for remote socket cache-to-cache data transfer, why data homed in reader socket shows higher latency than data homed in writer socket?

Question

Here is what I measured under Cascade lake platform:

Intel(R) Memory Latency Checker - v3.11
Command line parameters: --c2c_latency

Measuring cache-to-cache transfer latency (in ns)...
Remote Socket L2->L2 HITM latency (data address homed in writer socket)
                        Reader Numa Node
Writer Numa Node     0       1
            0        -   113.0
            1    112.6       -
Remote Socket L2->L2 HITM latency (data address homed in reader socket)
                        Reader Numa Node
Writer Numa Node     0       1
            0        -   176.5
            1    173.5       -

In this case, remote L2->L2 data access shows different latency. Data address homed in reader socket shows higher latency than in writer socket.

The results is correct. I just want to know how to explain these results. I think for both cases, data is transferred from remote L2 cache. However, it seems like CPU do more things when data address is homed in reader's socket?