Usermode busy waiting for performance gains #38
Description
After testing, it is possible for the usermode to no longer call WaitForSingleObject when awaiting a kernel response, but instead run a busy wait for a kernel response. This appears to improve read and write performance significantly, and we will no longer be dependent on the thread scheduler and time it takes to wake up the thread.
But this should be tested further and a more thorough solution should be implemented.
Just making the following changes, allowed for about a 600% performance improvement for reads and writes:
Average performance (~700k-900k per second):
