"make all" will compile four benchmarks, which are also individual make targets:

"singleVc"(default) - Vector classes of SSE vectorised single precision benchmark
"single" - the original SSE vectorised single precision benchmark
"double" - SSE vectorised double precision benchmark
"tbb" - multithreaded benchmark using Intel Threading Building Blocks with parallel_for() (based on "single")
"x87" - uses double precision x87 arithmetic, not vectorised

The "tbb" binary takes a number as argument, which is the number of tasks. #threads <= #tasks, depending #cores. e.g. "./tbb 4" will run 4 threads on 8 cores and 2 threads on 2 cores.

________
There are set of options, which one can choose in Makefile, in KF_APPROACH variable

KF algorithm:
fit is used by default
-DDAF - use Determenistic Anneling Filter

Extrapolation a method:

-DANALYTIC_EXTRAPOLATION - flexible extrapolation formula via Tailor expansion
-DRK4_EXTRAPOLATION - 4th order Runge-Kutta extrapolation

KF approach:
-DCONVENTIONAL - use conventional approach

-DSQUARE_ROOT1S - use Potter Square root method, with a square matrix
-DSQUARE_ROOT1 - use Potter Square root method, with a triangular matrix
-DSQUARE_ROOT2 - use Carlson Square root method (with a triangular matrix)

-DUD_FILTERING_S - use U-D filtering with single measurement filtration (via diagonalization)
-DUD_FILTERING_V - use U-D filtering with vector measurement filtration (only for DAF, fitter will use single measurement in any case)