changed copy_rows_fromd_by_idx a bit to make it clearer

1 files changed, 4 insertions, 4 deletions

diff --git a/nerv/lib/matrix/generic/cukernel.cu b/nerv/lib/matrix/generic/cukernel.cu
index 6c8e64a..d042d48 100644
--- a/nerv/lib/matrix/generic/cukernel.cu
+++ b/nerv/lib/matrix/generic/cukernel.cu
@@ -294,7 +294,7 @@ __global__ void cudak_(gen_col_idx)(MATRIX_ELEM *b,
 }
 
 __global__ void cudak_(copy_rows_by_idx)(const MATRIX_ELEM *a, MATRIX_ELEM *b,
-                                    const MATRIX_ELEM *idx, int b_begin,
+                                    const MATRIX_ELEM *idx,
                                     int nrow, int ncol, int stride) {
     int j = blockIdx.x * blockDim.x + threadIdx.x;
     int i = blockIdx.y * blockDim.y + threadIdx.y;
@@ -620,13 +620,13 @@ extern "C" {
     }
 
     void cudak_(cuda_copy_rows_by_idx)(const Matrix *a, Matrix *b,
-                                        const Matrix *idx, int b_begin) {
+                                        const Matrix *idx, int idx_begin) {
         dim3 threadsPerBlock(CUDA_THREADS_NN, 1);
         dim3 numBlocks(CEIL_DIV(b->ncol, threadsPerBlock.x), b->nrow);
         cudak_(copy_rows_by_idx)<<<numBlocks, threadsPerBlock>>> \
             (MATRIX_ELEM_PTR(a), MATRIX_ELEM_PTR(b),
-            MATRIX_ELEM_PTR(idx) + b_begin,
-             b_begin, b->nrow, b->ncol, b->stride / sizeof(MATRIX_ELEM));
+            MATRIX_ELEM_PTR(idx) + idx_begin,
+            b->nrow, b->ncol, b->stride / sizeof(MATRIX_ELEM));
         cudaStreamSynchronize(0);
     }
 }