x86_zen data is getting populated in the x86_64 dashboard in https://hud.pytorch.org/benchmark/compilers #6823
Description
I see the x86_zen data getting populated in the x86_64 dashboard in https://hud.pytorch.org/benchmark/compilers
Also the parsing of the log files is not happening correctly leading to the file names getting uploaded under x86_64
From our initial analysis it looks post this PR (#6793) the parsing logic has changed for the benchmarking data.
It now expects the file names to have arch as cpu and device as x86_zen instead of the earlier convention of zen_cpu_x86 (added in the test_perf_for_dashboard method - https://github.com/pytorch/pytorch/blob/main/.ci/pytorch/test.sh#L582)
Should we change it to cpu_x86_zen to align to the latest parsing logic?
cc @huydhn @desertfire @malfet