From 1c0fd9bb4726f28b5ccf4491b8d108b00c884ec3 Mon Sep 17 00:00:00 2001
From: Joseph Redmon <pjreddie@gmail.com>
Date: Thu, 30 Oct 2014 06:26:41 +0000
Subject: [PATCH] im2col slightly faster
---
src/im2col.cl | 69 +++++++++++++++++++++++-----------
1 files changed, 47 insertions(+), 22 deletions(-)
diff --git a/src/im2col.cl b/src/im2col.cl
index 8169e1a..e00e8f5 100644
--- a/src/im2col.cl
+++ b/src/im2col.cl
@@ -1,28 +1,17 @@
-float im2col_get_pixel(__global float *im, int height, int width, int channels,
- int batch, int row, int col, int channel, int pad)
-{
- row -= pad;
- col -= pad;
- if (row < 0 || col < 0 || row >= height || col >= width) return 0;
- int index = col + width*(row + height*(channel+batch*channels));
- return im[index];
-}
-
-__kernel void im2col(__global float *data_im, int batch,
+__kernel void im2col_pad(__global float *im, int batch,
int channels, int height, int width,
- int ksize, int stride, int pad, __global float *data_col)
+ int ksize, int stride, __global float *data_col)
{
int c,h,w,b;
- int height_col = (height - ksize) / stride + 1;
- int width_col = (width - ksize) / stride + 1;
+ int height_col = 1 + (height-1) / stride;
+ int width_col = 1 + (width-1) / stride;
int channels_col = channels * ksize * ksize;
- if (pad){
- height_col = 1 + (height-1) / stride;
- width_col = 1 + (width-1) / stride;
- pad = ksize/2;
- }
+
+ int pad = ksize/2;
+
int id = get_global_id(0);
+ int col_index = id;
w = id % width_col;
id /= width_col;
h = id % height_col;
@@ -35,9 +24,45 @@
int col_size = height_col*width_col*channels_col;
int w_offset = c % ksize;
int h_offset = (c / ksize) % ksize;
- int c_im = c / ksize / ksize;
+ int im_channel = c / ksize / ksize;
+ int im_row = h_offset + h * stride - pad;
+ int im_col = w_offset + w * stride - pad;
+
+ int im_index = im_col + width*(im_row + height*(im_channel+batch*channels));
+ float val = (im_row < 0 || im_col < 0 || im_row >= height || im_col >= width) ? 0 : im[im_index];
+
+ data_col[col_index] = val;
+}
+
+__kernel void im2col_nopad(__global float *im, int batch,
+ int channels, int height, int width,
+ int ksize, int stride, __global float *data_col)
+{
+ int c,h,w,b;
+ int height_col = (height - ksize) / stride + 1;
+ int width_col = (width - ksize) / stride + 1;
+ int channels_col = channels * ksize * ksize;
+
+ int id = get_global_id(0);
+ int col_index = id;
+ w = id % width_col;
+ id /= width_col;
+ h = id % height_col;
+ id /= height_col;
+ c = id % channels_col;
+ id /= channels_col;
+ b = id % batch;
+ id /= batch;
+
+ int col_size = height_col*width_col*channels_col;
+ int w_offset = c % ksize;
+ int h_offset = (c / ksize) % ksize;
+ int im_channel = c / ksize / ksize;
int im_row = h_offset + h * stride;
int im_col = w_offset + w * stride;
- int col_index = (c * height_col + h) * width_col + w + b*col_size;
- data_col[col_index] = im2col_get_pixel(data_im, height, width, channels, b, im_row, im_col, c_im, pad);
+
+ int im_index = im_col + width*(im_row + height*(im_channel+batch*channels));
+ float val = (im_row < 0 || im_col < 0 || im_row >= height || im_col >= width) ? 0 : im[im_index];
+
+ data_col[col_index] = val;
}
--
Gitblit v1.10.0