aaron/ispc
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Add foreach and foreach_tiled looping constructs

Browse Source 
These make it easier to iterate over arbitrary amounts of data
elements; specifically, they automatically handle the "ragged
extra bits" that come up when the number of elements to be
processed isn't evenly divided by programCount.

TODO: documentation
This commit is contained in:
Matt Pharr
2011-11-30 13:17:31 -08:00
parent b48775a549
commit 8bc7367109
32 changed files with 1120 additions and 78 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
examples/volume_rendering/volume.ispc
View File

@@ -310,11 +310,7 @@ volume_tile(uniform int x0, uniform int y0, uniform int x1,
// by 4.
for (uniform int y = y0; y < y1; y += 4) {
for (uniform int x = x0; x < x1; x += 4) {
// For each such tile, process programCount pixels at a time,
// until we've done all 16 of them. Thus, we're also assuming
// that programCount <= 16 and that 16 is evenly dividible by
// programCount.
for (uniform int o = 0; o < 16; o += programCount) {
foreach (o = 0 ... 16) {
// These two arrays encode the mapping from [0,15] to
// offsets within the 4x4 pixel block so that we render
// each pixel inside the block
@@ -324,8 +320,7 @@ volume_tile(uniform int x0, uniform int y0, uniform int x1,
2, 2, 3, 3, 2, 2, 3, 3 };
// Figure out the pixel to render for this program instance
int xo = x + xoffsets[o + programIndex];
int yo = y + yoffsets[o + programIndex];
int xo = x + xoffsets[o], yo = y + yoffsets[o];
// Use viewing parameters to compute the corresponding ray
// for the pixel