I recently learned about the LLC prefetcher for the intel Xeon processor family and read in the "4th Gen Intel® Xeon® Scalable Processor XCC (Codename Sapphire Rapids) Uncore Performance" manual about the LLCPrefData and the DRd_Pref and other prefetch events, I would like to know if the essential difference between these 2 events is that they prefetch to different destinations, i.e. to LLC and L2 cache respectively? Will these 2 events overlap (i.e. is it possible that an LLCPrefData request will be converted into a DRd_Pref request or something like that)? I've been monitoring the throughput of these 2 events lately and noticed that the throughput for the DRd_Pref event looks odd when LLC Prefetch is abled.
