==2993== Cachegrind, a cache and branch-prediction profiler
==2993== Copyright (C) 2002-2017, and GNU GPL'd, by Nicholas Nethercote et al.
==2993== Using Valgrind-3.19.0 and LibVEX; rerun with -h for copyright info
==2993== Command: _bin/cxx-opt-sh/osh -n --ast-format none benchmarks/testdata/abuild
==2993== Parent PID: 2986
==2993== 
--2993-- warning: L3 cache found, using its data for the LL simulation.
==2993== 
==2993== I   refs:      44,724,379
==2993== I1  misses:       216,062
==2993== LLi misses:         8,405
==2993== I1  miss rate:       0.48%
==2993== LLi miss rate:       0.02%
==2993== 
==2993== D   refs:      19,284,383  (11,461,653 rd   + 7,822,730 wr)
==2993== D1  misses:       332,622  (   226,397 rd   +   106,225 wr)
==2993== LLd misses:        59,395  (    12,703 rd   +    46,692 wr)
==2993== D1  miss rate:        1.7% (       2.0%     +       1.4%  )
==2993== LLd miss rate:        0.3% (       0.1%     +       0.6%  )
==2993== 
==2993== LL refs:          548,684  (   442,459 rd   +   106,225 wr)
==2993== LL misses:         67,800  (    21,108 rd   +    46,692 wr)
==2993== LL miss rate:         0.1% (       0.0%     +       0.6%  )