|
Name |
PQ I |
PQ II |
PQ II Pro |
PQ III |
Description |
|
QE
DMA Emulation |
|
|
+ |
+ |
•
Provides a
replacement for the on-chip DMA module, which lack the “End of Transfer”
capability.
•
Microcode is
managed and controlled by Frame Descriptors (a.k.a. Buffer Descriptors).
•
H/W handshake is
supported (e.g. SOT, EOT).
•
Ideal for
proprietary FPGAs that stream data to the QE in order to be encapsulated
into a supported protocol (e.g. Ethernet, ATM, TDM). |
|
System Fabric Plane protocol (SFP.1, iTDM)
|
|
+ |
+ |
+ |
Sometimes referred to as Pseudo-Wire, this is a multiplexed voice over
packet protocol.
Media
processing, such as voice conferencing and voice mail, is being
increasingly implemented using off-the-shelf host platforms. At low
densities, VoIP works well for this, but when the density scales up,
VoIP is no longer viable, as the processing overhead is too high. Even
for some lower density host media processing applications, the latency
of VoIP becomes an issue. SFT.1 is perfectly suited to connect high
density host media processing platforms into a larger system.
In
addition, as voice and data networks converge, TDM busses will migrate
to packet bus implementations. However, this migration will not occur
all at once, and existing telephony equipment will have to continue to
inter-work with the newer packet bus equipment for some time. SFP.1 is
perfectly positioned to enable this migration.
SFP.1
is also a perfect protocol for carrying many TDM voice/data calls over
the packet bus itself. |
|
Ethernet Port-to-Port forwarding |
|
+ |
|
|
Forwarding based on
exact match of IP destination address. Any FCC pair can be selected for
forwarding, with none-matched packets forwarded to standard HOST queue.
Supports up to 128 IP exact match addresses, or IP address/port number
combination. Supports three way forwarding (all tree ports).
A free demo is available here. |
|
Generic Framing Protocol (GFP)
|
+ |
+ |
+ |
+ |
Supports
ITU-T G.7041. Provides a set of functions
that support a generic interface to underlying frame representation
systems (FRSs). The interface layer allows mapping of variable length,
higher-layer client signals over a transport network like SDH/SONET. The
client signals can be protocol data unit (PDU) oriented (like IP/PPP or
Ethernet Media access control [MAC]) or can be block-code oriented (like
fiber channel). Unlike the popular HDLC protocol, where the bandwidth
expansion is non-deterministic, GFP uses only the information in its
header for frame delineation, allowing deterministic bandwidth and
reducing latency. |
|
Coalescing |
|
+ |
|
|
Enables
the system to continue to maintain real-time performance, whilst
reducing the cost of the context switching by sharing the overhead
associated with interrupts between a configurable number of events.
Dramatically reduces the host Interrupt overhead. |
|
MemTest |
+ |
+ |
+ |
+ |
A vital
utility for the PowerQUICC II family, which runs a series of memory
functionality tests. These are carried out via microcode (using SDMA)
and not from the host, avoiding the interface with the host’s
instruction fetch. |
|
PowerµFilter™ |
|
+ |
|
+ |
Boost performance
significantly, by handling the MAC or IP-screening on-the-fly at
microcode level, eliminating the need for storing unwanted frames into
the memory (by the CPM) as well as reloading it for further processing
(by the CPU). |
|
CPM Idle Slots |
+ |
+ |
+ |
+ |
Analysis tool that measures the idle slots of the CPM (A situation
where no requests to the CPM scheduler are pending). This utility
detects if idle slots are available during the application CPM activity.
It can be used to figure out whether the application CPM requirements
are met. |
|
PQx
Queued SPI |
+ |
+ |
|
+ |
Generates Slave
Select (SS) signal for Master SPI. Supports up to 16 slaves. |
|
FCC Ethernet Classifiers |
|
+ |
|
+ |
Enhances the
capability and boosts the performance of the CPM’s Ethernet protocol.
This is by allowing the CPM to classify the incoming Ethernet frames
according to the defined patterns and masks, and then insert the
incoming frame into the corresponding RxBD ring and generates an event
(per ring) if instructed to. Frames that fail the match are discarded.
|
|
Ethernet MultiRing |
+ |
+ |
|
+ |
Separates frames from different protocols into different buffer
descriptor rings, resulting in significant saving on Host processing
time, and enables the application to set up different priorities to
different protocols. Supports the following protocols:
ARP, ICMP,
TCP, NVP, UDP, IPv6,
IPv6OverIP. Can
be customized to use different or add other L3/L4 protocols (e.g. RTP
instead of NVP).
The package has
been successfully deployed by several major companies. |
|
IP Filter |
+ |
+ |
|
+ |
Boost Ethernet driver performance
significantly, by handling the IP screening on-the-fly at microcode
level. Stores only frames with an IP that matches a user defined list. frame can
be further sorted to up to eight BD rings providing priority policy
support.
|
|
FCC Ethernet WFQ |
+ |
+ |
|
+ |
Manages transmit QoS, by implementing a Weighted Fair Queue
(WFQ). It is a flow-based queuing algorithm that schedules low-volume
traffic first, while allowing high-volume traffic share the remaining
bandwidth. |
|
FCC Ethernet UDP Forwarding |
|
+ |
|
+ |
A
utility that illustrates the throughput boost of implementing uCode at
the CPM level. It shows the speed and cost-effectiveness of microcode
compared to the traditional methods of high-level code or costly,
dedicated hardware. This patch is applicable to any PQII or PQIII FCC
channel that is configured to operate in an Ethernet mode.
A 60-day evaluation copy is available here. |
|
Header’s MRBLR |
+ |
+ |
|
+ |
Special FCCE event is generated after desirable amount of bytes of the
incoming frame are received. Permits early header processing. |
|
IP/TCP/UDP Checksum Calculation |
+ |
+ |
|
+ |
Per BD request for IP/TCP/UDP. Saves substantial host
processing time! |
|
Fast Ethernet
Frame Time
Stamp |
+ |
+ |
+ |
+ |
Replaces
the CRC field in the received buffer (typically not required) with an
End-Of-Frame timestamp.
A 60-day evaluation copy is available here. |
|
MAC Filter
|
|
+ |
|
+ |
Boost Bridge performance significantly, by handling the MAC screening
on-the-fly at microcode level. Stores only frames with a MAC match with
a user list. frame can be further sorted to up to eight BD rings
suggesting priority policy support |
|
VLAN Support |
+ |
+ |
|
+ |
Incoming frame are sorted according to
their VLAN priority to eight BD rings. Another ring is available to
none-VLAN frames. Each ring has its own FCCE event. |
|
Buffer Pool manger |
+ |
+ |
+ |
+ |
Fundamental microcode utility. The Rx BD is
allocated a buffer from buffer pool(s), on demand. The application frees
the buffer back to pool by command, to avoiding possible mutual
inclusion. Overall performance enhancement due to faster buffer
administration, as well as fewer buffers, as they are allocated "on
demand". |
|
MII interface over SPI |
|
+ |
|
|
Needed when frequent messages are exchanged
with external logic (or PHY) along the MII interface. Especially useful
with Marvel(TM) PHYs. (CPLD is needed) |
|
Automatic Ethernet Flow
Control |
+ |
+ |
|
+ |
Automatically "pause frame" when Rx busy is
encountered. Can be configured to send pause only after N consecutive
busy states. |
|
IDMA Peripheral to
Memory or Memory to Peripheral in a none fly-back mode |
|
+ |
|
|
Capable of transferring data to/from
peripheral from/to memory. Time-out mechanism is implemented to push
data to destination in absent of DREQ during a configurable time-out
window. This feature is especially useful in peripheral to memory
transfer in a bandwidth sensitive system. |
|
Protocol Parser over
Ethernet |
+ |
+ |
|
+ |
Parses incoming Ethernet frames to up to
eight BD rings according to protocol type (UDP, TCP, ARP etc.)
Additional BD ring is dedicated to "none-in-list" protocol. Each ring
has private flag in the FCCE. |
