EMBEDDED SYSTEM ON PCI

Tejaswini
ECE II year

JAYAPRAKASH NARAYANA EDUCATIONAL SOCIETY

ABSTRACT:
A split-transaction protocol is
The presentation involves the success of
implemented with attributed packets that
the widely adopted PCI bus and
are prioritized and optimally delivered to
describes a higher performance next
their target. The new PCI Express
generation of I/O interconnect, called
Architecture comprehends a variety of
PCIExpress* Architecture that will serve
form factors to support smooth
as a standard local I/O bus for a wide
integration with PCI and to enable new
variety of future computing platforms.
system form factors. PCI Express
Key PCI attributes, such as its usage
Architecture will provide industry
model and software interfaces are
leading performance and
maintained whereas its bandwidth-
price/performance.
limiting, parallel bus implementation is
replaced by a long-life, fully serial
interface.
 The PCI bus has served us well for the last 10
years and it will play a major role in the next
few years. However, today’s and tomorrow’s
processors and I/O devices are demanding
INTRODUCTION: much higher I/O bandwidth than PCI 2.2 or
PCI-X can
deliver and it is time

To engineer a new generation of the PC platform supporting a variety of

PCI to serve as a standard I/O bus for application-specific buses alongside the
future generation platforms. There have PCI I/O expansion bus as shown in
been several efforts to create higher Figure 1.
bandwidth buses and this has resulted in
Figure 1. Today’s PC has multiple local buses with different requirements.

memory hub and an I/O hub since the

memory bus often changes with each
processor generation. One of the major
The processor system bus
functions of the chipset is to isolate these
continues to scale in both frequency and
ever-changing buses from the stable I/O
voltage at a rate that will continue for the
bus. Close investigation of the 1990’s
foreseeable future. Memory bandwidths
PCI signaling technology reveals a
have increased to keep pace with the
multi-drop, parallel bus implementation
processor. Indeed, as shown in Figure 1,
the chipset is typically partitioned as a
that is close to its practical limits of there is no baseline support for this
performance. time–dependant data within the PCI 2.2
or PCI-X specifications Today’s
Today’s software applications are more platforms, an example desktop PCI is
demanding of the platform hardware, shown in Figure 2, must also deal with
particularly the I/O subsystems. multiple concurrent
Streaming data from various video and Transfers at ever-increasing data rates.
audio sources are now commonplace on
the desktop and mobile machines and

Figure 2. Multiple concurrent data transfers.

EMBEDDED SYSTEM OVERVIEW: comparison to older systems with full

functional hardware or systems with
An embedded system is a special-
general purpose hardware and externally
purpose computer system, which is
loaded software. Embedded systems are
completely encapsulated by the device it
a combination of hardware and software
controls. An embedded system has
which facilitates mass production and
specific requirements and performs pre-
variety of application.
defined tasks, unlike a general-purpose
personal computer. An embedded What is PCI and what made to move PCI
system is a programmed hardware express?
device.A programmable hardware chip 1.Supports multiple market segments
is the 'raw material' and it is and emerging applications:
programmed with particular
applications. This is to be understood in
 Unifying I/O Architecture for Base mechanisms to enable Embedded
Desktop, Mobile, Server, and Communications applications.
Communications Platforms, 8.Non-Goals:
Workstations and Embedded Devices. Coherent interconnect for
2.Low cost and high volume: processors, memory interconnect, and
Cost at or below PCI cost cable interconnect for cluster solutions.
structure at the system level
3.PCI Compatible software model PCI EXPRESS* ARCHITECTURE
4.Boot existing operating systems OVERVIEW:
without any change. PCI compatible
configuration and device driver A PCI Express multi-drop,
interfaces. parallel bus topology contains a Host
5.Performance: Bridge and several endpoints (the I/O
Scalable performance via devices) as shown in Figure 3. Multiple
frequency and additional lanes. High point-to-point connections introduce a
Bandwidth per Pin. Low new element, the switch, into the I/O
Overhead. Low latency. system topology also shown in Figure 3.
6.Support multiple platform connection The switch replaces the multi-drop bus
types: and is used to provide fan-out for the I/O
Chip-to-chip, board-to-board via bus. A switch may provide peer-to-peer
connector, docking station and enable communication between different
new form factors. endpoints and this traffic, if it does not
7.Advanced features: involve cache-coherent memory
Comprehend different data types. transfers, need not be forwarded to the
Power Management. Quality Of Service. host bridge. The switch is shown as a
Hot Plug and Hot Swap support. Data separate logical element but it could be
Integrity and Error Handling. Extensible. integrated into a host bridge component.
 Figure 3. A switch is added to the system topology

The multiple, similar parallel buses of performance I/O. The switch is a logical
today’s platform are replaced with PCI element that may be implemented within
Express links with one or more lanes. a component that also contains a host
Each link is individually scalable by bridge, or it may be implemented as a
adding more lanes so that additional separate component. It is expected that
bandwidth may be applied to those links PCI will coexist in many platforms to
where it is required – such as graphics support today’s lower bandwidth
in the desktop platform and bus bridges applications until a compelling need,
(e.g. PCI Express-to-PCI-X) in the such as a new form factor, causes a full
server platform. A PCI Express switch migration to a fully PCI Express based
provides fanout capability and enables a platform
series of connectors for add-in, high

The server platform requires more I/O high bandwidth PCI Express links to
performance and connectivity including PCI-X slots, Gigabit Ethernet* and an
InfiniBand* fabric. Figure 5 shows how PCI Express for “inside the I/O and
PCI Express provides many of the same cluster interconnect, allows servers to
advantages for servers, as it does for transition from “parallel shared buses” to
desktop systems. The combination of a high speed serial interconnect

Figure 5. PCI Express-based Server/Workstation System

The networking communications different traffic types. It too would
platform could use multiple switches for benefit from a multiple PCI Express
increased connectivity and Quality of links that could be constructed as a
Service (QOS) for differentiation of modular I/O system

Figure 6. PCI Express-based Networking Communications System

The PCI Express Architecture is

specified in layers as shown in Figure 7.
PCI EXPRESS ARCHITECTURE: Compatibility with the PCI addressing
model (a load-store architecture with a
flat address space) is maintained to
ensure that all existing applications and these packets to create a highly reliable
drivers operate unchanged. PCI Express data transfer mechanism. The basic
configuration uses standard mechanisms physical layer consists of a dual-simplex
as defined in the PCI Plug-and-Play channel that is implemented as a
specification. The software layers will transmit pair and a receive pair. The
generate read and write requests that are initial speed of 2.5 Giga
transported by the transaction layer to transfers/second/direction provides a
the I/O devices using a packet-based, 200MB/s communications channel that
split-transaction protocol. The link layer is close to twice the classic PCI data
adds sequence numbers and CRC to rate.

Figure 7. The PCI Express Architecture is specified in layers

The remainder of this section will look transmit pair and a receive pair as shown
deeper into each layer starting at the in Figure 8. A data clock is embedded
bottom of the stack. using the 8b/10b-encoding scheme to
achieve very high data rates. The initial
frequency is 2.5 Giga
transfers/second/direction and this is
PHYSICAL LAYER: possible to increase with silicon

The fundamental PCI Express link technology advances of up to 10 Giga

consists of two, low-voltage, transfers/second/direction (the

differentially driven pairs of signals: a theoretical maximum for signals in

copper). The physical layer transports packets between the link layers of two
PCI Express agents.

The bandwidth of a PCI Express link may be frequency of operation by the two agents at each
linearly scaled by adding signal pairs to form end of the link. No firmware or operating system
multiple lanes. The physical layer supports x1, software is involved.
x2, x4, x8, x12, x16 and x32 lane widths and
splits the byte data as shown in Figure 9. Each
byte is transmitted, with 8b/10b encoding, across Figure 9. A PCI Express Link consists of
the lane(s). This data disassembly and re- one or more lanes
assembly is transparent to other layers. During
initialization, each PCI Express link is set up
following a negotiation of lane widths and

LINK LAYER:

The primary role of a link layer is

Figure 10. The Link Layer adds data
to ensure reliable delivery of the packet
integrity features
across the PCI Express link. The link
Most packets are initiated at the
layer is responsible for data integrity and
Transaction Layer (next section). A
adds a sequence number and a CRC to
credit-based, flow control protocol
the transaction layer packet as shown in
ensures that packets are only transmitted
Figure 10.
when it is known that a buffer is
available to receive this packet at the
other end. This eliminates any packet
retries, and their associated waste of bus
bandwidth due to resource constraints.
The Link Layer will automatically retry Software compatibility is of
a packet that was signaled as corrupted. paramount importance for a third
generation local I/O bus. There are two
facets of software compatibility;
TRANSACTION LAYER:
initialization, or enumeration, and run-
The transaction layer receives time. PCI has a robust initialization
read and write requests from the model wherein the operating system can
software layer and creates request discover all of the add-in hardware
packets for transmission to the link devices present and then allocate system
layer. All requests are implemented as resources, such as memory, I/O space
split transactions and some of the request and interrupts, to create an optimal
packets will need a response packet. The system environment. The PCI
transaction layer also receives response configuration space and the
packets from the link layer and matches programmability of I/O devices are key
these with the original software requests. concepts that are unchanged within the
Each packet has a unique identifier that PCI Express Architecture; in fact, all
enables response packets to be directed operating systems will be able to boot
to the correct originator. The packet without modification on a PCI Express-
format supports 32bit memory based platform.
addressing and extended 64bit memory
addressing. Packets also have attributes
such as “no-snoop”, “relaxed-ordering”
and “priority” which may be used to
optimally route these packets through
the I/O subsystem.

MECHANICAL FORM FACTORS:

The low signal-count of a PCI
Express link will enable both an
SOFTWARE LAYERS:
evolutionary approach to I/O subsystem
design and a new modular approach that new connector placed alongside the
will encourage new system partitioning. existing PCI or AGP connector in the
area previously occupied by those types
of connectors as in Figure 11.
EVOLUTIONARY DESIGN:
In addition to the base implementations
Initial implementations of PCI
of PCI Express-based add-in cards that
Express-based add-in cards will co-exist
replace AGP and PCI cards, other form
alongside the current PCI-form factor
factors are in progress of being
boards (both full-size and “half-height”).
developed. This NEWCARD will offer
As an example connection; a higher
integrated desktop and mobile external
bandwidth link, such as a sixteen-lane
expansion into a single
connection to a graphics card, will use a

standard. Slated for release later this desktop devices, similar to how USB
year, the NEWCARD standard is devices can be shared. Desktop users
targeted for OEMs seeking small will benefit from NEW Cards plug-n-
footprint for thinner mobile designs and play ease of use, which eliminates the
sealed systems for desktop designs. need to open the chassis to add new
NEWCARD is the next evolution of the features. The new standard is operating
PC Card, combining a smaller form system independent.
factor and faster performance with the
PC Cards reliability and ease of use.
NEWCARD supports hot swappable
device sharing between mobile and
scalable performance will enable it to
CONCLUSION: become a unifying I/O solution across a
broad range of platforms – desktop,
The PCI Express Architecture meets all mobile, server, communications,
of the requirements of a third generation workstations and embedded devices. A
I/O bus. It’s advanced features and PCI Express link is implemented using
multiple, point-to-point connections
called lanes and multiple lanes can be
used to create an I/O interconnect whose
bandwidth is linearly scalable. This
interconnect will enable flexible system
partitioning paradigms at or below the
current PCI cost structure. PCI Express
is software compatible with all existing
PCI-based software to enable smooth
integration within future systems.

BIBILOGRAPHY:
www.softwaresolutions.com
www.embeddedsystems.net
www.pciexpressarchitecture.net
Electronics Maker magazine.
PCI architecture by David H. Albenosi.