It's worth keeping in mind that Valgrind's cachegrind is a simulation. If you have access to a Linux system you can make use of the perf tool to get real, honest numbers about your program's cache performance and more. This is highly recommended and something we'll do throughout this book. Like git, perf is many tools arranged under a banner—perf—with its own subcommands that have their own options.
It is a tool well worth reading the documentation for. Anyhow, what does the standard look like under perf?
naive_hashmap > perf stat --event task-clock,context-switches,page-faults,cycles,instructions,branches,branch-misses,cache-references,cache-misses target/release/standard > /dev/null Performance counter stats for ...