binary support

.gitignore

@@ -53,3 +53,4 @@ docs/_build/
 
 
 *.out
 *.out
 *.full
 *.full
+*.npy
\ No newline at end of file

epgpu.cu

@@ -26,39 +26,6 @@ __device__ inline vec3 cross(const vec3& v1, const vec3& v2) {
 
 
 __device__ inline vec3 operator*(double a, const vec3& v) { return vec3(v.x * a, v.y * a, v.z * a); }
 __device__ inline vec3 operator*(double a, const vec3& v) { return vec3(v.x * a, v.y * a, v.z * a); }
 
 
-//---------------------------
-struct mat3 { // row-major matrix 4x4
-//---------------------------
-	vec3 rows[4];
-public:
-	__device__ mat3() {}
-	__device__ mat3(double m00, double m01, double m02,
-		double m10, double m11, double m12,
-		double m20, double m21, double m22,
-		double m30, double m31, double m32) {
-		rows[0][0] = m00; rows[0][1] = m01; rows[0][2] = m02; 
-		rows[1][0] = m10; rows[1][1] = m11; rows[1][2] = m12; 
-		rows[2][0] = m20; rows[2][1] = m21; rows[2][2] = m22; 
-		rows[3][0] = m30; rows[3][1] = m31; rows[3][2] = m32;
-	}
-	__device__ mat3(vec3 it, vec3 jt, vec3 kt) {
-		rows[0] = it; rows[1] = jt; rows[2] = kt;
-	}
-
-	__device__ vec3& operator[](int i) { return rows[i]; }
-	__device__ vec3 operator[](int i) const { return rows[i]; }
-	__device__ operator double*() const { return (double*)this; }
-};
-
-__device__ inline vec3 operator*(const vec3& v, const mat3& mat) {
-	return v.x * mat[0] + v.y * mat[1] + v.z * mat[2];
-}
-
-__device__ inline mat3 operator*(const mat3& left, const mat3& right) {
-	mat3 result;
-	for (int i = 0; i < 4; i++) result.rows[i] = left.rows[i] * right;
-	return result;
-}
 
 
 __device__ inline double signeddistance(const vec3& planeNN, const vec3& planeP, const vec3& Q) {
 __device__ inline double signeddistance(const vec3& planeNN, const vec3& planeP, const vec3& Q) {
     vec3 PQ = Q - planeP;
     vec3 PQ = Q - planeP;

tetrarun.py

 import sys, getopt
 import sys, getopt
 import numpy as np
 import numpy as np
+import cupy as cp
+import os
 from genax import gen_angels_to_pick, angles_alap, angles_ratet
 from genax import gen_angels_to_pick, angles_alap, angles_ratet
 from gpu import start_kernel
 from gpu import start_kernel
 from utils import convert, printresults, search, exact_one, exact_one_gpu, filter_gpu, writetofile, writetofile2
 from utils import convert, printresults, search, exact_one, exact_one_gpu, filter_gpu, writetofile, writetofile2
@@ -11,9 +13,10 @@ def main(argv):
    w = 3
    w = 3
    PLOT = False
    PLOT = False
    mpi = False
    mpi = False
+   binary = False
 
 
    try:
    try:
-      opts, args = getopt.getopt(argv,"hpn:v:w:d:o:m:")
+      opts, args = getopt.getopt(argv,"hpn:v:w:d:o:mb")
    except getopt.GetoptError as err:
    except getopt.GetoptError as err:
       print(err)
       print(err)
       print ('tetrarun.py -n <range division:int> -v <> -w <> -o <outputfile>')
       print ('tetrarun.py -n <range division:int> -v <> -w <> -o <outputfile>')
@@ -36,6 +39,8 @@ def main(argv):
          PLOT = True
          PLOT = True
       elif opt in ("-m", "--mpi"):
       elif opt in ("-m", "--mpi"):
          mpi = True
          mpi = True
+      elif opt in ("-b", "--bin"):
+         binary = True
 
 
    space = gen_angels_to_pick(n)
    space = gen_angels_to_pick(n)
 
 
@@ -43,9 +48,7 @@ def main(argv):
 
 
    Dx, Dy, Dz = angles_ratet(space, mpi)
    Dx, Dy, Dz = angles_ratet(space, mpi)
 
 
-   del space
-
-   res = start_kernel(Cx, Cy, Dx, Dy, Dz, v, w)
+   if outputfile and not binary:
 
 
       Cx_cpu = np.zeros(Cx.size, dtype=np.float64)
       Cx_cpu = np.zeros(Cx.size, dtype=np.float64)
       Cx.get(out=Cx_cpu)
       Cx.get(out=Cx_cpu)
@@ -57,13 +60,27 @@ def main(argv):
       Dy.get(out=Dy_cpu)
       Dy.get(out=Dy_cpu)
       Dz_cpu = np.zeros(Dz.size, dtype=np.float64)
       Dz_cpu = np.zeros(Dz.size, dtype=np.float64)
       Dz.get(out=Dz_cpu)
       Dz.get(out=Dz_cpu)
+
+      del space
+
+      res = start_kernel(Cx, Cy, Dx, Dy, Dz, v, w)
+
       mtx_cpu = np.zeros(res.size, dtype=np.int8).reshape(res.shape)
       mtx_cpu = np.zeros(res.size, dtype=np.int8).reshape(res.shape)
       res.get(out=mtx_cpu)
       res.get(out=mtx_cpu)
    
    
-   if outputfile:
       writetofile(outputfile+'.full', Cx_cpu, Cy_cpu, Dx_cpu, Dy_cpu, Dz_cpu, mtx_cpu, mpi)
       writetofile(outputfile+'.full', Cx_cpu, Cy_cpu, Dx_cpu, Dy_cpu, Dz_cpu, mtx_cpu, mpi)
       writetofile2(outputfile, Cx_cpu, Cy_cpu, Dx_cpu, Dy_cpu, Dz_cpu, mtx_cpu, mpi)
       writetofile2(outputfile, Cx_cpu, Cy_cpu, Dx_cpu, Dy_cpu, Dz_cpu, mtx_cpu, mpi)
 
 
+   if outputfile and binary:
+      res = start_kernel(Cx, Cy, Dx, Dy, Dz, v, w)
+      os.mkdir(outputfile)
+      cp.save(outputfile+'/Cx.npy', Cx)
+      cp.save(outputfile+'/Cy.npy', Cy)
+      cp.save(outputfile+'/Dx.npy', Dx)
+      cp.save(outputfile+'/Dy.npy', Dy)
+      cp.save(outputfile+'/Dz.npy', Dz)
+      cp.save(outputfile+'/mtx.npy', res)
+
    #printresults(res)
    #printresults(res)
    #print("Exact one 3-3")
    #print("Exact one 3-3")
    #exact_one_gpu(res, 3-1, 3-1)
    #exact_one_gpu(res, 3-1, 3-1)