Manual clamp

README.md

@@ -76,9 +76,9 @@ Tests performed on the image 5763x3842
 
 |                        | time(NEON) | Time(AVX) |
 |------------------------|:----------:|:---------:|
-| utils RGB->YUV 4:2:0   |   4.09ms   |  6.14ms   |
+| utils RGB->YUV 4:2:0   |   3.48ms   |  6.14ms   |
 | libyuv RGB->YUV 4:2:0  |   3.58ms   |  33.87ms  |
-| utils RGBA->YUV 4:2:0  |   4.88ms   |  7.34ms   |
+| utils RGBA->YUV 4:2:0  |   4.32ms   |  7.34ms   |
 | libyuv RGBA->YUV 4:2:0 |   4.87ms   |  23.48ms  |
 | utils RGBA->YUV 4:2:2  |   4.83ms   |  7.08ms   |
 | libyuv RGBA->YUV 4:2:2 |   5.90ms   |  35.23ms  |

app/src/main.rs

@@ -95,11 +95,11 @@ fn main() {
     // let mut bytes_16: Vec<u16> = src_bytes.iter().map(|&x| (x as u16) << 4).collect();
 
     let start_time = Instant::now();
-    rgb_to_yuv_nv12(
-        &mut bi_planar_image,
+    rgb_to_yuv420(
+        &mut planar_image,
         &src_bytes,
         rgba_stride as u32,
-        YuvRange::Full,
+        YuvRange::Limited,
         YuvStandardMatrix::Bt601,
     )
     .unwrap();
@@ -253,11 +253,11 @@ fn main() {
     // let rgba_stride = width as usize * 4;
     // let mut rgba = vec![0u8; height as usize * rgba_stride];
 
-    yuv_nv12_to_rgb(
-        &fixed_biplanar,
+    yuv420_to_rgb(
+        &fixed_planar,
         &mut rgba,
         rgba_stride as u32,
-        YuvRange::Full,
+        YuvRange::Limited,
         YuvStandardMatrix::Bt601,
     )
     .unwrap();

src/neon/mod.rs

@@ -36,6 +36,7 @@ mod rgb_to_ycgco_r;
 mod rgb_to_yuv_p16;
 mod rgba_to_nv;
 mod rgba_to_yuv;
+mod rgba_to_yuv420;
 mod y_p16_to_rgba16;
 mod y_to_rgb;
 mod ycgco_to_rgb;
@@ -62,6 +63,7 @@ pub(crate) use rgb_to_ycgco_r::neon_rgb_to_ycgcor_row;
 pub(crate) use rgb_to_yuv_p16::{neon_rgba_to_yuv_p16, neon_rgba_to_yuv_p16_rdm};
 pub(crate) use rgba_to_nv::{neon_rgbx_to_nv_row, neon_rgbx_to_nv_row_rdm};
 pub(crate) use rgba_to_yuv::{neon_rgba_to_yuv, neon_rgba_to_yuv_rdm};
+pub(crate) use rgba_to_yuv420::{neon_rgba_to_yuv420, neon_rgba_to_yuv_rdm420};
 pub(crate) use y_p16_to_rgba16::neon_y_p16_to_rgba16_row;
 pub(crate) use y_to_rgb::{neon_y_to_rgb_row, neon_y_to_rgb_row_rdm};
 pub(crate) use ycgco_to_rgb::neon_ycgco_to_rgb_row;

src/neon/neon_simd_support.rs

@@ -83,14 +83,6 @@ pub(crate) unsafe fn vweight_laneq_x2<const LANE0: i32, const LANE1: i32>(
     (lo, hi)
 }
 
-#[inline(always)]
-pub(crate) unsafe fn vmullq_s16(v: int16x8_t, q: int16x8_t) -> (int32x4_t, int32x4_t) {
-    (
-        vmull_s16(vget_low_s16(v), vget_low_s16(q)),
-        vmull_high_s16(v, q),
-    )
-}
-
 #[inline(always)]
 pub(crate) unsafe fn vmullq_laneq_s16<const LANE: i32>(
     v: int16x8_t,

src/neon/rgba_to_yuv.rs

@@ -48,7 +48,6 @@ pub(crate) unsafe fn neon_rgba_to_yuv_rdm<
     start_cx: usize,
     start_ux: usize,
     width: usize,
-    compute_uv_row: bool,
 ) -> ProcessedOffset {
     let chroma_subsampling: YuvChromaSubsampling = SAMPLING.into();
     let source_channels: YuvSourceChannels = ORIGIN_CHANNELS.into();
@@ -192,7 +191,7 @@ pub(crate) unsafe fn neon_rgba_to_yuv_rdm<
             vst1q_u8(v_ptr.get_unchecked_mut(ux..).as_mut_ptr(), cr);
 
             ux += 16;
-        } else if (chroma_subsampling == YuvChromaSubsampling::Yuv420 && compute_uv_row)
+        } else if (chroma_subsampling == YuvChromaSubsampling::Yuv420)
             || (chroma_subsampling == YuvChromaSubsampling::Yuv422)
         {
             let r1 = vreinterpretq_s16_u16(vshlq_n_u16::<V_SCALE>(vrshrq_n_u16::<1>(vpaddlq_u8(
@@ -250,7 +249,6 @@ pub(crate) unsafe fn neon_rgba_to_yuv<
     start_cx: usize,
     start_ux: usize,
     width: usize,
-    compute_uv_row: bool,
 ) -> ProcessedOffset {
     let chroma_subsampling: YuvChromaSubsampling = SAMPLING.into();
     let source_channels: YuvSourceChannels = ORIGIN_CHANNELS.into();
@@ -460,7 +458,7 @@ pub(crate) unsafe fn neon_rgba_to_yuv<
 
             ux += 16;
         } else if chroma_subsampling == YuvChromaSubsampling::Yuv422
-            || (chroma_subsampling == YuvChromaSubsampling::Yuv420 && compute_uv_row)
+            || (chroma_subsampling == YuvChromaSubsampling::Yuv420)
         {
             let r1 = vreinterpretq_s16_u16(vrshrq_n_u16::<1>(vpaddlq_u8(r_values_u8)));
             let g1 = vreinterpretq_s16_u16(vrshrq_n_u16::<1>(vpaddlq_u8(g_values_u8)));