When some function is called from multiple points in a program, all the inputs are merged and the result is therefore same for all these function calls. Simple parallelization would be to execute all these calculation for every function call separately. This would not only increase precision, but also can be executed in parallel, since it is basically execution of the same instruction over different input data.
When some function is called from multiple points in a program, all the inputs are merged and the result is therefore same for all these function calls. Simple parallelization would be to execute all these calculation for every function call separately. This would not only increase precision, but also can be executed in parallel, since it is basically execution of the same instruction over different input data.