PQOS tool measures 0.0 for MBL and MBR with MLC on AMD systems

When using the pqos tool to monitor memory bandwidth, both MBL (Memory Bandwidth Local) and MBR (Memory Bandwidth Remote) are reported as 0.0, even while a memory-intensive workload such as MLC is running.

Steps to Reproduce
Boot an AMD system with Linux kernel v6.18 or later.
Ensure the system supports the mbm_event monitoring mode and that it is enabled.
Run a memory bandwidth workload (for example, MLC).
In parallel, run the pqos monitoring command:
#sudo pqos -m all:[0-191]

Example output:

CORE IPC MISSES LLC[KB] MBL[MB/s] MBR[MB/s]
0-191 0.01 109937k 385600.0 0.0 0.0

Observed Behavior

The MBL and MBR counters remain at 0.0 despite ongoing memory traffic generated by MLC.

Additional Information

The issue is observed when the system is operating in mbm_event mode and the number of available MBM counters (num_mbm_cntrs) is limited to 32.
#cat /sys/fs/resctrl/info/L3_MON/mbm_assign_mode
[mbm_event]
default

#cat /sys/fs/resctrl/info/L3_MON/num_mbm_cntrs
0=32;1=32

This issue appears to be specific to AMD systems that support and have enabled the mbm_event assignment mode.

Start with the pqos monitoring path exercised by `sudo pqos -m all:[0-191]`, focusing on AMD systems using `mbm_event` with 32 available MBM counters. Reproduce the issue with MLC and the provided resctrl settings, then verify that MBL and MBR report the generated memory traffic instead of 0.0.

Written by the indexing model from the issue text.