How to measure L1 and L2 cache hit rate with analyzing memory objects?

[CKang5]

Hello, everyone:

I am a new user of Intel Vtune. I want to measure the L1 and L2 cache hit rate on intel Xeon E5-2698 v4 (Broadwell-EP) with the Data Linear Address Profiling.

So I'm trying to generate a simple memory trace with the analysis type like "analyze memory objects".
here is vtune command line
"amplxe-cl -collect memory-access -knob dram-bandwidth-limits=false -knob analyze-mem-objects=true -data-limit=0 -- COMMAND"
and here is the result

I know that the Data Linear Address Profiling also support L1 and L2 cache hardware events called MEM_LOAD_UOPS_RETIRED.L1_HIT and MEM_LOAD_UOPS_RETIRED.L2_HIT.

but I can see the only LLC Miss count in the VTune.

Is there any way to measure L1 and L2 cache hit rate in the "Memory access" analysis type?

[Dmitry_R_Intel1]

Hello,

You'll need to use custom analysis and manually specify all the events. For example:

amplxe-cl -collect-with runsa -knob event-config=CPU_CLK_UNHALTED.REF_TSC,MEM_LOAD_UOPS_RETIRED.L1_HIT_PS,MEM_LOAD_UOPS_RETIRED.L1_MISS_PS,MEM_LOAD_UOPS_RETIRED.L3_HIT_PS,MEM_LOAD_UOPS_RETIRED.L3_MISS_PS,MEM_UOPS_RETIRED.ALL_LOADS_PS,MEM_UOPS_RETIRED.ALL_STORES_PS,MEM_LOAD_UOPS_RETIRED.L2_HIT_PS:sa=100003,MEM_LOAD_UOPS_RETIRED.L2_MISS_PS -knob collectMemBandwidth=true -knob dram-bandwidth-limits=true -knob collectMemObjects=true