How to profile LLC misses


I am trying to profile the LLC misses on my Jetson TX2, with the JetPack 3.3.
For recording cache misses I am exploiting the perf tool, installed as reported in this tutorial:

the problem is that LLC counter is not supported, for example if I try to profile an executable:

perf stat -d ./lu.exe 

 Performance counter stats for './lu.exe':

       5017.181600      task-clock (msec)         #    0.795 CPUs utilized          
             2,357      context-switches          #    0.470 K/sec                  
                14      cpu-migrations            #    0.003 K/sec                  
            28,978      page-faults               #    0.006 M/sec                  
     9,042,670,509      cycles                    #    1.802 GHz                    
   <not supported>      stalled-cycles-frontend  
   <not supported>      stalled-cycles-backend   
     6,198,291,733      instructions              #    0.69  insns per cycle        
   <not supported>      branches                 
         4,012,545      branch-misses             #    0.00% of all branches        
     2,991,571,379      L1-dcache-loads           #  596.265 M/sec                  
       191,461,394      L1-dcache-load-misses     #    6.40% of all L1-dcache hits  
   <not supported>      LLC-loads                
   <not supported>      LLC-load-misses          

       6.310669418 seconds time elapsed

and if I run perf list, LLC-loads and LLC-load-misses are not listed.


LLC events are not present in Cortex-A57 and added later. Cortex A76 has LL_CACHE_RD, LL_CACHE_MISS_RD.
Related patch added recently

there are no other ways to profile LLC misses on TX2/TX1?