|Non-Confidential||PDF version||ARM 100052_0001_00_en|
|Home > Performance Optimization and Monitoring > RN-I performance events > Bandwidth at RN-I bridges|
External devices connect to a CCN-502 at an RN-I bridge.
To monitor the behavior of the system, the following events measure the read bandwidth at each RN-I bridge:
|RDataBeats_Port0||Number of RData beats, RVALID and RREADY, dispatched on port 0. This is a measure of the read bandwidth.|
|RDataBeats_Port1||Number of RData beats, RVALID and RREADY, dispatched on port 1. This is a measure of the read bandwidth.|
|RDataBeats_Port2||Number of RData beats, RVALID and RREADY, dispatched on port 2. This is a measure of the read bandwidth.|
Because CMOs are sent through the read channel, their responses are included in these events.
Calculate the read bandwidth as follows:
Where AXIDataBeatSize is the number of bytes for each AXI beat. In most cases, this is the same size as the AXI bus.
RXDATFLITV measures the bandwidth that an RN-I bridge sends to the interconnect.
To measure the actual bandwidth that an RN-I bridge sends to the interconnect, and not the useful bandwidth the external devices can use, this event counts the number of received data flit requests that the bridge receives through the data channel:
|RXDATFLITV||Number of RXDAT flits received. This event is a measure of the true read data bandwidth. It excludes CMOs, because CMO completions return to the RN-I through the response channel, but includes replayed requests.|
This event includes the replayed requests because of the read data buffer decoupled scheme.
Calculate the actual read bandwidth as follows:
TXDATFLITV monitors the number of data flits that the RN-I bridge sends out.
In a similar way to the read actual bandwidth event, this event monitors the number of data flits that the RN-I bridge sends out, to measure the actual write bandwidth that is sent to the interconnect:
|TXDATFLITV||Number of TXDAT flits dispatched. This event is a measure of the write bandwidth.|
Calculate the write bandwidth as follows:
To improve efficiency when using PMU events and signals, TXREQFLITV_TOTAL combines the read and write bandwidth estimation in a single event.
TXREQFLITV_TOTAL achieves this by monitoring the number of request flits that are sent from an RN-I bridge:
|TXREQFLITV_TOTAL||Number of TXREQ flits dispatched. This event is a measure of the total request bandwidth.|
To use this event correctly, you must know the average request data size for both reads and writes in your system. If the AXI masters issue a mixture of request sizes, you must estimate the average size of read and writes, using the PMU in an AXI master or an AXI interrupt controller.
Calculate the total bandwidth as follows: