From 67f40ab195520b78934da2127434c0da304b02e1 Mon Sep 17 00:00:00 2001 From: anapt Date: Tue, 6 Feb 2018 16:21:34 +0200 Subject: [PATCH] dimensions fix --- mean_shift_cuda/meanshift_gpu_utils.cu | 2 +- mean_shift_cuda/meanshift_kernels.cu | 12 ++++++++++-- 2 files changed, 11 insertions(+), 3 deletions(-) diff --git a/mean_shift_cuda/meanshift_gpu_utils.cu b/mean_shift_cuda/meanshift_gpu_utils.cu index 555709c..af9454c 100644 --- a/mean_shift_cuda/meanshift_gpu_utils.cu +++ b/mean_shift_cuda/meanshift_gpu_utils.cu @@ -361,7 +361,7 @@ void calculate_norm(Matrix d_mean_shift_vector, double *current_norm){ dim3 dimGrid; do { dimBlock.x = requested_block_size; - dimBlock.y = d_mean_shift_vector.width; + dimBlock.y = 1; dimGrid.x = (d_mean_shift_vector.height + dimBlock.x - 1) / dimBlock.x; dimGrid.y = 1; diff --git a/mean_shift_cuda/meanshift_kernels.cu b/mean_shift_cuda/meanshift_kernels.cu index f0d1c7b..dfb6496 100644 --- a/mean_shift_cuda/meanshift_kernels.cu +++ b/mean_shift_cuda/meanshift_kernels.cu @@ -86,13 +86,21 @@ __global__ void norm(Matrix mean_shift_vector, double *current_norm) { // by accumulating results into cell_value double cell_value = 0; int row = blockIdx.x * blockDim.x + threadIdx.x; - int col = blockIdx.y * blockDim.y + threadIdx.y; // performs calculations only if thread's indexes are within matrix bounds - if (row * mean_shift_vector.width + col >= mean_shift_vector.width * mean_shift_vector.height){ + if (row >= denominator.height){ return; } +// for (int column = 0; column < kernel_matrix.width; ++column){ +// cell_value += kernel_matrix.elements[row * kernel_matrix.width + column]; +// } + denominator.elements[row] = cell_value; + // performs calculations only if thread's indexes are within matrix bounds +// if (row * mean_shift_vector.width + col >= mean_shift_vector.width * mean_shift_vector.height){ +// return; +// } + for (int element_index = 0; element_index < mean_shift_vector.width; ++element_index){ cell_value += mean_shift_vector.elements[row * mean_shift_vector.width + element_index] * mean_shift_vector.elements[row * mean_shift_vector.width + element_index];