diff --git a/README.md b/README.md
index f08fb1b..0e3938c 100644
--- a/README.md
+++ b/README.md
@@ -1,4 +1,14 @@
-clang-17 1brc.c -Wall -Werror -Wno-unused-parameter -std=c17 -march=native -mtune=native -Ofast -o 1brc
+#### Buiding
+`clang-17 1brc.c -std=c2x -march=native -mtune=native -Ofast -o 1brc`
 
-For optimal ~8 core results:
-./1brc measurments.txt 8
+AVX2 is required. GCC also works, but its binary is ~15% slower.
+
+#### Running
+`./1brc measurments.txt n`
+
+_n_ is the number of threads to use
+
+Work is spread evenly across threads with no rebalancing. If using hypethreading, _n_ must equal the number of available logical cores for best performance. Use `taskset` to limit the available logical cores.
+
+#### Implementation summary
+[Fastest known solution: 0.577s (8 core Zen2); C with heavy AVX2](https://github.com/gunnarmorling/1brc/discussions/710)