size

genax.py

@@ -26,6 +26,8 @@ extern "C"
 __global__ 
 void parosit(const double* x1, const double* x2, double* a, double* b, const int m, const double PI) {
     int tid = blockDim.x * blockIdx.x + threadIdx.x;
+    if (m <= tid || m*m <= tid*m+m-1)
+        return;
     float alpha = x1[tid];
     if (m*m <= tid*m+m-1)
         return;
@@ -47,9 +49,9 @@ extern "C"
 __global__ 
 void parosit2(const double* x1, const double* x2, double* a, double* b, const int m, const double PI) {
     int tid = blockDim.x * blockIdx.x + threadIdx.x;
-    float alpha = x1[tid];
-    if (m*m <= tid*m+m-1)
+    if (m <= tid || m*m <= tid*m+m-1)
         return;
+    float alpha = x1[tid];
     for (int i = 0; i < m; i++) {
         float betha = x2[i];
         if ((alpha + betha) < PI && alpha > 0.0) {

gpu.py

@@ -9,7 +9,7 @@ ep_pontok_module = cp.RawModule(code=code, options=('--std=c++11',), name_expres
 fun = ep_pontok_module.get_function(kers[0])
 
 def start_kernel(Cx, Cy, Dx, Dy, Dz, v, w):
-
+    print("Res size (byte): ", Cx.size*Dx.size*4*4)
     egyensulyi_mtx = cp.zeros((Cx.size*Dx.size, 4, 4), dtype=cp.int8)
     numBlock = int((Cx.size*Dx.size + 256 - 1) / 256)
     fun((numBlock,), (256,), (v, w, Cx, Cy, Dx, Dy, Dz, Cx.size, Dx.size, egyensulyi_mtx))