all
previous devices of the depth-cascaded chain must then be
set to 0. Similarly, the last CAM on the SRAM bus
(LRAM) signal, bit 8 in the Configuration register, must be
set to 1 in the last Harmony of the depth-cascaded chain
connected to the SRAM bus. The LRAM bit of the other
Harmony devices must then be set to 0.
16 K
When the TLSZ field is set to one, the MF, /MM, and MV
signals will have five clock cycles of latency. When a single
Harmony is used, the TLSZ field of the Instruction register
can be set to 0 to reduce the latency from five or six to four
clock cycles. Harmony will perform all of the necessary
arbitration to decide which device will drive the SRAM
bus. Although the latency of the searches will increase
proportionally with the table sizes, the search rate will
remain constant. For each search, if the device determines a
match within the device, then the CSO[1] and CSO[0]
signals will be asserted. See Table 24 on page 19 for TLSZ
configurations.
16383
CFG = 00
(68-Bit Configuration)
Figure 6: 68-Bit (Half-Word) Configuration
Multiple Logical Tables of Different Widths
The logical tables in the Harmony device are configured as
equal width tables but some applications justify different
table widths. The Harmony device may be configured, by
quadrant, to support different width logical tables within the
same search engine as long as the total number of bits in all
combined tables does not exceed the device’s maximum
Rev. 0 April 4, 2001
7
MUAD "Harmony" 1M and 2M Ternary CAMs
Register Descriptions
REGISTER DESCRIPTIONS
Harmony contains sixteen Comparand registers, the
Global Mask register (consisting of nine registers), Result,
Instruction, Information, Burst Read, Burst Write, Next
Free Address, and Configuration registers. Table 3
provides an overview of the Harmony registers. The
registers are ordered in ascending address order. Each
register group is described in the following subsections.
Table 3: Register Overview
Address
0–31
32–47
48–55
56
Abbreviation
CMPR[0–31]
GMR[0-7]
RR[0–7]
INSTR
Type Name
16 136-bit Comparand Registers. Stores comparands from the DQ bus for writing later.
R/W Eight 136-bit Global Mask Registers
Eight Result Registers
R/W Instruction Register
Information Register
R
R
57
INFO
R
58
RBAR
R/W Burst Read Address Register
R/W Burst Write Address Register
59
WBAR
60
NFA
R
Next Free Address Register
61
CONFIG
N/A
R/W Configuration Register
62–63
N/A
Reserved
DQ[67:0]
I/O
Configuration
Register
Burst Read
Burst Write
Burst Length
Burst Length
Information
Register
Instruction
Register
MUX
MUX
Comparand
Register 32x
Mask Reg 16x
MUX
MUX
Word
Word
Word
Word
Mask In
Data In
Mask In
Data Out
Mask Out
Burst Write
Address
CAM
Core
Burst Read
Address
Address
Register
Address In
MUX
Address Out
MUX
MUX
Result
Register 8x
NFA
Register
MUX
Figure 8: Internal Datapath Diagram of the Harmony Registers
8
April 4, 2001 Rev. 0
Register Descriptions
MUAD "Harmony" 1M and 2M Ternary CAMs
Comparand Registers
The device contains eight 136-bit Comparand registers
dynamically selected in every SEARCH operation to store
the comparand presented on the DQ bus. These registers
will later be used by the WRITE NEXT FREE
ADDRESS.
In Cycle A of the SEARCH instruction, Harmony stores
the SEARCH data bits[135:68]) in the even-number
Comparand register. In Cycle B, Harmony stores the
SEARCH data bits[67:0] in the odd-numbered Comparand
register. See Figure 9.
136
Index
135
67
0
0
1
0
2
4
6
1
3
5
7
15
30
31
Figure 9: Comparand Registers Address and Usage
Global Mask Registers
The device contains eight 136-bit Global Mask registers
dynamically selected in every SEARCH operation to
select the search subfield. Figure 10 specifies the address
of these registers. The 3-bit global search or write index
supplied on the Instruction bus applies eight global masks
during the SEARCH and WRITE operations, as shown in
Figure 10.
A mask bit in the Global Mask registers is used during
SEARCH and WRITE operations. In SEARCH
operations, setting the mask bit to 1 enables compares;
setting the mask bit to 0 disables compares (forced match)
at the current bit position. In WRITE operations to the data
or mask array, setting the mask bit to 1 enables writes;
setting the mask bit to 0 disables writes at the current bit
position.
Note: In a 68-bit configuration, the host must program the even
and odd mask register with the same value; Harmony uses
even-numbered mask registers as global masks.
During a SEARCH operation, the search data bit (S), data
bit (D), mask bit (M) and the global mask bit (G) are used
in the following manner to generate a match at that bit
position (see Table 4).
136
Index
135
0
0
1
2
3
4
5
6
7
0
2
4
6
8
10
12
14
1
3
5
7
9
11
13
15
Table 4: Bit Position Map
G
0
1
1
1
1
1
M
x
S
x
D
x
Match
1
1
1
0
0
1
0
1
1
1
1
x
x
0
0
1
1
0
1
0
1
SEARCH and WRITE Command Global Mask Selection
Figure 10: Addressing the Global Mask Register
Rev. 0 April 4, 2001
9
MUAD "Harmony" 1M and 2M Ternary CAMs
Register Descriptions
Result Registers
The device contains eight Result registers to hold the
lowest match address (LMA) found during a SEARCH
operation. The SEARCH instruction specifies which
Result register to use by parsing the Result register index
in Cycle B of the SEARCH instruction.
Subsequently, the host uses this register to access the mask
of the word at the LMA or external SRAM using the index
as part of the address (see SRAM Addressing on page 4).
The device with a valid bit set performs a READ or
WRITE operation. All other devices suppress the
operation.
Table 5: Result Register (1M)
Bit(s) Name
Initial Description
Value
[13:0] INDEX
x
Index. This is the address of the 68-bit entry where a successful search occurs. The device updates this
field if it has a successful search. In 136-bit, the LSB is 0; in a 272-bit configuration, the two LSBs are 00.
[30:14]
[31]
N/A
0
0
Reserved
VALID
Valid. The device sets this bit to 1 if it is a global winner (first Harmony downstream with a hit) in a SEARCH
operation.
[67:32]
N/A
0
Reserved
Table 6: Result Register (2M)
Bit(s) Name
Initial Description
Value
[14:0] INDEX
x
Index. This is the address of the 68-bit entry where a successful search occurs. The device updates this
field if it has a successful search. In 136-bit, the LSB is 0; in a 272-bit configuration, the two LSBs are 00.
[30:15]
[31]
N/A
0
0
Reserved
VALID
Valid. The device sets this bit to 1 if it is a global winner (first Harmony downstream with a hit) in a SEARCH
operation.
[67:32]
N/A
0
Reserved
Instruction Register
Table 7: Instruction Register
Bit(s) Name
Initial Description
Value
[0]
[1]
SRST
DEVE
0
Software Reset. If 1, this bit resets the device with the same effect as a hardware reset. Internally, it
generates a reset pulse lasting for eight CLK cycles. This bit automatically resets to a 0 if written with a 1.
Device Enable. If 0, the device does not perform any SEARCH, WRITE, WRITE NEXT FREE ADDRESS,
and READ operations, and it keeps the SRAM bus in 3-state condition, forcing the cascade interface outputs
to 0.
0
[3:2]
TLSZ
01
Table Size. The host must program this field to configure the chips into a table of a certain size. This field
affects the pipeline latency of the SEARCH and WRITE NEXT FREE ADDRESS operations as well as the
accesses to the SRAM (SADR[21:0], /CE, /OE, /WE, /ALE, MV, MF, /MM, and /ACK). Once programmed,
the search latency stays constant.
Number of Devices
00:1 device
01:2-4 devices
10:5-8 devices
11:Reserved
CLK
4
5
6
[6:4]
RLAT
000
Latency of Hit Signals. This field adds latency to the MF, MV, /MM, and /ACK signals by the following
number of CLK cycles during SEARCH and SRAM READ operations.
000:0
001:1
010:2
011:3
100:4
101:5
110:6
111:7
[7]
LCAM
0
Last CAM in Table. This device is the last CAM in the depth-cascaded table. In the event of a search failure,
the device with this bit set drives the hit signals as follows.
MF = 0, MV = 1.
During non-search cycles, the device with this bit set drives the signals as follows.
MF = 0, MV = 0.
10
April 4, 2001 Rev. 0
Register Descriptions
MUAD "Harmony" 1M and 2M Ternary CAMs
Table 7: Instruction Register (continued)
Bit(s) Name
Initial Description
Value
[8]
LRAM
0
Last CAM on this SRAM Bus. This device is the last CAM on this SRAM bus. In cycles where a CAM does
not drive the SRAM bus, the device with this bit set drives the SRAM bus (SADR, /CE, and /WE) in their
inactive state. This bit sets a default driver for the SRAM control signals (SADR, /CE, /WE, and /OE).
Note: /OE is always asserted or deasserted.
[16:9]
CFG 10000101 Table Configuration. The device is internally divided into four quadrants of 8K x 68, each of which can be
configured as 8K x 68, 4K x 136, or 2K x 272 as follows.
00:8K x 68
01:4K x 136
10:2K x 272
11:Reserved
Bits[10:9] apply to configuring the 1st quadrant in the address space.
Bits[12:11] apply to configuring the 2nd quadrant in the address space.
Bits[14:13] apply to configuring the 3rd quadrant in the address space.
Bits[16:15] apply to configuring the 4th quadrant in the address space.
[67:17]
N/A
0
Reserved
Information Register
Table 8: Information Register
Bit(s) Name
Initial
Value
Description
[11:0] MFD
000100110011
Manufacturer ID. These bits include the JTAG bit LSB = 1.
[27:12] DEVID 1010110100000010 Device Identification. These bits indicate the device identification number.
[31:28] RVSN
0001
Revision Number. This is the current device revision number. This number increments by 1 for
each revision of the device.
[67:32] N/A
Reserved
Burst Read Address Register
The Burst Read Address register fields must be programmed before each BURST READ operation.
Table 9: Burst Read Address Register
Bit(s) Name
Initial Description
Value
[13:0] ADDR
0
Address. This is the starting address of the location of the CAM array during a BURST READ operation. It
automatically increments by 1 for each successive READ of the CAM array.
[18:14]
N/A
0
0
Reserved
[27:19] BLEN
Length of Burst Access. The device can READ from 4 up to 511 locations in a burst mode. Up to the
maximum number of devices, this decrements to back 0.
[67:28]
N/A
0
Reserved
Burst Write Address Register
The Burst Write Address register fields must be programmed before BURST WRITE operations.
Table 10: Burst Write Address Register
Bit(s) Name
Initial Description
Value
[13:0] AADR
0
Address. This is the starting address of the location of the CAM array during a BURST WRITE operation. It
automatically increments by 1 for each successive WRITE of the CAM array.
[18:14]
N/A
0
0
Reserved.
[27:19] BLEN
Length of Burst Access. The device provides the capability to WRITE from 4 up to 511 locations in a burst
mode. Up to the maximum number of devices, this decrements to back 0.
Reserved.
[67:28]
N/A
0
Rev. 0 April 4, 2001
11
MUAD "Harmony" 1M and 2M Ternary CAMs
Register Descriptions
Next Free Address Register
Bit 0 of each, regardless of width, entry is a special bit
designated for use in the operation of the WRITE NEXT
FREE ADDRESS instruction. In 68-bit configurations, bit
0 indicates whether a location is full (bit set to 1) or empty
( bit set to 0). Every WRITE and WRITE NEXT FREE
ADDRESS instruction loads the address of first 68-bit
location that contains a 0 in the entry’s bit 0. This is stored
in the Next Free Address register. If the bits of the LSB in
a device are set to 1, Harmony asserts FO[1:0] to 11. FF
should also be set to 1 in each data word.
In 136-bit configuration, the (LSB) of this register is
always set to 0. Regardless of the configured word width,
the host must set bit 0 of each word to 0 (empty) or 1 (full)
to indicate the full/empty status of each entry.
Configuration Register
Table 11: Configuration Register
Bit(s) Name
Initial Description
Value
[0]
SENE
0
Enable Search Enable. When this bit is set to 1, /SEN[3:0] are enabled. These pins are individual quadrant
enables. When SENE is set to 0, the /SEN[3:0] pins have no effect.
Multi-Match Enable. When this bit is set to 1, the /MM output is enabled. Else, the /MM is disabled (3-state).
SCLK Enable. When this bit is set to 1, the SCLK output is enabled. Else, the SCLK output is disabled
(3-state).
[1]
[2]
MME
SCE
0
[67:3]
N/A
0
Reserved.
12
April 4, 2001 Rev. 0
Harmony Instructions
MUAD "Harmony" 1M and 2M Ternary CAMs
HARMONY INSTRUCTIONS
A master device, such as a controller, issues instructions to
Harmony using the Op Code Valid (OPV) signal and the
Instruction bus. The following subsections describe the
functions of the instructions.
Instruction Codes
Harmony implements four basic instructions (see Table 12).
OP[1:0] apply the instructions to the device while keeping
the command valid (OPV) signal high for two CLK cycles.
These two CLK cycles are designated as Cycle A and Cycle
B.
The OP[8:2] field passes the parameters of the instruction in
Cycles A and B. The controller must align the instructions
with the CLK signal.
Table 12: Harmony Instructions
Code Instruction Description
00
01
10
READ
Reads one of the following: CAM Array, Register locations or external SRAM.
Writes one of the following: CAM Array, Register locations or external SRAM.
WRITE
SEARCH Searches the CAM array for a desired pattern using the specified register from the Global Mask register array and
local mask associated with each data cell.
11
WRITE
NEXT
The device can write up to 16 comparands for internal storage. The device’s controller inserts these entries at the
next free address (as specified by the NFA register) using the WRITE NEXT FREE ADDRESS instruction.
FREE
ADDRESS
Instructions and Instruction Parameters
Table 13 lists the Instruction bus fields that contain
Harmony instruction parameters and their respective
cycles. Each instruction is described separately in
subsections following this table.
Table 13: Instruction Parameters
Instruction Cyc
8
7
6
5
4
3
2
1
0
SADR[21]1 SADR[20]1 SADR[19]1
READ
A
B
A
B
A
0
0
0
0 = Single
1 = Burst
0
0
0
0
0
0
0
0
0 = Single
1 = Burst
0
0
0
1
0
1
1
0
SADR[21]1 SADR[20]1 SADR[19]1
WRITE
Global Mask Register Index
Global Mask Register Index
0 = Single
1 = Burst
0
0
0
0 = Single
1 = Burst
SEARCH
SADR[21]
SADR[20]
SADR[19] Global Mask Register Index
68-bit or 136-bit: 0
272-bit:
1 in 1st Cycle
0 in 2nd Cycle
B
A
B
Result Register Index[2:0]
Comparand Register Index
1
1
1
0
1
1
WRITE
NEXT
FREE
SADR[21]
0
SADR[20]
0
SADR[19] Comparand Register Index
Mode
0:68-bit
1:136-bit
Comparand Register Index
ADDRESS2
Notes:
1.
2.
For SRAM read/write only.
The WRITE NEXT FREE ADDRESS instruction is not supported when the table width is 272 bits.
Rev. 0 April 4, 2001
13
MUAD "Harmony" 1M and 2M Ternary CAMs
Harmony Instructions
BURST READ Instruction
READ Instruction
The burst length (BLEN) field of the Burst Read Address
(RBAR) register determines the latency of the BURST
READ instruction. The BURST READ instruction
completes in four clock cycles plus twice the number of
burst accesses. Note that, before initiating the BURST
READ instruction, the host must first program the BURST
READ Address register with the start address and the
length of transfer. The following sequence delineates the
clock cycles required of the BURST READ instruction:
The READ instruction, configured as a SINGLE READ
(OP[2] = 0) or as a BURST READ (OP[2] = 1), will read
the CAM array, synchronous random access memory
(SRAM), or register location. The SINGLE READ
instruction operates in six clock cycles. However, the
BURST READ requires two additional clock cycles for
each successive READ instruction. Refer to Tables 14, 15,
16, and 17 for the READ address formats.
SINGLE READ Instruction
In the first cycle, the host ASIC configures the OP[1:0]
(OP[2] = 1), using OPV = 1 and applies the READ
instruction, while the DQ bus supplies the address. The
host will then select the device for which UID[4:0]
matches the DQ[25:21] lines, or the last chained device
when DQ[25:21] = 11111. The host will also supply
SADR[21:19] on OP[8:6] in first cycle of the BURST
READ instruction if the READ instruction has been
applied to an external SRAM.
During the first cycle, the host ASIC configures the
OP[1:0] (OP[2] = 0), using OPV = 1 and applies the
READ instruction, while the DQ bus supplies the address.
The host selects the device for which UID[4:0] matches
the DQ[25:21] lines, or the last chained Harmony of the
cascade when DQ[25:21] = 11111. The host ASIC also
will supply SADR[21:19] on OP[8:6] in the first cycle of
the READ instruction, if the READ has been applied to an
external SRAM.
For the next two cycles the host will 3-state (HIGHZ)
DQ[67:0]. In the fourth cycle, the device selected by the
host will drive DQ[67:0] to signal the end of transfer, and
deassert /ACK from Z to low. In the fifth cycle, the
selected device (selected by the host) will drive the data to
be read from the addressed location on DQ[67:0] and
assert /ACK signal back to high. These fourth and fifth
cycles are repeated until all of the specified accesses in the
burst length (BLEN) field of the Burst Read Address
register have been depleted.
For the next two cycles the host ASIC will hold the
DQ[67:0] bus in a 3-stated, high Z mode. Afterwards, in
the fourth cycle, the device selected by the host will drive
the DQ[67:0] bus and pull the /ACK signal from Z to low.
In the fifth cycle, the device selected by the host will drive
data to be read from the addressed location on DQ[67:0]
and assert the /ACK signal to high.
Lastly, the selected device 3-states DQ[67:0] and deasserts
/ACK to low. Upon the termination of the last cycle, the
selected device 3-states the /ACK, completes the SINGLE
READ instruction, and prepares Harmony for the next
instruction.
On the last transfer, the selected device drives DQ[67:0] to
a 3-stated position, asserts the end of transfer (EOT) signal
to high and deasserts /ACK to low. Upon the termination
of the last cycle, cycle 4 + 2n (where n is the number of
burst accesses), the selected device 3-states the /ACK
signal, completes the BURST READ instruction, and
prepares Harmony for the next instruction.
Table 14: Read Instruction Parameters
Instruction
Parameter
OP[2]
Read Instruction Description
0
SINGLE READ
BURST READ
Reads a single location of the CAM array, external SRAM, or device registers. All access
information is applied on the DQ bus.
1
Reads a block of locations from the CAM array as a burst. The internal register (RBAR) specifies
the starting address and the length of the data transfer from the CAM array, and it
auto-increments the address for each access. All other access information is applied on the DQ
bus.
Note: The device registers and external SRAM can only be read in single-read mode.
14
April 4, 2001 Rev. 0
Harmony Instructions
MUAD "Harmony" 1M and 2M Ternary CAMs
Table 15: CAM or SRAM Read Address Format
DQ
DQ
DQ
DQ
DQ
DQ
DQ
DQ
[67:30]
[29]
[28:26]
[25:21] [20]
[19]
[18:14]
[13:0]
Reserved 0:Direct
Result Register Index
UID
UID
0
1
0:Data
1:Mask
Reserved If DQ[29] is 0, this field carries address of
data location. If DQ[29] is 1, the Result
register specified on DQ[28:26] supplies
the address of the data location.
1:Indirect (Applicable if DQ[29] is
indirect)
Reserved 0:Direct
Result Register Index
1:Indirect (Applicable if DQ[29] is
indirect)
0
Reserved If DQ[29] is 0, this field carries address of
SRAM’s location. If DQ[29] is 1, the Result
register specified on DQ[28:26] supplies
the address of the SRAM’s data location.
Note: DQ[25] should be set to 1.
Table 16: Internal Registers Read Address Format
DQ[67:26]
DQ[25:21]
DQ[20:19]
11:Register
DQ[18:6]
DQ[5:0]
Reserved
UID
Reserved
Register Address
Table 17: CAM Read Address for BURST READ
DQ[67:26] DQ[25:21] DQ[20]
Reserved UID
DQ[19]
DQ[18:14] DQ[13:0]
0
0:Data
1:Mask
Reserved Don’t Care. These 14 bits come from the internal register (Burst Read
Address register) which increments for each access.
BURST WRITE Instruction
WRITE Instruction
The BURST WRITE instruction operates for the number of
burst accesses specified by the burst length (BLEN) field of
the Burst Write Address register plus two additional clock
cycles. Note that, before initiating the BURST WRITE
instruction, the host must program the Burst Write Address
register with the start address and the length of transfer as
indicated in the BLEN field. The following summarizes the
sequence of the BURST WRITE instruction
The WRITE instruction can be configured for SINGLE
WRITE (OP[2] = 0) or BURST WRITE (OP[2] = 1)
instruction of a CAM array, register location, or external
SRAM locations or using the internal auto-incrementing
Burst Write Address register, of the CAM array locations.
The SINGLE WRITE instruction can be completed in only
three-cycles. However, the BURST WRITE operation
requires an additional cycle for each successive WRITE.
In the first cycle, the host applies the WRITE instruction on
the OP[1:0] (OP[2] = 1), using OPV = 1 and supplied
address on the DQ bus. The host also supplies the index to
the Global Mask register to mask the WRITE to the CAM
array locations in OP[5:3]. The host will then select the
device for which UID[4:0] match the DQ[25:21], or all
devices when DQ[25:21] = 11111.
SINGLE WRITE Instruction
In the first cycle, the host applies the WRITE instruction on
the OP[1:0] (OP[2] = 0), using OPV=1 and the supplied
address on the DQ bus. The host will also supply the index
to the Global Mask register to mask the WRITE instruction
to the CAM array’s location in OP[5:3]. For the WRITE
instruction, the host selects the device for which UID[4:0]
match DQ[25:21] or all connected devices when DQ[25:21]
= 11111.
In the second cycle, the host drives DQ[67:0] with the data
to be written to the CAM array location of the selected
device. On DQ[67:0], the host will only write the data to the
corresponding subfield that has its mask bit set to 1 in the
Global Mask register. This is specified by the index
OP[5:3] and supplied in the first cycle.
In the second cycle, the host drives DQ[67:0] with the data
to be written to the CAM array, external SRAM, or register
location of the selected device. The third cycle is an idle
cycle, and soon after this idle period the device is ready for
the next instruction.
From the third cycle to number of burst accesses (indicated
th
by the BLEN field) plus one additional access, n cycle +1,
the host drives DQ[67:0] with the data to be written to the
CAM array’s next location (addressed by the
auto-increment address field of the Burst Write Address
register).
Rev. 0 April 4, 2001
15
MUAD "Harmony" 1M and 2M Ternary CAMs
Harmony Instructions
This is specified by OP[5:3] index and supplied by the
first cycle. The host drives the end of transfer signal to low
th
Two cycles after the n cycle, the host will drive the end
of transfer signal to low. Afterward, when the cycle
terminates, the host drives the end of transfer signal to a Z
state, and prepares Harmony for the next instruction.
th
from the third to the n cycle. Afterward, the host drives
th
this same signal to high one cycle after the n cycle. This
value, n, is specified by the BLEN field of the Burst Write
Address register.
Table 18: CAM or SRAM (SINGLE WRITE) Write Address Format for 1M Harmony
DQ
DQ
DQ
DQ
DQ
DQ
DQ
DQ
[67:30]
[29]
[28:26]
[25:21] [20]
[19]
[18:14]
[13:0]
Reserved 0:Direct Result Register Index
1:Indirect (Applicable if DQ[29] is
indirect)
UID
UID
0
0:Data
1:Mask
Reserved If DQ[29] is 0, this field carries the address of
the data location.
If DQ[29] is 1, the Result register specified by
DQ[28:26] supplies the address of the data
location.
Reserved 0:Direct Result Register Index
1:Indirect (Applicable if DQ[29] is
indirect)
1
0
Reserved If DQ[29] is 0, this field carries address of the
SRAM’s location.
If DQ[29] is 1, the successful search register
specified by DQ[28:26] supplies the address
of the SRAM’s location.
Table 19: CAM or SRAM (SINGLE WRITE) Write Address Format for 2M Harmony
DQ
DQ
DQ
DQ
DQ
DQ
DQ
DQ
[67:30]
[29]
[28:26]
[25:21] [20]
[19]
[18:15]
[14:0]
Reserved 0:Direct Result Register Index
1:Indirect (Applicable if DQ[29] is
indirect)
UID
UID
0
0:Data
1:Mask
Reserved If DQ[29] is 0, this field carries the address of
the data location.
If DQ[29] is 1, the Result register specified by
DQ[28:26] supplies the address of the data
location.
Reserved 0:Direct Result Register Index
1:Indirect (Applicable if DQ[29] is
indirect)
1
0
Reserved If DQ[29] is 0, this field carries address of the
SRAM’s location.
If DQ[29] is 1, the Result register specified by
DQ[28:26] supplies the address of the SRAM’s
location.
Table 20: Internal Registers Write Address Format
DQ[67:26]
DQ[25:21]
DQ[20:19]
DQ[18:6]
DQ[5:0]
Reserved
UID
11:Register
Reserved
Register Address
Table 21: CAM (BURST WRITE) Write Address Format
DQ
[67:26]
DQ
[25:21]
DQ
[20]
DQ
[19]
DQ
[18:14]
DQ
[13:0]
Reserved
UID
0
0:Data
1:Mask
Reserved Don’t care. These 14 bits come from the internal register (Burst Write
Address register), which increments with each access.
16
April 4, 2001 Rev. 0
Harmony Instructions
MUAD "Harmony" 1M and 2M Ternary CAMs
All SRAM interface signals, MV, MF, and /MM will shift
to the right for different values of TLSZ. Additionally, MV,
MF, and /MM shift to the right for different values of
RLAT. See Tables 24 and 25 for the TLSZ and RLAT shift
values.
SEARCH Instructions
Full Word Searches
In the first cycle of full word searches, the host will drive
the OPV high and apply the instruction on OP[8:0]. For
the SEARCH operation, OP[5:3] carries the index to the
Global Mask register, whereas OP[8:6] carries the address
to be matched on SADR[21:19]. DQ[67:0] will then
transport the data to compare with the CAM array’s
[135:68] field.
Note: In the 68-bit configuration, the host must supply the same
data on DQ[67:0] during the first and second cycles.
Double-Word Searches
The double word SEARCH instruction completes in four
clock cycles after an initial latency search of four clock
cycles; however, since the instruction is pipelined, searches
In the second cycle, the host drives the OPV high and
applies the instruction to OP[8:0], whereas OP[5:2]
transports the index to the Comparand registers and then
sends the full, 136-bit, word (which was presented during
the first and second cycles) to the DQ bus. The OP[8:6]
carries the index to the Result register to store the
matching index and the match valid flag, whereas
DQ[67:0] carries the data to be compared with CAM array
bits 0 through 67. The resultant of the SEARCH
instruction will then appear as a pipelined, SRAM READ
cycle.
can be performed every two cycles
.
In the first cycle, the host drives the OPV to high and
applies the instruction on OP[8:0]. In this cycle OP[2]
must be set to 1. OP[5:3] carries the Global Mask register
index to be applied to field [271:136] of the search data
whereas DQ[67:0] carries the data to be compared with the
CAM array’s field of [271:204].
In the second cycle, the host also drives the OPV to high
and applies the instruction on OP[8:0], whereas DQ[67:0]
carries the data to be compared with the CAM array’s field
of [203:136].
The pipelined SEARCH instruction completes in two
clock cycles. All SRAM interface signals, MV, MF, and
/MM shift to the right for different values of TLSZ.
Additionally, MV, MF, and /MM shift to the right for
different values of RLAT. See Tables 24 and 25 for the
TLSZ and RLAT shift values.
In the third cycle, the host continues to drive the OPV to
high and to apply the instruction on OP[8:0]. In this cycle
OP[2] must be set to 0. OP[5:3] carries the Global Mask
register index to be applied to field [135:0] of the search
data. OP[8:6] carries the address to be supplied on
SADR[21:19] if the device has a successful match. The
DQ[67:0] carries the data to be compared with the
[135:68] field of the CAM array.
Note: The word in use must have bit 0 set to one, whereas an
empty word must have bit 0 set to 0.
Half-Word Searches
In the first cycle of half-word searches, the host drives the
OPV high and applies the instruction to OP[8:0]. For the
SEARCH instruction, the OP[5:3] carries the index to the
Global Mask register, whereas OP[8:6] carries the address
to be matched on SADR[21:19]. The DQ[67:0] will then
transport the data to be compared with the CAM array’s
[67:0] field.
In the fourth cycle, the host continues to drive OPV high
and apply the instruction on OP[8:0]. The DQ[67:0]
carries the data to be compared against the [67:0] field of
the CAM array.
In the 272-bit configuration, the SEARCH instruction will
be completed in four clock cycles. The SEARCH
instruction results appear as a pipelined SRAM READ
cycle with its latency measured from the second cycle of
the instruction.
In the second cycle, the host drives the OPV high and
applies the instruction to the OP[8:0] field. The OP[5:2]
transports the index to the Comparand registers to be stored,
and then sends the half, 68-bit word (which was presented
during the first and second cycles) to the DQ bus. The
OP[8:6] will transport the index to the Result register to
store the match index and the match valid flag. The full
word SEARCH instruction completes in four clock cycles
after an initial latency search of four clock cycles; however,
since the instruction is pipelined, searches can be performed
every two cycles.
For all SRAM interface signals, MV and MF will shift to
the right for different values of TLSZ. Additionally, MV,
MF, and /MM shift to the right for different values of
RLAT. See Tables 24 and 25 for the TLSZ and RLAT shift
values.
Rev. 0 April 4, 2001
17
MUAD "Harmony" 1M and 2M Ternary CAMs
Harmony Instructions
Write Next Free Address
instruction only supports 68 or 136 words and does not
support multiple search tables of different widths in depth
cascaded configurations. In others words, all tables must be
single, equal width tables.
The WRITE NEXT FREE ADDRESS instruction can be
completed in two clock cycles; the following delineates
the instruction sequence.
In the first half of the first cycle, the host applies the
WRITE NEXT FREE ADDRESS instruction on OP[1:0]
and sets the instruction data valid to one (OPV = 1).
OP[5:2] specifies the index of the even and odd comparand
registers that will be written in the 136-bit configuration. In
the 68-bit configuration, the even numbered comparands
are specified by this index. OP[8:6] transports the bits to be
driven on SADR[21:19] during the SRAM WRITE. In the
second half of the first cycle, the host continues to drive
OPV to 1, OP[1:0] to 11, and OP[5:2] with the even and
odd comparand indexes. OP[6] equals 0 for a half-word
searches of the next free address, and equals one for full
word searches of the next free address.
SRAM Addressing
Index[13:0] (for 1M Harmony) and Index[14:0] (for 2M
Harmony) contains the address, of a half-word entries, that
results in a successful match; when configured, it is this
address that resides on the full and double-word page
boundaries, respectively.
SADR[13:0] (for 1M Harmony) and SADR[14:0] (for 2M
Harmony) contains the address supplied on the DQ bus
during device READ or WRITE accesses. See Tables 22
and 23 for the SRAM addressing.
SRAM READ or WRITE Accesses
SRAM READ
In the second cycle, the host will set the instruction data
valid signal to 0 (OPV = 0). After the completion of the
second cycle, the CAM is ready for the next instruction.
The search latency of the SRAM WRITE instruction is the
same as the search latency to the SRAM READ instruction;
it is measured from the second cycle of the WRITE NEXT
FREE ADDRESS instruction.
The SRAM READ enables and accesses associative data
contained in external SRAM. An SRAM READ instruction
completes in six cycles and the following delineates the
instruction sequence.
In the first cycle, the host applies the READ instruction on
OP[1:0], and sets the instruction data valid to one
(OPV = 1). The DQ bus then supplies the appropriate
address, sets DQ[20:19] to 10, and selects the SRAM
address. The host selects the device for which the UID[4:0]
matches DQ[25:21] and supplies SADR[21:19] to OP[8:6].
In the second and third cycles, the host 3-states DQ[67:0].
When the host applies the WRITE NEXT FREE
ADDRESS instruction specifying the appropriate
comparand register, Harmony writes the specified
comparand in the next free location in the depth-cascaded
table. The next free location is the first entry in a Harmony
with its bit [0] set to 0. If all the entries within the first
Harmony are occupied (bit[0] = 1), then the first entry with
bit[0] = 0 in a downstream Harmony in a group of cascaded
devices is the next free location. In 136-bit configuration,
bit[0] of both the even and the odd locations are both 0
when empty or both 1 when filled.
In the fourth cycle, the selected device starts to drive
DQ[67:0] and drives the acknowledge signal, /ACK, from
HIGHZ to low. In the fifth cycle, the selected device drives
the READ address on SADR[21:0]; it also drives /ACK
high, /CE low, and /ALE low. In the sixth cycle, the
selected device 3-states /CE, SADR, and the DQ bus and
continues to drive /ACK low. At the end of sixth cycle, the
selected device 3-states /ACK.
When configured for depth cascading, the FF signal
indicates to the host when no more entries can be written,
and when all entries within a group of cascaded devices are
occupied. Harmony updates the signal to the CAM array
after each WRITE or WRITE NEXT FREE ADDRESS
instruction.
SADR[13:0] contains the address supplied on the DQ bus
during access to Harmony. Furthermore, OP[8:6] transports
signals from the instruction bus to the SRAM[21:19]
address bus.
SRAM WRITE
When configured for full words, the WRITE NEXT FREE
ADDRESS instruction writes to both the even and odd
Comparand registers in the data locations and uses the Next
Free Address register as the address. It generates a WRITE
to the external SRAM and also uses the Next Free Address
register as a portion of the SRAM address.
The SRAM WRITE instruction enables and writes
associative data contained in external SRAM. An SRAM
WRITE instruction completes in three clock cycles on the
DQ bus.
In the first cycle, the host ASIC applies the WRITE
instruction on CMD[1:0] (with CMD[2] = 0), and sets the
command valid signal to one (CMDV = 1). The DQ bus
then supplies the SRAM address, sets DQ[20:19] to 10. The
host ASIC selects the device for which the UID[4:0]
matches DQ[25:21]; it selects the device with LCAM bit set
when DQ[25:21] = 11111. In the second and third are
necessary wait cycles.
When configured for half-words, the WRITE NEXT FREE
ADDRESS instruction writes only to the even comparand
register within the data location and uses the NFA register
as the address. It generates a WRITE to the external SRAM
and also uses the Next Free Address register as a portion of
the SRAM address. The WRITE NEXT FREE ADDRESS
18
April 4, 2001 Rev. 0
Harmony Instructions
MUAD "Harmony" 1M and 2M Ternary CAMs
Table 22: SRAM Addressing for 1M Harmony
Instruction
SRAM Operation
Read
21
C8
C8
C8
C8
C8
20
C7
C7
C7
C7
C7
19
C6
C6
C6
C6
C6
18
1
[17:14]
UID[3:0]
UID[3:0]
UID[3:0]
UID[3:0]
UID[3:0]
[13:0]
SEARCH
Index[13:0]
NFA[13:0]
SADR[13:0]
SADR[13:0]
RR[13:0]
WRITE NEXT FREE ADDRESS
READ
Write
1
Read
1
WRITE
Write
1
Indirect Access
Write/Read
1
Table 23: SRAM Addressing for 2M Harmony
Instruction
SRAM Operation
Read
21
C8
C8
C8
C8
C8
20
C7
C7
C7
C7
C7
19
1
[18:15]
[14:0]
SEARCH
UID[3:0]
UID[3:0]
UID[3:0]
UID[3:0]
UID[3:0]
Index[14:0]
NFA[14:0]
SADR[14:0]
SADR[14:0]
RR[14:0]
WRITE NEXT FREE ADDRESS
READ
Write
1
Read
1
WRITE
Write
1
Indirect Access
Write/Read
1
Table 24: Right-Shift of Signals for TLSZ Values
TLSZ
00
Number of CLK Cycles
Number of Devices
0
1
2
1
01
2 - 8
5 - 8
10
Table 25: Right-Shift of Signals for RLAT Values
RLAT
000
001
010
011
100
101
110
111
Number of CLK Cycles
0
1
2
3
4
5
6
7
Rev. 0 April 4, 2001
19
MUAD "Harmony" 1M and 2M Ternary CAMs
Application Information
APPLICATION INFORMATION
the designer must decide between one-cycle searches,
which are not fully associative, or two-cycle searches that
are fully associative. Where the above approach is
unacceptable, or when IPv4 using CIDR addresses, two
alternatives are available that provide fully associative
searches. The first alternative simply stores single 34-bit
entries into each 68-bit half-word. This ensures that
searches are completed in one-cycle at the expense of less
efficient memory utilization. The second alternative
performs two linear search operations: one on the high
order and the other on the low order 34-bits. This second
approach provides more efficient memory utilization at the
expense of reduced search speeds.
34-Bit Word Applications
Some applications (e.g. IPv4, IPv4 CIDR, MPLS and
ATM entries) will warrant smaller content addressable
widths, such as 34-bit one-quarter words. For IPv4
(non-CIDR), MPLS and ATM addresses, a one-cycle
technique that is not fully associative can be used. To
ensure that table look-ups are completed in one-cycle, the
designer must determine which of the entries are stored in
the high quarter-word (bits 34 through 67) or low
quarter-word (bits 0 through 33) based upon a single bit of
the value to be stored (e.g., bit 0 of the network address).
Based upon the value of the selected address bit to be
found, while performing search operations, the Global
Mask register can be used to restrict the search to the high
or low quadrant of the 68-bit half-word.
Hence, 32-bit data must be entered in two iterations,
masking out the bits of the right side then the left side of
the device. Since the search operations must be performed
twice, thus, decreasing the speed to 50 million searches
per second instead of the usual 100 million searches per
second for the 68-bit and 136-bit configurations.
Furthermore, in the case where a match may be produced
on both halves of Harmony, the desired information of the
left half has the higher priority. For example, if a
successful match is found within the left half then the right
half will not be searched. Hence, information must be
written on the left half first then the right half.
Hence, the rationale is to perform two search operations of
Harmony’s content addressable array, where the first
search will be performed on the bits 0 through 33 with the
Mask register configured as zeroes, or "don’t cares" in bits
34 through 67. The second search will be performed upon
bits 34 through 67 while bits 0 through 33 will be masked
out using the Global Mask register.
Harmony can perform both 68-bit and 136-bit fully
associative searches in a single clock cycle. However, for
34-bit searches (with two 34-bit entries per 68-bit word),
20
April 4, 2001 Rev. 0
Timing Diagrams
MUAD "Harmony" 1M and 2M Ternary CAMs
TIMING DIAGRAMS
Single Location Read Timing Diagram
Cycle 1
Cycle 2
Cycle 3
Cycle 4
Cycle 5
Cycle 6
CLK
PHASE
OPV
READ
OP[8:0]
DQ
Address
X
Data
/ACK
Figure 11: Single Location Read Timing Diagram
Write Cycle Timing Diagram
Cycle 0
Cycle 1
Cycle 2
Cycle 3
Cycle 4
CLK
PHASE
OPV
Write
OP[8:0]
DQ
Address
Data
X
Figure 12: Write Cycle Timing Diagram
Rev. 0 April 4, 2001
21
MUAD "Harmony" 1M and 2M Ternary CAMs
Timing Diagrams
Data and Mask READ (BLEN = 4) for a Group of Cascaded Devices Timing Diagram
Cycle 1
Cycle 2
Cycle 3
Cycle 4
Cycle 5
Cycle 6
Cycle 7
Cycle 8
Cycle 9
Cycle 10 Cycle 11 Cycle 12
CLK
PHASE
OPV
READ
OP[8:0]
DQ
Address
FF
Data0
FF
Data1
FF
Data2
FF
Data3
/ACK
EOT
Figure 13: Data and Mask READ (BLEN = 4) for a Group of Cascaded Devices Timing Diagram
Data and Mask WRITE (BLEN = 4) for a Group of Cascaded Devices Timing Diagram
Cycle 1
Cycle 2
Cycle 3
Cycle 4
Cycle 5
Cycle 6
CLK
PHASE
OPV
Write
OP[8:0]
DQ
Address
Data0
Data1
Data2
Data3
X
EOT
Figure 14: Data and Mask WRITE (BLEN = 4) for a Group of Cascaded Devices Timing Diagram
22
April 4, 2001 Rev. 0
Timing Diagrams
MUAD "Harmony" 1M and 2M Ternary CAMs
68-Bit SEARCH Operation Timing Diagram
Cycle 1
Cycle 2
Cycle 3
Cycle 4
Cycle 5
Cycle 6
Cycle 7
Cycle 8
Cycle 9
Cycle 10
CLK
PHASE
SCLK
OPV
Search1
Search2
Search3
Search4
OP[1:0]
OP[8:2]
DQ
A
B
D1
D2
D3
D4
/SEN[3:0]
SADR[21:0]
/CE
A1
A2
A3
A4
/WE
/OE
MF
/MM
Hit
Miss
Hit
Miss
MV
CFG = 00000000, RLAT = 000, TLSZ = 00, LRAM = 0, LCAM = 0
Figure 15: 68-Bit SEARCH Operation Timing Diagram
Rev. 0 April 4, 2001
23
MUAD "Harmony" 1M and 2M Ternary CAMs
Timing Diagrams
136-Bit SEARCH Operation Timing Diagram
Cycle 1
Cycle 2
Cycle 3
Cycle 4
Cycle 5
Cycle 6
Cycle 7
Cycle 8
Cycle 9
Cycle 10
CLK
PHASE
SCLK
OPV
Search1
Search2
Search3
Search4
OP[1:0]
OP[8:2]
DQ
A
B
/SEN[3:0]
SADR[21:0]
/CE
A1
A2
A3
A4
/WE
/OE
MF
/MM
Hit
Miss
Hit
Miss
MV
CFG = 01010101, RLAT = 000, TLSZ = 00, LRAM = 0, LCAM = 0
Figure 16: 136-Bit SEARCH Operation Timing Diagram
24
April 4, 2001 Rev. 0
Timing Diagrams
MUAD "Harmony" 1M and 2M Ternary CAMs
272-Bit SEARCH Operation Timing Diagram
Cycle 1
Cycle 2
Cycle 3
Cycle 4
Cycle 5
Cycle 6
Cycle 7
Cycle 8
Cycle 9
Cycle 10
CLK
PHASE
SCLK
OPV
A
B
C
D
Search1
B
Search2
OP[1:0]
OP[8:2]
DQ
A
C
D
D0
D1
D2
D3
D0
D1
D2
D3
/SEN[3:0]
SADR[21:0]
/CE
A1
A2
/WE
/OE
MF
/MM
Hit
Miss
MV
CFG = 10101010, RLAT = 000, TLSZ = 00, LRAM = 0, LCAM = 0
Figure 17: 272-Bit SEARCH Operation Timing Diagram
Rev. 0 April 4, 2001
25
MUAD "Harmony" 1M and 2M Ternary CAMs
Timing Diagrams
Arbitration for a Group of Cascaded Devices Timing Diagram
Cycle 1
Cycle 2
Cycle 3
Cycle 4
Cycle 5
Cycle 6
Cycle 7
Cycle 8
Cycle 9
Cycle 10
CLK
PHASE
SCLK
OPV
Search1
Search2
Search3
Search4
OP[8:0]
DQ
D0
D1
D0
D1
D0 D1 D0
D1
A1
A2
A3
A4
SADR[21:0]
/CE
/WE
/OE
Hit
Miss
Hit
Hit
CSO[1:0]
MF
/MM
MV
RLAT = 000, TLSZ = 01, LRAM = 0, LCAM = 0
Figure 18: Arbitration for a Group of Cascaded Devices Timing Diagram
WRITE Timing Diagram
Cycle 1
Cycle 2
Cycle 3
Cycle 4
Cycle 5
Cycle 6
Cycle 7
Cycle 8
Cycle 9
Cycle 10
CLK
PHASE
SCLK
OPV
Write1
X
Write2
OP[1:0]
OP[8:2]
DQ
Comp1
Comp2
X
2
X
X
1A
1B
X
X
X
A1
A2
SADR[21:0]
/CE
/WE
/OE
TLSZ = 00, LRAM = 0
Figure 19: WRITE Timing Diagram
26
April 4, 2001 Rev. 0
Timing Diagrams
MUAD "Harmony" 1M and 2M Ternary CAMs
SRAM READ Cycle Timing Diagram
Cycle 1
Cycle 2
Cycle 3
Cycle 4
Cycle 5
Cycle 6
CLK
PHASE
SCLK
OPV
Read
OP[8:0]
DQ
Address
X
/CE
/WE
/OE
Address
SADR
/ACK
RLAT = 000, TLSZ = 00, LRAM = 0, LCAM = 0
Figure 20: SRAM READ Cycle Timing Diagram
Rev. 0 April 4, 2001
27
MUAD "Harmony" 1M and 2M Ternary CAMs
Timing Diagrams
SRAM WRITE Cycle TIming Diagram
Cycle 1
Cycle 2
Cycle 3
Cycle 4
Cycle 5
Cycle 6
Cycle 7
CLK
PHASE
SCLK
OPV
OP[8:0]
DQ
Address
X
X
/CE
/WE
/OE
A1
SADR
RLAT = 000, TLSZ = 00, LTAM = 0, LCAM = 0
Figure 21: SRAM WRITE Cycle Timing Diagram
28
April 4, 2001 Rev. 0
Timing Diagrams
MUAD "Harmony" 1M and 2M Ternary CAMs
AC Timing Waveforms
CLK
PHASE
T
IHCH
T
ISCH
Signal Group 1
Signal Group 2
Signal Group 3
T
ICHCH
T
ICSCH
T
CKHOV
T
CKHOV
T
CKHSHZ
Signal Group 4
Signal Group 5
T
CKHSV
T
CKHDZ
T
CKHSV
T
CKHDV
Signal Group 1: PHASE, OP[8:0], OPV, /RST
SIgnal Group 2: CSI[6:0], FI[6:0]
Signal Group 3: CSO[1:0], FO[1:0], FF, MF, /MM, MV, /SEN[3:0]
Signal Group 4: /CE, /OE, /WE, /ALE
Signal Group 5: DQ, /ACK, EOT
Figure 22: AC Timing Waveforms
Rev. 0 April 4, 2001
29
MUAD "Harmony" 1M and 2M Ternary CAMs
Electrical Specifications
ELECTRICAL SPECIFICATIONS
This section describes the electrical specifications,
capacitance, operating conditions, and DC characteristics
for the Harmony devices.
Electrical Characteristics
Symbol Parameter
Conditions
Min
Max
Unit
I
Input Leakage Current
0 ≤ V ≤ V
-10
10
µA
LI
IN DDQ
Output Leakage Current1
Output Low Voltage
I
0 ≤ V
≤ V
-10
2.4
10
µA
V
LO
OUT DDQ
V
8mA, V
= 3.3V
= 3.3V
0.4
OL
DDQ
DDQ
V
Output High Voltage
4mA, V
V
OH
1.8 V Supply Current2
3.3 V Supply Current2
I
TBD
TBD
mA
mA
DD
I
DDQ
1.
2.
Applies only for outputs in 3-state.
Average operating current at maximum frequency. Transient peak currents may exceed these values.
Capacitance
Symbol
Parameter
Max
Unit
pF1
pF2
C
Input Capacitance
Output Capacitance
6
IN
C
6
OUT
1.
2.
f = 1 MHz, V =0 V
IN
f = 1 MHz, V
OUT
=0 V
Operating Conditions
Symbol Parameter
Min
Max
Unit
V
Operating Voltage for IO
Operating Supply Voltage
3.0
1.65
2.0V
-0.3
0
3.6
V
DDQ
V
1.95
V
V
DD
Input High Voltage1
V
V
+0.3
IH
DDQ
Input Low Voltage2
V
0.8
70
V
IL
T
Ambient Operating Temperature
Supply Voltage Tolerance
°C
A
-10%
+10%
Notes:
1.
2.
Maximum allowable applies to overshoot only (V
Minimum allowable applies to undershoot only.
is 3.3 V supply).
DDQ
30
April 4, 2001 Rev. 0
Package
MUAD "Harmony" 1M and 2M Ternary CAMs
PACKAGE
Pin # 1 Corner
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20
A
B
C
A
B
C
D
D
E
F
E
F
G
H
J
G
H
J
K
L
K
L
F
E
M
N
P
R
T
M
N
P
R
T
U
V
W
Y
U
V
W
Y
D
e
A
L
L1
b
Seating Plane
272-Pin BGA Dimensions
Dim. A
Dim. b
Dim. D
Dim. E
Dim. e
Dim. F
Dim. L
0.50
L1
Min.
26.80
27.00
27.20
Nom.
Max.
1.17
0.04
24.00
1.27
24.13
0.60
30° TYP.
0.70
Rev. 0 April 4, 2001
31
MUAD "Harmony" 1M and 2M Ternary CAMs
Ordering
ORDERING
Part Number
Density
Clock Speed
Package
Temperature
Voltage
MUAD16K136-66B272C
MUAD16K136-83B272C
MUAD16K136-10B272C
66 MHz
83 MHz
100 MHz
272-Pin BGA
272-Pin BGA
272-Pin BGA
0 - 70°
0 - 70°
0 - 70°
1.8/3.3V
1.8/3.3V
1.8/3.3V
2Mbit
MUAD8K136-66B272C
MUAD8K136-83B272C
MUAD8K136-10B272C
66 MHz
83 MHz
100 MHz
272-Pin BGA
272-Pin BGA
272-Pin BGA
0 - 70°
0 - 70°
0 - 70°
1.8/3.3V
1.8/3.3V
1.8/3.3V
1Mbit
MUSIC Semiconductors reserves the right to make changes to its products and
specifications at any time in order to improve on performance, manufacturability or
reliability. Information furnished by MUSIC is believed to be accurate, but no
responsibility is assumed by MUSIC Semiconductors for the use of said information, nor
for any infringements of patents or of other third-party rights which may result from said
use. No license is granted by implication or otherwise under any patent or patent rights of
any MUSIC company.
MUSIC Semiconductors’ agent or distributor:
© Copyright 2001, MUSIC Semiconductors
Worldwide Headquarters
Asian Headquarters
European Headquarters
MUSIC Semiconductors
P. O. Box 184
MUSIC Semiconductors
1521 California Circle
Milpitas, CA 95035
MUSIC Semiconductors
Special Export Processing Zone
Carmelray Industrial Park
Canlubang, Calamba, Laguna
Philippines
6470 ED Eygelshoven
The Netherlands
USA
Tel: 408 869-4600
Tel: +31 43 455-2675
Fax: +31 43 455-1573
Fax: 408 942-0837
Tel: +63 49 549-1480
http://www.musicsemi.com
email: info@musicsemi.com
USA Only: 800 933-1550 Tech Support
888 226-6874 Product Info
Fax: +63 49 549-1024
Sales Tel/Fax: +632 723-6215
32
April 4, 2001 Rev. 0
相关型号:
SI9130DB
5- and 3.3-V Step-Down Synchronous ConvertersWarning: Undefined variable $rtag in /www/wwwroot/website_ic37/www.icpdf.com/pdf/pdf/index.php on line 217
-
VISHAY
SI9135LG-T1
SMBus Multi-Output Power-Supply ControllerWarning: Undefined variable $rtag in /www/wwwroot/website_ic37/www.icpdf.com/pdf/pdf/index.php on line 217
-
VISHAY
SI9135LG-T1-E3
SMBus Multi-Output Power-Supply ControllerWarning: Undefined variable $rtag in /www/wwwroot/website_ic37/www.icpdf.com/pdf/pdf/index.php on line 217
-
VISHAY
SI9135_11
SMBus Multi-Output Power-Supply ControllerWarning: Undefined variable $rtag in /www/wwwroot/website_ic37/www.icpdf.com/pdf/pdf/index.php on line 217
-
VISHAY
SI9136_11
Multi-Output Power-Supply ControllerWarning: Undefined variable $rtag in /www/wwwroot/website_ic37/www.icpdf.com/pdf/pdf/index.php on line 217
-
VISHAY
SI9130CG-T1-E3
Pin-Programmable Dual Controller - Portable PCsWarning: Undefined variable $rtag in /www/wwwroot/website_ic37/www.icpdf.com/pdf/pdf/index.php on line 217
-
VISHAY
SI9130LG-T1-E3
Pin-Programmable Dual Controller - Portable PCsWarning: Undefined variable $rtag in /www/wwwroot/website_ic37/www.icpdf.com/pdf/pdf/index.php on line 217
-
VISHAY
SI9130_11
Pin-Programmable Dual Controller - Portable PCsWarning: Undefined variable $rtag in /www/wwwroot/website_ic37/www.icpdf.com/pdf/pdf/index.php on line 217
-
VISHAY
SI9137
Multi-Output, Sequence Selectable Power-Supply Controller for Mobile ApplicationsWarning: Undefined variable $rtag in /www/wwwroot/website_ic37/www.icpdf.com/pdf/pdf/index.php on line 217
-
VISHAY
SI9137DB
Multi-Output, Sequence Selectable Power-Supply Controller for Mobile ApplicationsWarning: Undefined variable $rtag in /www/wwwroot/website_ic37/www.icpdf.com/pdf/pdf/index.php on line 217
-
VISHAY
SI9137LG
Multi-Output, Sequence Selectable Power-Supply Controller for Mobile ApplicationsWarning: Undefined variable $rtag in /www/wwwroot/website_ic37/www.icpdf.com/pdf/pdf/index.php on line 217
-
VISHAY
SI9122E
500-kHz Half-Bridge DC/DC Controller with Integrated Secondary Synchronous Rectification DriversWarning: Undefined variable $rtag in /www/wwwroot/website_ic37/www.icpdf.com/pdf/pdf/index.php on line 217
-
VISHAY
©2020 ICPDF网 联系我们和版权申明