flow 1.3.0
Kokkos cut-cell IBM incompressible Navier-Stokes solver + pnm pore extraction
Loading...
Searching...
No Matches
bench_bottom_factor.cpp File Reference

Timing probe of the direct pressure bottom's factor launch (mg_bottom_factor, doc/vof_step_performance_design.md §13.4.3, §14 H-1) on the bottom-direct test problems: walls-y (bottom 16x12x8, walls on y: P = 12, b = 128 – the bubble column's bottom) and periodic (16x12x8 all periodic: P = 16, b = 96, with the border). More...

#include <algorithm>
#include <cmath>
#include <cstdio>
#include <cstdlib>
#include <cstring>
#include <Kokkos_Core.hpp>
#include <string>
#include <type_traits>
#include <vector>
#include "mac_cutcell_mg.hpp"

Go to the source code of this file.

Functions

int main (int argc, char **argv)
 

Detailed Description

Timing probe of the direct pressure bottom's factor launch (mg_bottom_factor, doc/vof_step_performance_design.md §13.4.3, §14 H-1) on the bottom-direct test problems: walls-y (bottom 16x12x8, walls on y: P = 12, b = 128 – the bubble column's bottom) and periodic (16x12x8 all periodic: P = 16, b = 96, with the border).

Not a correctness gate.

bench_bottom_factor [reps] [case: walls-y | periodic] [dump file] [T ...]

Per team size T (default 1 2 4 8 16 24; 0 = the production default): min / median ms over reps refactors of the same operator, and whether the factor's storage (Q, Y, e, s, stat) is bitwise identical to the first run's. A host backend times its host schedule, then the team algorithm (the device one, the host schedule's bitwise oracle) at the same sizes. With a dump file the first T's storage is written there (raw bytes, Q|Y|e|s|stat), so two builds can be compared with cmp.

Definition in file bench_bottom_factor.cpp.

Function Documentation

◆ main()