Thanks for sending the Harris paper.
Just now I had the time to read it.
Attached I sent a screenshot of the performance chart presented there.
Did you noticed the difference between the Mutex and the New version?
What do you think is the O() of each one?
And about the Valois version?
Just looking in this chart, It's possible to conclude if it's lock free or no? Why?
Dmitry?