Commit ffebd2f996 for ffmpeg
commit ffebd2f996347ec9035f4ffea23b830aa2c277ea
Author: Ramiro Polla <ramiro.polla@gmail.com>
Date: Wed Nov 4 21:44:49 2026 +0100
swscale/ppc: remove the AltiVec float yuv2plane1
The vector loop treats the signed 19-bit intermediates as unsigned, so
negative values are clamped to 65535 (1.0f) instead of 0. This has gone
unnoticed because nothing produces negative values for float gray outputs
yet. The next commit enables range expansion of limited range input for
these outputs, which does, and the output would no longer match the C
code.
Remove it and let ppc use the C implementation.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Ramiro Polla <ramiro.polla@gmail.com>
diff --git a/libswscale/ppc/swscale_altivec.c b/libswscale/ppc/swscale_altivec.c
index 0eb155bde0..7ac91b5d29 100644
--- a/libswscale/ppc/swscale_altivec.c
+++ b/libswscale/ppc/swscale_altivec.c
@@ -27,7 +27,6 @@
#include "libswscale/swscale.h"
#include "libswscale/swscale_internal.h"
#include "libavutil/attributes.h"
-#include "libavutil/intfloat.h"
#include "yuv2rgb_altivec.h"
#include "libavutil/ppc/util_altivec.h"
@@ -106,131 +105,11 @@
#endif /* HAVE_BIGENDIAN */
-#define SHIFT 3
-
-#define get_pixel(val, bias, signedness) \
- (bias + av_clip_ ## signedness ## 16(val >> shift))
-
-static void
-yuv2plane1_float_u(const int32_t *src, float *dest, int dstW, int start)
-{
- static const int shift = 3;
- static const float float_mult = 1.0f / 65535.0f;
- int i, val;
- uint16_t val_uint;
-
- for (i = start; i < dstW; ++i){
- val = src[i] + (1 << (shift - 1));
- val_uint = get_pixel(val, 0, uint);
- dest[i] = float_mult * (float)val_uint;
- }
-}
-
-static void
-yuv2plane1_float_bswap_u(const int32_t *src, uint32_t *dest, int dstW, int start)
-{
- static const int shift = 3;
- static const float float_mult = 1.0f / 65535.0f;
- int i, val;
- uint16_t val_uint;
-
- for (i = start; i < dstW; ++i){
- val = src[i] + (1 << (shift - 1));
- val_uint = get_pixel(val, 0, uint);
- dest[i] = av_bswap32(av_float2int(float_mult * (float)val_uint));
- }
-}
-
-static void yuv2plane1_float_altivec(const int32_t *src, float *dest, int dstW)
-{
- const int dst_u = -(uintptr_t)dest & 3;
- const int add = (1 << (SHIFT - 1));
- const int clip = (1 << 16) - 1;
- const float fmult = 1.0f / 65535.0f;
- const vec_u32 vadd = (vec_u32) {add, add, add, add};
- const vec_u32 vshift = (vec_u32) vec_splat_u32(SHIFT);
- const vec_u32 vlargest = (vec_u32) {clip, clip, clip, clip};
- const vec_f vmul = (vec_f) {fmult, fmult, fmult, fmult};
- const vec_f vzero = (vec_f) {0, 0, 0, 0};
- vec_u32 v;
- vec_f vd;
- int i;
-
- yuv2plane1_float_u(src, dest, dst_u, 0);
-
- for (i = dst_u; i < dstW - 3; i += 4) {
- v = vec_ld(0, (const uint32_t *) &src[i]);
- v = vec_add(v, vadd);
- v = vec_sr(v, vshift);
- v = vec_min(v, vlargest);
-
- vd = vec_ctf(v, 0);
- vd = vec_madd(vd, vmul, vzero);
-
- vec_st(vd, 0, &dest[i]);
- }
-
- yuv2plane1_float_u(src, dest, dstW, i);
-}
-
-static void yuv2plane1_float_bswap_altivec(const int32_t *src, uint32_t *dest, int dstW)
-{
- const int dst_u = -(uintptr_t)dest & 3;
- const int add = (1 << (SHIFT - 1));
- const int clip = (1 << 16) - 1;
- const float fmult = 1.0f / 65535.0f;
- const vec_u32 vadd = (vec_u32) {add, add, add, add};
- const vec_u32 vshift = (vec_u32) vec_splat_u32(SHIFT);
- const vec_u32 vlargest = (vec_u32) {clip, clip, clip, clip};
- const vec_f vmul = (vec_f) {fmult, fmult, fmult, fmult};
- const vec_f vzero = (vec_f) {0, 0, 0, 0};
- const vec_u32 vswapbig = (vec_u32) {16, 16, 16, 16};
- const vec_u16 vswapsmall = vec_splat_u16(8);
- vec_u32 v;
- vec_f vd;
- int i;
-
- yuv2plane1_float_bswap_u(src, dest, dst_u, 0);
-
- for (i = dst_u; i < dstW - 3; i += 4) {
- v = vec_ld(0, (const uint32_t *) &src[i]);
- v = vec_add(v, vadd);
- v = vec_sr(v, vshift);
- v = vec_min(v, vlargest);
-
- vd = vec_ctf(v, 0);
- vd = vec_madd(vd, vmul, vzero);
-
- vd = (vec_f) vec_rl((vec_u32) vd, vswapbig);
- vd = (vec_f) vec_rl((vec_u16) vd, vswapsmall);
-
- vec_st(vd, 0, (float *) &dest[i]);
- }
-
- yuv2plane1_float_bswap_u(src, dest, dstW, i);
-}
-
-#define yuv2plane1_float(template, dest_type, BE_LE) \
-static void yuv2plane1_float ## BE_LE ## _altivec(const int16_t *src, uint8_t *dest, \
- int dstW, \
- const uint8_t *dither, int offset) \
-{ \
- template((const int32_t *)src, (dest_type *)dest, dstW); \
-}
-
-#if HAVE_BIGENDIAN
-yuv2plane1_float(yuv2plane1_float_altivec, float, BE)
-yuv2plane1_float(yuv2plane1_float_bswap_altivec, uint32_t, LE)
-#else
-yuv2plane1_float(yuv2plane1_float_altivec, float, LE)
-yuv2plane1_float(yuv2plane1_float_bswap_altivec, uint32_t, BE)
-#endif
-
av_cold void ff_sws_init_swscale_altivec(SwsInternal *c)
{
+#if HAVE_BIGENDIAN
enum AVPixelFormat dstFormat = c->opts.dst_format;
-#if HAVE_BIGENDIAN
if (c->srcBpc == 8 && c->dstBpc <= 14) {
c->hyScale = c->hcScale = hScale_real_altivec;
}
@@ -241,12 +120,6 @@ av_cold void ff_sws_init_swscale_altivec(SwsInternal *c)
}
#endif
- if (dstFormat == AV_PIX_FMT_GRAYF32BE) {
- c->yuv2plane1 = yuv2plane1_floatBE_altivec;
- } else if (dstFormat == AV_PIX_FMT_GRAYF32LE) {
- c->yuv2plane1 = yuv2plane1_floatLE_altivec;
- }
-
/* The following list of supported dstFormat values should
* match what's found in the body of ff_yuv2packedX_altivec() */
if (!(c->opts.flags & (SWS_BITEXACT | SWS_FULL_CHR_H_INT)) && !c->needAlpha) {