Skip to content

Speed up some conversions #9807

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
akx wants to merge 2 commits into
base: main
Choose a base branch
from
+27 −37
Open

Speed up some conversions #9807

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
ed54845
Speed up i2l with branchless operations so the loop autovectorizes
akx Jul 21, 2026
b31fed7
Speed up palettized-to-RGB* conversions
akx Jul 21, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
64 changes: 27 additions & 37 deletions src/libImaging/Convert.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -587,17 +587,11 @@ l2i(UINT8 *out_, const UINT8 *in, int xsize) {

static void
i2l(UINT8 *out, const UINT8 *in_, int xsize) {
int x;
for (x = 0; x < xsize; x++, out++, in_ += 4) {
for (int x = 0; x < xsize; x++, out++, in_ += 4) {
INT32 v;
memcpy(&v, in_, sizeof(v));
if (v <= 0) {
*out = 0;
} else if (v >= 255) {
*out = 255;
} else {
*out = (UINT8)v;
}
// Branchless saturation
*out = (UINT8)(v <= 0 ? 0 : (v >= 255 ? 255 : v));
}
}

Expand Down Expand Up @@ -980,27 +974,30 @@ pa2f(UINT8 *out_, const UINT8 *in, int xsize, ImagingPalette palette) {
}
}

// Set the alpha channel of the UINT32 `v` in-place to the given value.
#ifdef WORDS_BIGENDIAN
#define SET_ALPHA_32(v, alpha) v = ((v & 0xFFFFFF00u) | (alpha))

@radarhere radarhere Aug 20, 2026

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
#define SET_ALPHA_32(v, alpha) v = ((v & 0xFFFFFF00u) | (alpha))
#define SET_ALPHA_32(v, alpha) v = ((v & 0xFFFFFF00u) | alpha)

@akx akx Aug 21, 2026 •
edited
Loading

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Same here - defensive parentheses if alpha expands to an expression.

EDIT: Prior art in e.g. the B16/L16/S16 macros in _imaging.c.

#else
#define SET_ALPHA_32(v, alpha) v = ((v & 0x00FFFFFFu) | ((UINT32)(alpha) << 24))

@radarhere radarhere Aug 20, 2026

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
#define SET_ALPHA_32(v, alpha) v = ((v & 0x00FFFFFFu) | ((UINT32)(alpha) << 24))
#define SET_ALPHA_32(v, alpha) v = ((v & 0x00FFFFFFu) | ((UINT32)alpha << 24))

@akx akx Aug 21, 2026 •
edited
Loading

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Rather not, because alpha could be a more complex expression.

EDIT: Prior art in e.g. the B16/L16/S16 macros in _imaging.c.

#endif

static void
p2rgb(UINT8 *out, const UINT8 *in, int xsize, ImagingPalette palette) {
int x;
for (x = 0; x < xsize; x++) {
const UINT8 *rgb = &palette->palette[*in++ * 4];
*out++ = rgb[0];
*out++ = rgb[1];
*out++ = rgb[2];
*out++ = 255;
for (int x = 0; x < xsize; x++, in++, out += 4) {
UINT32 v;
memcpy(&v, &palette->palette[in[0] * 4], sizeof(v));
SET_ALPHA_32(v, 0xFF);
memcpy(out, &v, sizeof(v));
}
}

static void
pa2rgb(UINT8 *out, const UINT8 *in, int xsize, ImagingPalette palette) {
int x;
for (x = 0; x < xsize; x++, in += 4) {
const UINT8 *rgb = &palette->palette[in[0] * 4];
*out++ = rgb[0];
*out++ = rgb[1];
*out++ = rgb[2];
*out++ = 255;
for (int x = 0; x < xsize; x++, in += 4, out += 4) {
UINT32 v;
memcpy(&v, &palette->palette[in[0] * 4], sizeof(v));
SET_ALPHA_32(v, 0xFF);
memcpy(out, &v, sizeof(v));
}
}

Expand All @@ -1026,25 +1023,18 @@ pa2hsv(UINT8 *out, const UINT8 *in, int xsize, ImagingPalette palette) {

static void
p2rgba(UINT8 *out, const UINT8 *in, int xsize, ImagingPalette palette) {
int x;
for (x = 0; x < xsize; x++) {
const UINT8 *rgba = &palette->palette[*in++ * 4];
*out++ = rgba[0];
*out++ = rgba[1];
*out++ = rgba[2];
*out++ = rgba[3];
for (int x = 0; x < xsize; x++, in++, out += 4) {
memcpy(out, &palette->palette[in[0] * 4], 4);
}
}

static void
pa2rgba(UINT8 *out, const UINT8 *in, int xsize, ImagingPalette palette) {
int x;
for (x = 0; x < xsize; x++, in += 4) {
const UINT8 *rgb = &palette->palette[in[0] * 4];
*out++ = rgb[0];
*out++ = rgb[1];
*out++ = rgb[2];
*out++ = in[3];
for (int x = 0; x < xsize; x++, in += 4, out += 4) {
UINT32 v;
memcpy(&v, &palette->palette[in[0] * 4], sizeof(v));
SET_ALPHA_32(v, in[3]);
memcpy(out, &v, sizeof(v));
}
}

Expand Down
Loading