more optimization attempts

bfps/cpp/Lagrange_polys.hpp

@@ -28,14 +28,14 @@
 
 #define LAGRANGE_POLYS
 
-void beta_Lagrange_n1(int deriv, double x, double *poly_val);
-void beta_Lagrange_n2(int deriv, double x, double *poly_val);
-void beta_Lagrange_n3(int deriv, double x, double *poly_val);
-void beta_Lagrange_n4(int deriv, double x, double *poly_val);
-void beta_Lagrange_n5(int deriv, double x, double *poly_val);
-void beta_Lagrange_n6(int deriv, double x, double *poly_val);
-void beta_Lagrange_n7(int deriv, double x, double *poly_val);
-void beta_Lagrange_n8(int deriv, double x, double *poly_val);
+void beta_Lagrange_n1(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_Lagrange_n2(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_Lagrange_n3(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_Lagrange_n4(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_Lagrange_n5(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_Lagrange_n6(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_Lagrange_n7(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_Lagrange_n8(const int deriv, const double x, double *__restrict__ poly_val);
 
 #endif//LAGRANGE_POLYS
 

bfps/cpp/interpolator.hpp

@@ -40,9 +40,9 @@
 #define INTERPOLATOR
 
 typedef void (*base_polynomial_values)(
-        int derivative,
-        double fraction,
-        double *destination);
+        const int derivative,
+        const double fraction,
+        double *__restrict__ destination);
 
 template <class rnumber, int interp_neighbours>
 class interpolator
@@ -59,7 +59,7 @@ class interpolator
                 base_polynomial_values BETA_POLYS);
         ~interpolator();
 
-        void operator()(double t, int *xg, double *xx, double *dest, int *deriv = NULL);
+        void operator()(double t, int *__restrict__ xg, double *__restrict__ xx, double *__restrict__ dest, int *deriv = NULL);
         /* destroys input */
         int read_rFFTW(void *src);
 };

bfps/cpp/particles.hpp

@@ -107,19 +107,19 @@ class particles
         /* an Euler step is needed to compute an estimate of future positions,
          * which is needed for synchronization.
          * */
-        void jump_estimate(double *jump_length);
-        void get_rhs(double *x, double *rhs);
-        void get_rhs(double t, double *x, double *rhs);
+        void jump_estimate(double *__restrict__ jump_length);
+        void get_rhs(double *__restrict__ x, double *__restrict__ rhs);
+        void get_rhs(double t, double *__restrict__ x, double *__restrict__ rhs);
 
         int get_rank(double z); // get rank for given value of z
         void synchronize();
-        void synchronize_single_particle_state(int p, double *x, int source_id = -1);
-        void get_grid_coordinates(double *x, int *xg, double *xx);
+        void synchronize_single_particle_state(int p, double *__restrict__ x, int source_id = -1);
+        void get_grid_coordinates(double *__restrict__ x, int *__restrict__ xg, double *__restrict__ xx);
         void sample_vec_field(
             interpolator<rnumber, interp_neighbours> *vec,
             double t,
-            double *x,
-            double *y,
+            double *__restrict__ x,
+            double *__restrict__ y,
             const bool synch = false,
             int *deriv = NULL);
         inline void sample_vec_field(interpolator<rnumber, interp_neighbours> *field, double *vec_values)

bfps/cpp/spline_n1.hpp

@@ -28,9 +28,9 @@
 
 #define SPLINE_N1
 
-void beta_n1_m0(int deriv, double x, double *poly_val);
-void beta_n1_m1(int deriv, double x, double *poly_val);
-void beta_n1_m2(int deriv, double x, double *poly_val);
+void beta_n1_m0(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n1_m1(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n1_m2(const int deriv, const double x, double *__restrict__ poly_val);
 
 #endif//SPLINE_N1
 

bfps/cpp/spline_n2.hpp

@@ -28,11 +28,11 @@
 
 #define SPLINE_N2
 
-void beta_n2_m0(int deriv, double x, double *poly_val);
-void beta_n2_m1(int deriv, double x, double *poly_val);
-void beta_n2_m2(int deriv, double x, double *poly_val);
-void beta_n2_m3(int deriv, double x, double *poly_val);
-void beta_n2_m4(int deriv, double x, double *poly_val);
+void beta_n2_m0(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n2_m1(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n2_m2(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n2_m3(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n2_m4(const int deriv, const double x, double *__restrict__ poly_val);
 
 #endif//SPLINE_N2
 

bfps/cpp/spline_n3.hpp

@@ -28,13 +28,13 @@
 
 #define SPLINE_N3
 
-void beta_n3_m0(int deriv, double x, double *poly_val);
-void beta_n3_m1(int deriv, double x, double *poly_val);
-void beta_n3_m2(int deriv, double x, double *poly_val);
-void beta_n3_m3(int deriv, double x, double *poly_val);
-void beta_n3_m4(int deriv, double x, double *poly_val);
-void beta_n3_m5(int deriv, double x, double *poly_val);
-void beta_n3_m6(int deriv, double x, double *poly_val);
+void beta_n3_m0(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n3_m1(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n3_m2(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n3_m3(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n3_m4(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n3_m5(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n3_m6(const int deriv, const double x, double *__restrict__ poly_val);
 
 #endif//SPLINE_N3
 

bfps/cpp/spline_n4.hpp

@@ -28,15 +28,15 @@
 
 #define SPLINE_N4
 
-void beta_n4_m0(int deriv, double x, double *poly_val);
-void beta_n4_m1(int deriv, double x, double *poly_val);
-void beta_n4_m2(int deriv, double x, double *poly_val);
-void beta_n4_m3(int deriv, double x, double *poly_val);
-void beta_n4_m4(int deriv, double x, double *poly_val);
-void beta_n4_m5(int deriv, double x, double *poly_val);
-void beta_n4_m6(int deriv, double x, double *poly_val);
-void beta_n4_m7(int deriv, double x, double *poly_val);
-void beta_n4_m8(int deriv, double x, double *poly_val);
+void beta_n4_m0(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n4_m1(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n4_m2(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n4_m3(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n4_m4(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n4_m5(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n4_m6(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n4_m7(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n4_m8(const int deriv, const double x, double *__restrict__ poly_val);
 
 #endif//SPLINE_N4
 

bfps/cpp/spline_n5.hpp

@@ -28,17 +28,17 @@
 
 #define SPLINE_N5
 
-void beta_n5_m0(int deriv, double x, double *poly_val);
-void beta_n5_m1(int deriv, double x, double *poly_val);
-void beta_n5_m2(int deriv, double x, double *poly_val);
-void beta_n5_m3(int deriv, double x, double *poly_val);
-void beta_n5_m4(int deriv, double x, double *poly_val);
-void beta_n5_m5(int deriv, double x, double *poly_val);
-void beta_n5_m6(int deriv, double x, double *poly_val);
-void beta_n5_m7(int deriv, double x, double *poly_val);
-void beta_n5_m8(int deriv, double x, double *poly_val);
-void beta_n5_m9(int deriv, double x, double *poly_val);
-void beta_n5_m10(int deriv, double x, double *poly_val);
+void beta_n5_m0( const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n5_m1( const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n5_m2( const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n5_m3( const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n5_m4( const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n5_m5( const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n5_m6( const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n5_m7( const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n5_m8( const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n5_m9( const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n5_m10(const int deriv, const double x, double *__restrict__ poly_val);
 
 #endif//SPLINE_N5
 

bfps/cpp/spline_n6.hpp

@@ -28,19 +28,19 @@
 
 #define SPLINE_N6
 
-void beta_n6_m0(int deriv, double x, double *poly_val);
-void beta_n6_m1(int deriv, double x, double *poly_val);
-void beta_n6_m2(int deriv, double x, double *poly_val);
-void beta_n6_m3(int deriv, double x, double *poly_val);
-void beta_n6_m4(int deriv, double x, double *poly_val);
-void beta_n6_m5(int deriv, double x, double *poly_val);
-void beta_n6_m6(int deriv, double x, double *poly_val);
-void beta_n6_m7(int deriv, double x, double *poly_val);
-void beta_n6_m8(int deriv, double x, double *poly_val);
-void beta_n6_m9(int deriv, double x, double *poly_val);
-void beta_n6_m10(int deriv, double x, double *poly_val);
-void beta_n6_m11(int deriv, double x, double *poly_val);
-void beta_n6_m12(int deriv, double x, double *poly_val);
+void beta_n6_m0( const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n6_m1( const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n6_m2( const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n6_m3( const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n6_m4( const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n6_m5( const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n6_m6( const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n6_m7( const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n6_m8( const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n6_m9( const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n6_m10(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n6_m11(const int deriv, const double x, double *__restrict__ poly_val);
+void beta_n6_m12(const int deriv, const double x, double *__restrict__ poly_val);
 
 #endif//SPLINE_N6
 

tox_plain.ini

@@ -12,10 +12,10 @@ commands =
     #python tests/test_plain.py -n 256 --run --initialize --ncpu 8 --niter_todo 8 --precision single --wd "data/single"
     python tests/test_plain.py \
         -n 128 --run --initialize --ncpu 4 \
-        --nparticles 0 --niter_todo 24 \
+        --nparticles 100 --niter_todo 24 \
         --precision single --wd "data/single" \
-        --kMeta 3.0 \
-        --multiplejob
+        --kMeta 3.0 #\
+        #--multiplejob
     #python tests/test_plain.py -n 32 --run --initialize --multiplejob --ncpu 2 --nparticles 16 --niter_todo 64 --precision single --wd "data/single"
     #python tests/test_plain.py -n 32 --run --initialize --multiplejob --ncpu 2 --nparticles 16 --niter_todo 64 --precision double --wd "data/double"