Intel NETWORK PROCESSOR IXP2800 - Manual (Page 376)

376 Hardware Reference Manual
Intel
®
IXP2800 Network Processor
Performance Monitor Unit
11.1.2 Motivation for Choosing CHAP Counters
The Chipset Hardware Architecture Performance (CHAP) counters enable statistics gathering of
internal hardware events in real-time. This implementation provides users with direct event
counting and timing for performance monitoring purposes, and provides enough visibility into the
internal architecture to perform utilization studies and workload characterization.
This implementation can also be used for chipset validation, higher-performing future chipsets, and
applications tuned to the current chipset. The goal is that this will benefit both internal and external
hardware and software development. The primary motivation for selecting the CHAP architecture
for use in the IXP2800 Network Processor product family is that it has been designed and validated
in several Intel desktop chipsets and the framework also provides a software suite that may be
reused with little modification.
Figure 136. Performance Monitor Interface Block Diagram
Performance Monitoring
Unit
APB Bus for
Read/Write of
CHAP Registers
Status Conditions
for Interrupts from
CHAP Counter 0
Status Conditions
for Interrupts from
CHAP Counter 1
Status Conditions
for Interrupts from
CHAP Counter N-2
Status Conditions
for Interrupts from
CHAP Counter N-1
Event Multiplexer
Control to
Hardware Blocks
Events from
Hardware Block A
Events from
Hardware Block B
Events from
Hardware Block C
Events from
Hardware Block D