Empowering Intelligent Optical Networks
H i g h l i g h t s
Product Brief
PRS Q-80G, PRS C48X, and PRS C192X
Switch Fabric
PB PRS Q-80G / v0.5 / 10-14-2004
Empowering Intelligent Networks
1
Advanced Technology for
Low Latency 80 Gbps to 320 Gbps
Platforms
The AMCC PRS Q-80G switch chip, along
with the PRS C48X and PRS C192X
switch fabric interface chips, is designed to
address the needs of a wide range of high
bandwidth, demanding applications that
handle any mix of real-time and best effort
traffic in the Metro Ethernet, WAN
multi-services and enterprise (data
centers, NAS/SAN, high-end LANs) space.
The PRS Q-80G shared-memory switch
builds upon the successful PRS Q-64G,
providing improved multicast capabilities,
reduced latency, enhancements to more
efficiently handle TDM traffic, and
higher-speed 3.2-Gbps serial links that
provide the overspeed required for
wire-speed performance. Using
0.13-micron technology, this solution
delivers 80 Gbps to 320 Gbps of user
bandwidth in a low-power, compact switch
fabric design, relieving the pressure of
space and power limitations facing
equipment manufacturers, network service
providers, and enterprise customers.
The PRS Q-80G enables faster time-to-
market for OEMs addressing the Quality of
Service (QoS) requirements of real-time
voice and video applications and the best
effort service needs of file transfer and
e-mail applications, while still leveraging
previous carrier investment by providing
the capability to switch IP and TDM traffic
concurrently on the same platform.
The PRS Q-80G is an integral part of a
complete system solution with AMCC's
integrated network processor and
full-featured traffic manager products
(nP3700), framers, PHYs, PowerPC
processors, and software tools. To facilitate
the evaluation and prototyping of complete
solutions, AMCC also offers a 320-Gbps
reference switch board featuring the
AMCC PRS Q-80G.
PRS Q-80G Switch, PRS C48X and
PRS C192X Fabric Interfaces
The PRS Q-80G device performs the
switch core function, in configurations of
two, four, or eight devices, depending on
the switch capacity needs. The PRS C48X
and PRS C192X devices perform the
queuing manager function of the
PRS Q-80G based switches. An adapter
card can be equipped with either a
PRS C48X interfacing a multi-GE / OC-48
network processor or a PRS C192X
interfacing up to four multi-GE / OC-48
network processors or a 10-Gbps / OC-192
network processor.
High Scalability and Wide Range
of Port Speeds for Complete and
Evolving Product Lines
The PRS Q-80G switch, when used in
conjunction with the PRS C48X and
PRS C192X fabric interface devices,
provides 80 Gbps (two chips), 160 Gbps
(four chips), or 320 Gbps (eight chips)
full-duplex aggregate user bandwidth with
a per-port speed-up factor of two, or eight
for PRS C48X-based ports, to match
wire-speed requirements for all possible
cases with all line protocols, fully enabling
system services such as security, deep
packet inspection, web caching, and SSL
in demanding Packet Over SONET (POS)
and Ethernet applications.
The PRS Q-80G enables the design of
single-stage switch fabrics with up to 32
ports for network processors that support
OC-192c, 10-Gbps, and multi-GE
applications, or up to 128 ports for network
processors that support OC-48c and
multi-GE applications, or any intermediate
mix of those ports.
The PRS Q-80G also has built-in 20-Gbps
and 40-Gbps clear channel attachment
capabilities, requiring a customer ASIC or
FPGA to interface with a 20-Gbps or
40-Gbps protocol engine.
Full Bandwidth Multicast and
Real Time Traffic for Platforms
Scaling from 80 Gbps to 320
Gbps
New generation, non-blocking
shared-memory switch fabric chipset
for high bandwidth demanding
applications such as Packet Over
SONET (POS) or Ethernet
Full bandwidth multicast to support
the growing demand for audio and
video distribution; the efficient bicast
capability enables line card
protection designs and port
snooping
Real-time traffic support for TDM
traffic and IP traffic switching on the
same platform
Up to 320 Gbps of full-duplex
aggregate user bandwidth with a per
port speed-up factor of two
Redundant switch plane operation,
with load sharing capability and
packet lossless scheduled
switchover
Backward compatibility of the
PRS Q-80G switch with the
PRS C192 fabric interface, and of
the PRS C48X and PRS C192X
fabric interface chips with the
PRS 64Gu and PRS Q-64G
switches, enabling optimized
development costs and reuse of
existing line cards and switch cards
Highly-scalable solutions from
5 Gbps to 320 Gbps, with the
PRS 5G, PRS 20G, PRS 80G, and
PRS Q-80G, preserving customer
investments in network processor
cards
Complete multi-GE / OC-48c
solutions with the AMCC nP3700
Integrated Network Processor /
Traffic Manager, framers, and PHYs
Validated solutions with third-party
network processor devices
Product Overview
Product Brief
PRS Q-80G, PRS C48X, and PRS C192X
Switch Fabric
PB PRS Q-80G / v0.5 / 10-14-2004
2
Empowering Intelligent Networks
Full-Bandwidth Multicast to
Support the Growing Demand for
Audio and Video Distribution
The PRS Q-80G multicast with QoS copes
with the challenges of multicast and
broadcast traffic, that is, group addressing,
scalability without duplication, and
isochronous (near real-time) traffic. It is an
increasingly valued capability for growing
internet applications, such as distribution
of audio and video material.
The PRS Q-80G, like the other PRS
shared-memory switches, provides a
built-in efficient multicast function. To
enable multicasting and maximize product
resources utilization, the PRS switch
executes a scheme of store once (single
cell copy in switch core memory) and
transmit multiple times (replication at
sending): the multicast cell is placed in
shared memory and the memory location
index is stored in the output queues
corresponding to the target output ports
and cell priority. The PRS multicast
mechanism allows cells to be easily
processed, with cells being scheduled at
different times for different output ports.
Multicast traffic has its own flow control,
isolated from the unicast traffic.
In addition, the full multicast addressing
capability of the PRS Q-80G enables full
bandwidth multicast, including broadcast.
This support provides an efficient bicast
capability for port snooping and
applications requiring line card protection.
Real-Time Traffic Support to
Enable Single Network
Infrastructures
To enable voice and data network
consolidation over a single network
infrastructure, the PRS Q-80G provides
real-time traffic support that addresses the
technical challenges (timing and latency)
of circuit emulation in access networks: the
PRS Q-80G Switch Fabric protects TDM
traffic over the other flows and has a
deterministic low latency.
Output Queued Shared Memory
Architecture for Best Performance
Output queuing is the best architecture for
performance because it is non-blocking
and provides full outbound throughput.
Memory sharing optimizes the use of high
speed RAM. The PRS Q-80G architecture,
with cells being sliced between multiple
switch devices, offers low latency.
Multiple Scheduling Disciplines to
Meet QoS Requirements
The scheduling discipline for retrieving
cells from the switch core memory can be
suited to the application QoS by selecting
among strict priority scheduling, credit
table (weighted round-robin), and
exhaustive highest priority.
PRS Q-80G Switch with C48X and C192X Fabric Interfaces
4 x OC-12
GE
GE
4 x GE
4 x OC-12
OC-48
Fibre
10G / OC-192
OC-48
PRS C48X
PRS C192X
PRS C192X
NP / TM
NP / TM
OC-48
NP / TM
OC-48
NP / TM
OC-48
NP / TM
OC-48
NP / TM
Protection Switch
Channel
Active Switch
PRS Q-80G
(two, four, or eight per plane)
Line Card
Line Card
Line Card
CSIX / PL3
CSIX / PL3
CSIX
20 Gbps
20 Gbps
20 Gbps
Backplane
8x3.2 Gbps
8x3.2 Gbps
8x3.2 Gbps
8x3.2 Gbps
8x3.2 Gbps
8x3.2 Gbps
Serial Links at
3.2 Gbps (with
8b/10b coding)
nP3700 (SPI-4.2)
nP3700 (SPI-4.2)
nP3700 (SPI-4.2)
nP3700 (SPI-4.2)
Empowering Intelligent Optical Networks
Product Brief
PRS Q-80G, PRS C48X, and PRS C192X
Switch Fabric
PB PRS Q-80G / v0.5 / 10-14-2004
Empowering Intelligent Networks
3
Switch Redundancy and Load
Sharing for High Availability and
Maintenance without Service
Interruption
The PRS Q-80G operates in redundant
switch plane configurations that address
the high-availability requirements of
service providers and five-nine
applications. If an element in one plane
fails, the other plane takes over. On the
line adapters, the PRS C48X and
PRS C192X Fabric Interface chips provide
a redundant path with the switch planes,
enabling traffic load sharing and
switchover between the two switch planes,
including maintenance switchover without
packet loss.
Backward Compatibility Enabling
Flexible Evolution Paths and
Investment Optimization
The PRS Q-80G, PRS C48X, and
PRS C192X chipset offers backward
compatibility enabling co-population of
both existing and new line cards in the
same chassis: the PRS Q-80G switch chip
SerDes can operate at 2.5 Gbps per serial
link (instead of 3.2 Gbps) to support line
cards based on the previous PRS C192
fabric interface chip, while the PRS C48X
and PRS C192X interface chips SerDes
can operate at 2.5 Gbps per serial link
(instead of 3.2 Gbps) to operate with
switch cards based on the previous
PRS 64Gu or PRS Q-64G switch chips.
This twofold compatibility provides very
flexible evolution paths from existing
systems while protecting system vendor
and end-user investments in existing line
cards and switch cards.
Complete System Solutions with
the AMCC's Broad Range of
Products
The PRS Q-80G switch core chip, along
with the PRS C48X and PRS C192X fabric
interface chips, is an integral part of a
complete system solution with the AMCC
nP3700, framers, and software tools,
enabling simple design of complete
solutions for POS, ATM, and Ethernet.
The nP3700, a 5-Gbps network processing
and traffic management solution, is the
highest performance integrated solution in
the industry. In addition to
high-performance packet processing and
fine-grained traffic management, the
nP3700 includes specialized coprocessors
that perform classification, policing, and
coherent database management for
unparalleled line-rate performance. The
nP3700 supports high-performance
memory interfaces such as RLDRAM II,
DDR DRAM, and DDR SRAM. The
devices are offered in different speed
grades to provide a range of performance
and cost options tuned to the application.
ATCATM Standard-based
Evaluation Platforms and Tools for
Time-to-Market
To help system vendors to evaluate and
prototype complete solutions, shorten
project cycles, mitigate development risks,
and optimize their development costs,
evaluation platforms, including a
PRS Q-80G-based 320-Gbps reference
switch board, standards-based ATCA
versions of the PRS 64Gu and PRS 80G
Switch Fabric evaluation boards, switch
core reference designs with the AMCC
PowerPC 405EP control processor,
reference software to operate switch
boards, board design services, and board
manufacturing capabilities complement the
AMCC PRS offering.
Validated Multi-Vendor Solutions
with 3rd-Party NP/TM Devices
To allow system vendors to leverage
validated technical solutions in focusing on
the development of their product offering,
solutions including the AMCC PRS switch
fabric, along with a choice of third-party
network processors, have been tested for
inter-operability validation. For example,
the PRS ATCA 80-Gbps Switch Fabric
evaluation boards have been validated in
ATCA platforms with ATCA blades from
major network processor vendors. AMCC
PRS evaluation platforms are enabled for
attachment of other suppliers' network
processor platforms, in order to allow
system vendors to evaluate multi-vendor
solutions.
PRS Q-80G Switch Core Device
Features
The PRS Q-80G switch core device offers
the following features:
Non-blocking, single-stage,
shared-memory switch
Device aggregate throughput: 80 Gbps
Two-chip switch: 80-Gbps aggregate
user bandwidth (8x8 ports at 10-Gbps/
OC-192c speed)
Four-chip switch: 160-Gbps aggregate
user bandwidth (16x16 ports at 10-Gbps/
OC-192c speed)
Eight-chip switch: 320-Gbps aggregate
user bandwidth (32x32 ports at 10-Gbps/
OC-192c speed)
2.5-Gbps ports with speed-up factor of
eight (in conjunction with PRS C48X
fabric interface)
10-Gbps ports with speed-up factor of
two (in conjunction with PRS C192X
fabric interface)
20-Gbps clear channel ports with
speed-up factor of two (in conjunction
with an ASIC or FPGA as fabric
interface)
40-Gbps clear channel ports with
speed-up factor of two (in conjunction
with an ASIC or FPGA as fabric
interface)
10-Gbps port operation in 4x 2.5-Gbps
mode (in conjunction with PRS C192X
fabric interface)
Embedded SerDes: 3.2-Gbps
high-speed serial links, XAUI compliant,
with 8b/10b encoding for link
synchronization and supervision
Multicast and broadcast without packet
duplication in shared memory (packet
replication at sending)
Full multicast and broadcast addressing
TDM traffic support
Traffic priorities (CoS): up to four
(configurable number)
Output queue scheduling: strict priority
scheduling, credit table (weighted
round-robin), exhaustive highest priority
first
AMCC reserves the right to make changes to its products, or to discontinue any product or service without notice, and advises its customers to obtain the latest
version of relevant information to verify, before placing orders, that the information being relied upon is current.
AMCC is a registered trademark of Applied Micro Circuits Corporation. Copyright 2004 Applied Micro Circuits Corporation. All Rights Reserved.
Other company, product, and service names may be trademarks or service marks of others.
6290 Sequence Drive San Diego, CA 92121 Tel: 858 450-9333 Fax: 858 450-9885 http://www.amcc.com
For technical support, please call 800 840-6055 or email productsupport@amcc.com
Product Brief
PRS Q-80G, PRS C48X, and PRS C192X
Switch Fabric
PB PRS Q-80G / v0.5 / 10-14-2004
4
Empowering Intelligent Networks
Redundant switch plane operations: hot
standby or load sharing mode,
programmable scheduled switchover
(packet lossless)
Flow control based on a grant
mechanism, subport flow control
Programmable flow control thresholds
for output queues, shared memory,
multicast/broadcast
Fixed-length packets with configurable
length (64, 72, or 80 bytes)
Programmable packet header size: two,
three, or five bytes, independent of the
switch configuration, containing
destination bitmap, packet priority, and
switch redundancy support information,
all protected by a parity bit
Local control processor interface (serial
host interface)
Reception (transmission) of control
packets sent to (by) the local processor
on any input (output) port
Switch control software compatibility with
PRS Q-64G switch core
Backward compatibility with the
PRS C192 fabric interface: SerDes
operation at 2.5 Gbps (speed-up factor
of 1.6)
Inter-operability in PRS Q-80G chassis:
line cards with any of the PRS C48X,
PRS C192, and PRS C192X devices
PRS C48X and PRS C192X Fabric
Interface Device Features
Port queuing manager (one per line
card) between one or multiple network
processors and a PRS 80G or
PRS Q-80G switch core
PRS 80G or PRS Q-80G interface:
embedded 3.2-Gbps SerDes, with 8b/
10b coding, and a speed-up factor of
two, or eight between the PRS Q-80G
and PRS C48X
PRS C48X: CSIX-L1 or PL3-based
(POS-PHY Level 3 / OIF-SPI3-01.0)
32-bit wide interface for multi-GE /
OC-48 network processor
PRS C192X: CSIX-L1 64-bit wide or
128-bit wide interface for 10-Gbps /
OC-192 network processor or, in subport
mode, 4x CSIX-L1 or PL3-based
(POS-PHY Level 3 / OIF-SPI3-01.0)
32-bit wide interfaces for multi-GE /
OC-48 network processors
PRS C192X operation in subport (or
quad PRS C48X) mode with PRS 64Gu,
PRS 80G, PRS Q-64G, and PRS Q-80G
switch cores
Dual-switch attachment for redundant
switch plane operations
Ingress virtual output queuing (VOQ),
with up to 1024 unicast queues and eight
(PRS C48X) or 32 (PRS C192X)
multicast queues, preventing
head-of-line blocking
Up to four levels of traffic priority (CoS)
and four programmable ingress queue
thresholds (ingress CSIX / PL3 flow
control per destination and priority)
Shared ingress buffer and egress buffer
capacities (configurable), with sharing
capability between planes
Switch interface ingress packet
scheduling: strict priority scheduling,
credit table (weighted round-robin),
exhaustive highest priority first
End-to-end packet payload protection,
with optional cyclic redundancy check
insertion
Internal loopback support for both the
CSIX / PL3 interface and switch interface
Backward compatibility with PRS 64Gu
switch core: SerDes operation at 2.5
Gbps, with speed-up factor of 1.6
Backward compatibility with PRS Q-64G
switch core: SerDes operation at 2.5
Gbps, with speed-up factor of 1.6 (PRS
C192X) or 6.4 (PRS C48X)
Inter-operability in PRS 64Gu and PRS
Q-64G chassis: line cards with any of the
PRS C48 (PRS 64Gu chassis only),
PRS C48X, PRS C192, and PRS C192X
devices
Specifications
PRS Q-80G
PRS C48X
PRS C192X
Technology
0.13 m
0.13 m
0.13 m
Supply voltage
1.5 V, 2.5 V
1.5 V, 2.5 V
1.5 V, 2.5 V
Typical power @100% traffic
14 Watts (master)
16 Watts (slave)
4.3 Watts
8 Watts
Junction C
0C to 125C
0C to 125C
0C to 125C
Package
624-ball CBGA
899-ball FC-PBGA
937-ball CBGA
mm x mm (pitch)
32.5x32.5 (1.27 mm)
31x31 (1 mm)
32.5 x 32.5 (1 mm)
PDF 
ZIP