print speed; the current kernel is able to render 8844 triangles with 480,000 photons in roughly 3.5 seconds

1 files changed, 7 insertions, 9 deletions

diff --git a/test.py b/test.py
index 014c2db..2ff399c 100644
--- a/test.py
+++ b/test.py
@@ -64,25 +64,23 @@ rotate(np.int32(mesh.size), mesh_gpu, np.float32(-np.pi/2), gpuarray.vec.make_fl
 
 translate(np.int32(mesh.size), mesh_gpu, gpuarray.vec.make_float3(0,30,0), block=(256,1,1), grid=(mesh.size//256+1,1))
 
-t0 = time.time()
+
 for i in range(100):
     rotate(np.int32(x.size), x_gpu, np.float32(np.pi/50), gpuarray.vec.make_float3(0,0,1), block=(256,1,1), grid=(width*height//256+1,1))
 
     rotate(np.int32(p.size), p_gpu, np.float32(np.pi/50), gpuarray.vec.make_float3(0,0,1), block=(256,1,1), grid=(width*height//256+1,1))
 
+    t0 = time.time()
     intersect(np.int32(x.size), x_gpu, p_gpu, np.int32(mesh.size//3), mesh_gpu, pixel_gpu, block=(256,1,1), grid=(width*height//256+1,1))
-
     cuda.Context.synchronize()
 
+    elapsed = time.time() - t0
+
+    print '%i triangles, %i photons, %f sec; (%f photons/s)' % \
+        (mesh.size//3, pixel.size, elapsed, pixel.size/elapsed)
+
     cuda.memcpy_dtoh(pixel, pixel_gpu)
     pygame.surfarray.blit_array(screen, pixel.reshape(size))
     pygame.display.flip()
 
-
-elapsed = time.time() - t0
-
-print '%i triangles, %i photons, %f sec; (%f photons/s)' % \
-    (mesh.size//3, pixel.size, elapsed, pixel.size/elapsed)
-
-
 raw_input('press enter to exit')