doc/float.txt

purpose:
=======
 increase CPUs workload - verify that results of some math functions are stable

Floating testcases are related to :
----------------------------------
- trigonometric (acos, asin, atan, atan2, cos, sin, tan),
- Euclidean distance function (hypot), 
- Computes the natural logarithm of the gamma function (lgamma),
- Functions that manipulate floating-point numbers (modf, ldexp, frexp),
- exponential and logarithmic functions (exp, log, log10),
- hyperbolic (cosh, sinh, tanh),
- Bessel (j0, j1, y0, y1),
- and power functions (ceil, fabs, floor, fmod, pow, sqrt).

The testcase is splitted into 5 sets:
-------------------------------------
 float_bessel, float_exp_log, float_iperb, float_power, float_trigo

need:
----
- all tests need data files to execute:
  - Input Files located to the same Directory as the corresponding binary 
         Are in the form <function>_inp.ref
		  Or in the form d<function>
  - Output Files located to the same Directory as the corresponding binary 
         Are in the form <function>_out.ref     
		  Or in the form r<function>

 example: acos function works with dacos file, (input datafile) and with rcos 
 file (result/output datafile).  
 please refer to float_*.c files for a full list. 

algorithm:
---------
		 /* for one set - similar for all sets */
 thread_code:
		 read input data  /* read input datafile */
		 computes data 
		 check_error /* compares results to expected results: output datafile */
 main:
		 parse args
		 /* create and start threads */
		 from 1 up to number of function 
		  from 1 up to number of thread
		 		 pthread_create/start
		  end
		 end
		 /* wait for all threads to end */
		 from 1 up to number of thread
		 		 pthread_join
		 end
		 if "ok" returns 0 else returns 1