Reference Number: 327043-001
119
Intel® Xeon® Processor E5-2600 Product Family Uncore Performance Monitoring
2.9
R3QPI Performance Monitoring
2.9.1
Overview of the R3QPI Box
R3QPI is the interface between the Intel® QPI Link Layer, which packetizes requests, and the Ring.
R3QPI is the interface between the ring and the Intel® QPI Link Layer. It is responsible for
translating between ring prototcol packets and flits that are used for transmitting data across the
Intel® QPI interface. It performs credit checking between the local Intel® QPI LL, the remote Intel®
QPI LL and other agents onthe local ring.
The R3QPI agent provides several functions:
• Interface between Ring and Intel® QPI:
One of the primary attributes of the ring is its ability to convey Intel® QPI semantics with no
translation. For example, this architecture enables initiators to communicate with a local Home
agent in exactly the same way as a remote Home agent on another socket. With this philosophy,
the R3QPI block is lean and does very little with regards to the Intel® QPI protocol aside from
mirror the request between the ring and the Intel® QPI interface.
• Intel® QPI routing:
In order to optimize layout and latency, both full width Intel® QPI interfaces share the same ring
stop. Therefore, a Intel® QPI packet might be received on one interrface and simply forwarded
along on the other Intel® QPI interface. The R3QPI has sufficient routing logic to determine if a
request, snoop or response is targeting the local socket or if it should be forwarded along to the
other interface. This routing remains isolated to R3QPI and does not impede traffic on the Ring.
• Intel® QPI Home Snoop Protocol (with early snoop optimizations for DP):
The R3QPI agent implements a latency-reducing optimization for dual sockets which issues
snoops within the socket for incoming requests as well as a latency-reducing optimzation to
return data satisfying Direct2Core (D2C) requests.
2.9.2
R3QPI Performance Monitoring Overview
Each R3QPI Link in the uncore supports event monitoring through three 44b wide counters
(R3_Ly_PCI_PMON_CTR/CTL{2:0}). Each of these three counters can be programmed to count
almost any R3QPI event (see NOTE for exceptions). the R3QPI counters can increment by a maximum
of 5b per cycle (???).
For information on how to setup a monitoring session, refer to
Section 2.1, “Uncore Per-Socket
Performance Monitoring Control”
.
Note:
Only counter 0 can be used for tracking occupancy events. Only counter 2 can be used
to count ring events.
