to get better performance with version 5 client run multiple instances
for example with a dual Xeon E5-2690 I get about 380 Col/min with 32 threads
but running two instances of ptsminer_64bits with 16 threads each yields 220 col/min
I also set the affinity of each process ( using procexp) so each instance runs from a single cpu, seems to add about 10+/- col/min
so 440 col/min vs 380