better benchmarking

src/liblbm_f32.fut

@@ -10,6 +10,19 @@ module sim_d3q19_reg_bfl_f32 = mk_sim d3q19_reg_bfl_f32
 
 entry main_d3q19_reg = sim_d3q19_reg_bfl_f32.run
 
+-- Test the sub1 function.
+-- ==
+-- input { 21i64 21i64 21i64 27i64 3i64 1.5f32 1000i64 } 
+-- auto output
+let main (nx: i64) (ny: i64) (nz: i64) (q: i64) (d: i64) (omega: f32) (max_t: i64): [nx][ny][nz][q]f32 = 
+    let fin = replicate nx (replicate ny (replicate nz (replicate q 0f32)))
+    let dist = replicate nx (replicate ny (replicate nz (replicate q (-1.0f32))))
+    let u = replicate nx (replicate ny (replicate nz (replicate d 0f32)))
+
+    in sim_d3q27_reg_bfl_f32.run fin u dist omega max_t
+
+
+
 
 -- -- executes Lattice-Boltzmann it times and
 -- let lattice_boltzmann [nx][ny][nz] (fin: [nx][ny][nz](lbm_model.vec_q real)) (vel: [nx][ny][nz](lbm_model.vec_d real)) (dists: [nx][ny][nz](lbm_model.vec_q real)) (omega: real) (max_t: i64):  [nx][ny][nz](lbm_model.vec_q real) =
@@ -41,13 +54,3 @@ entry main_d3q19_reg = sim_d3q19_reg_bfl_f32.run
 --     in (lbm_model.to_array_q f, rho, (lbm_model.to_array_d u))
 
 
--- Testing main
--- ==
--- entry: test_main
--- compiled input { 213 123 0.8 1 }
--- auto output
-
--- entry test_main (nx: i64)(ny: i64)(tau: real)(it: i64): [9][]real =
--- 	let f_ini = init_f_in(nx)(ny)
--- 	let f = lattice_boltzmann(nx)(ny)(f_ini)(tau)(it)
---     in from_tuple_to_array(f)