==2964== Cachegrind, a cache and branch-prediction profiler
==2964== Copyright (C) 2002-2017, and GNU GPL'd, by Nicholas Nethercote et al.
==2964== Using Valgrind-3.19.0 and LibVEX; rerun with -h for copyright info
==2964== Command: _bin/cxx-opt-sh/osh -n --ast-format none benchmarks/testdata/test-cmd-util.sh
==2964== Parent PID: 2957
==2964== 
--2964-- warning: L3 cache found, using its data for the LL simulation.
==2964== 
==2964== I   refs:      96,114,800
==2964== I1  misses:       437,015
==2964== LLi misses:         8,698
==2964== I1  miss rate:       0.45%
==2964== LLi miss rate:       0.01%
==2964== 
==2964== D   refs:      41,646,734  (24,452,968 rd   + 17,193,766 wr)
==2964== D1  misses:       639,738  (   429,414 rd   +    210,324 wr)
==2964== LLd misses:       116,916  (    13,702 rd   +    103,214 wr)
==2964== D1  miss rate:        1.5% (       1.8%     +        1.2%  )
==2964== LLd miss rate:        0.3% (       0.1%     +        0.6%  )
==2964== 
==2964== LL refs:        1,076,753  (   866,429 rd   +    210,324 wr)
==2964== LL misses:        125,614  (    22,400 rd   +    103,214 wr)
==2964== LL miss rate:         0.1% (       0.0%     +        0.6%  )