SerenityOS/serenity
1
0 Fork 0
mirror of https://github.com/SerenityOS/serenity.git synced 2026-11-07 04:29:30 -10:00
Code Issues Packages Projects Releases Wiki Activity

LibGfx/JPEG2000/IDWT: Use raw pointers instead of op[] in _1D_EXTR

Browse Source 
Takes `Build/lagom/bin/image -o out.tiff ~/Downloads/balloon.jpf` from
542.3 ms ± 7.0 ms to 527.2 ms ± 4.7 ms under hyperfine on my system
(2.7% less). When run under `/usr/bin/time -l`, the changes are:

Before:

         13956342230  instructions retired
          2322276362  cycles elapsed

After:

         13251294364  instructions retired
          2270195534  cycles elapsed

The absolute win on the lossless file is again a bit smaller
(due to smaller image dimensions), and the relative win is tiny
there (due to it spending much more time in bitplane decoding and
less in IDWT).
This commit is contained in:
Nico Weber committed 2026-07-26 10:12:58 -04:00
1 parent 83a0675316
commit 32a71fa3aa
1 file changed
+9 -3
Download Patch File Download Diff File Expand all files Collapse all files
Userland/Libraries/LibGfx/ImageFormats/JPEG2000InverseDiscreteWaveletTransform.h
+9 -3
View File
@@ -275,14 +275,20 @@ inline void _1D_EXTR(Transformation transformation, IDWTOutput& a, int start, in
return i0 + min(mod(i - i0, 2 * (i1 - i0 - 1)), 2 * (i1 - i0 - 1) - mod(i - i0, 2 * (i1 - i0 - 1)));
};
VERIFY(static_cast<int>(buffers.scanline_buffer.size()) >= buffers.scanline_runway + (i1 - i0) + 8);
float* dst = buffers.scanline_buffer.data() + buffers.scanline_runway;
VERIFY(start + (i1 - 1 - i0) * delta <= static_cast<int>(a.data.size()));
float const* src = a.data.data() + start;
// PSE is the identity function for i0 <= i < i1, so only call it on the edges.
int i = 0;
for (int l = i0 - i_left; l < i0; ++l, ++i)
buffers.scanline_buffer[buffers.scanline_runway + i] = a.data[start + (PSE(l, i0, i1) - i0) * delta];
dst[i] = src[(PSE(l, i0, i1) - i0) * delta];
for (int l = i0; l < i1; ++l, ++i)
buffers.scanline_buffer[buffers.scanline_runway + i] = a.data[start + (l - i0) * delta];
dst[i] = src[(l - i0) * delta];
for (int l = i1; l < i1 + i_right; ++l, ++i)
buffers.scanline_buffer[buffers.scanline_runway + i] = a.data[start + (PSE(l, i0, i1) - i0) * delta];
dst[i] = src[(PSE(l, i0, i1) - i0) * delta];
buffers.scanline_start = buffers.scanline_runway + i_left;
}