High-performance, scalable technology for
enterprise-network and Internet applications
PowerNP NP4GS3 Network Processor
Highlights
Advanced, programmable
switching and routing system
on a single chip
High performance and capabili-
ties for complex applications,
operating at wire speed
Comprehensive solution with
integrated MACs, tree searches,
16 picoprocessors and software
offerings, improving time to
market
Supports large memory, using
DDR-SDRAMs, allows cost-
efficient designs for high-level
packet forwarding and filtering
Rich enablement environment
and proven technology to help
deliver high quality products
more rapidly with lower power
product designs
Complex functions, delivered
at wire speed
The IBM PowerNP NP4GS3 network
processor is an advanced, robust,
programmable, high-performance
solution for the most demanding routing
applications. A new high-end member of
the IBM PowerNP family of network
processors, the NP4GS3 implements
copper interconnect technology and
integrates a switching engine, search
engine, frame processors, and
multiplexed MACs. Designed to satisfy
enterprise, core, and edge networking
and Internet requirements, at wire speed,
the NP4GS3 processor is scalable to
meet increasing bandwidth and func-
tional demands. It can deliver complex
functions, such as quality of service
(QoS), scheduling, flow control, and
differentiated services.
Manufacturers who take advantage of
the IBM NP4GS3 in their product
designs can benefit from leadership
semiconductor technologies and proven
network experience. IBM offers a rich
suite of hardware and software
enablement tools and services to help
reduce time to market, and ensure
quality products that meet functional
specifications.
The NP4GS3 enables product differen-
tiation through the combination of
specific features for routing networking
applications including:
16 programmable picoprocessors
providing 2128 MIPS aggregate
processing capability
Embedded IBM PowerPC
processor
providing additional design flexibility for
network interconnect devices
Hardware accelerators performing
functions such as tree searches, frame
forwarding, frame filtering and frame
alteration
40 Fast Ethernet/4-Gb MACs with SMII,
GMII, and 8B/10B interfaces supporting
industry standard PHY components
OC-48c, OC-48, four OC-12, and sixteen
OC-3 integrated Packet over Sonet
(POS) interfaces supporting industry
standard POS framers
Advanced flow control designed to help
prevent TCP collapse
Hardware support for port mirroring for
remote traffic monitoring
Supporting multi-threads to improve
performance
Programmable, high-
performance processing
The exceptional processing capability
of the IBM PowerNP NP4GS3 is enabled
from its embedded processor complex,
which is made up of 16 picoprocessors,
multiple hardware accelerators, and a
PowerPC microprocessor. A pair of
picoprocessors share hardware
accelerators in a configuration that can
process up to 32 frames in parallel and
provide an aggregate 2,128 MIPs. The
hardware accelerators perform tree
searches, frame forwarding, frame
filtering, frame alteration, and
other functions.
The IBM PowerNP NP4GS3 network
processor