Improve README for race and atomic exercise

Author: bernhardmgruber

code/atomic/README.md

@@ -0,0 +1,13 @@
+
+## Instructions
+
+You know this program already from "racing".
+It tries to increment an integer 200 times in two threads.
+Last time, we fixed the race condition using a lock, but now we'll try atomics.
+
+Tasks:
+- Replace the counter 'a' by an atomic.
+  Run the program, and check for race conditions.
+- Go back to 'racing', and check the execution time of the atomic vs the lock solution,
+  e.g. using `time ./atomic`
+  You might have to increase the number of tries if it completes too fast.

code/atomic/atomic.cpp

@@ -2,17 +2,6 @@
 #include <iostream>
 #include <thread>
 
-/*
- * You know this program already from "racing". It tries to increment an integer 200 times in two threads.
- * Last time, we fixed the race condition using a lock, but now we'll try atomics.
- *
- * Tasks:
- * - Replace the counter 'a' by an atomic. Run the program, and check for race conditions.
- * - Go back to 'racing', and check the execution time of the atomic vs the lock solution, e.g. using
- *   time ./atomic
- *   You might have to increase the number of tries if it completes too fast.
- */
-
 int main() {
   int nError = 0;
 

code/atomic/solution/atomic.sol.cpp

@@ -2,17 +2,6 @@
 #include <iostream>
 #include <thread>
 
-/*
- * You know this program already from "racing". It tries to increment an integer 200 times in two threads.
- * Last time, we fixed the race condition using a lock, but now we'll try atomics.
- *
- * Tasks:
- * - Replace the counter 'a' by an atomic. Run the program, and check for race conditions.
- * - Go back to 'racing', and check the execution time of the atomic vs the lock solution, e.g. using
- *   time ./atomic
- *   You might have to increase the number of tries if it completes too fast.
- */
-
 int main() {
   int nError = 0;
 

code/race/README.md

@@ -1,6 +1,9 @@
 
 ## Instructions
 
-* compile, run many times, see what happens
-* use a mutex to fix the issue
-* see the difference in execution time
+* Compile, run many times, see what happens
+    * E.g. in bash, use: `while true; do ./racing; done`
+* (Optional) You can use `valgrind --tool=helgrind ./racing` to proof your assumption
+* Use a mutex to fix the issue
+* See the difference in execution time
+* (Optional) Check agan with `valgrind` if the problem is fixed