Search:

Having more threads than cores could cause excessive overhead switching between threads. One thread per core could be better.
I don't know what HyperThreading means.

Try some experiments and...

Another minor point: Do these expressions one time at the a loop level, not inside the b loop
aPlusOneSq = (a+1)*(a+1)
aTimes8M1= 8*a-1 or a<<3 - 1