Merge pull request #46 from awxkee/dev

Reworking f16

.github/workflows/build_push.yml

@@ -23,14 +23,14 @@ jobs:
       - uses: actions/checkout@v4
       - uses: actions-rust-lang/setup-rust-toolchain@v1
       - run: rustup target add aarch64-unknown-linux-gnu x86_64-unknown-linux-gnu i686-unknown-linux-gnu powerpc-unknown-linux-gnu riscv64gc-unknown-linux-gnu
-      - run: RUSTFLAGS="-C target-feature=+neon,-fp16" cargo build --target aarch64-unknown-linux-gnu --features half
-      - run: RUSTFLAGS="-C target-feature=+neon,+fp16" cargo build --target aarch64-unknown-linux-gnu --features half
+      - run: RUSTFLAGS="-C target-feature=+neon,-fp16" cargo build --target aarch64-unknown-linux-gnu
+      - run: RUSTFLAGS="-C target-feature=+neon,+fp16" cargo +nightly build --target aarch64-unknown-linux-gnu --features nightly_f16
       - run: RUSTFLAGS="-C target-feature=+sse4.1" cargo build --target i686-unknown-linux-gnu
       - run: cargo build --target powerpc-unknown-linux-gnu
       - run: cargo build --target riscv64gc-unknown-linux-gnu
       - run: RUSTFLAGS="-C target-feature=+sse4.1" cargo build --target x86_64-unknown-linux-gnu
-      - run: RUSTFLAGS="-C target-feature=+sse4.1,+f16c" cargo build --features half --target x86_64-unknown-linux-gnu
-      - run: RUSTFLAGS="-C target-feature=+avx2,+f16c" cargo build --features half --target x86_64-unknown-linux-gnu
+      - run: RUSTFLAGS="-C target-feature=+sse4.1,+f16c" cargo +nightly build --features nightly_f16 --target x86_64-unknown-linux-gnu
+      - run: RUSTFLAGS="-C target-feature=+avx2,+f16c" cargo +nightly build --features nightly_f16 --target x86_64-unknown-linux-gnu
       - run: RUSTFLAGS="-C target-feature=+avx2" cargo build --target x86_64-unknown-linux-gnu
 
   clippy:
@@ -42,7 +42,18 @@ jobs:
     steps:
       - uses: actions/checkout@v4
       - uses: dtolnay/rust-toolchain@stable
-      - run: cargo clippy
+      - run: cargo clippy -- -D warnings
+
+  clippy_nightly:
+    name: Clippy Nightly
+    strategy:
+      matrix:
+        os: [ ubuntu-latest, macos-latest ]
+    runs-on: ${{ matrix.os }}
+    steps:
+      - uses: actions/checkout@v4
+      - uses: dtolnay/rust-toolchain@stable
+      - run: cargo clippy --all-features -- -D warnings
 
   tests:
     name: Testing
@@ -71,6 +82,18 @@ jobs:
       - run: cargo fuzz run resize_plane -- -max_total_time=30
       - run: cargo fuzz run colorspaces -- -max_total_time=10
 
+  fuzz_rgba_8bit_no_rdm:
+    name: Fuzzing 8bit wo RDM
+    runs-on: macos-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: dtolnay/rust-toolchain@nightly
+      - run: cargo install cargo-fuzz
+      - run: cargo fuzz run resize_rgba --no-default-features -- -max_total_time=30
+      - run: cargo fuzz run resize_rgb --no-default-features -- -max_total_time=30
+      - run: cargo fuzz run resize_cbcr8 --no-default-features -- -max_total_time=30
+      - run: cargo fuzz run resize_plane --no-default-features -- -max_total_time=30
+
   fuzz_rgba_high_bit:
     name: Fuzzing High bit-depth
     strategy:
@@ -82,6 +105,7 @@ jobs:
       - uses: dtolnay/rust-toolchain@nightly
       - run: cargo install cargo-fuzz
       - run: cargo fuzz run resize_rgba_u16 -- -max_total_time=30
+      - run: cargo fuzz run resize_rgba_f16 -- -max_total_time=30
       - run: cargo fuzz run resize_rgb_u16 -- -max_total_time=30
       - run: cargo fuzz run resize_cbcr16 -- -max_total_time=30
       - run: cargo fuzz run resize_plane_u16 -- -max_total_time=30

Cargo.toml

@@ -2,7 +2,7 @@ workspace = { members = ["app", "wasm", "fuzz", "app/accelerate"], exclude = ["p
 
 [package]
 name = "pic-scale"
-version = "0.4.2"
+version = "0.5.0"
 edition = "2021"
 description = "High performance image scaling"
 readme = "README.md"
@@ -18,15 +18,16 @@ rust-version = "1.82.0"
 
 [dependencies]
 colorutils-rs = {version = "0.7.4", optional = true}
-half = { version = "2.4.1", optional = true, features = ["alloc", "std", "num-traits"] }
 num-traits = { version = "0.2.19", features = ["std"] }
 rayon = "1.10.0"
 
 [target.'cfg(any(target_os = "macos", target_os = "ios"))'.dependencies]
 libc = "0.2.158"
 
 [features]
-default = ["colorspaces"]
+default = ["colorspaces", "rdm"]
 colorspaces = ["dep:colorutils-rs"]
 nightly_avx512 = []
-nightly_avx512fp16 = ["nightly_avx512"]
+nightly_avx512fp16 = ["nightly_avx512"]
+nightly_f16 = []
+rdm = []

app/Cargo.toml

@@ -5,8 +5,7 @@ edition = "2021"
 
 [dependencies]
 image = { version = "0.25.5", features = ["default"] }
-#image = { path= "../../../RustroverProjects/image", features = ["default", "avif", "avif-native"] }
-pic-scale = { path = "..", features = ["half"], default-features = true }
+pic-scale = { path = "..", features = ["nightly_f16"], default-features = false }
 fast_image_resize = { version = "5.0.0", features = [] }
 half = { version = "2.4.1", default-features = true }
 accelerate = {path = "accelerate/"}
@@ -16,7 +15,7 @@ libc = "0.2.169"
 criterion = "0.5.1"
 #image = { version = "0.25.2", features = ["default"] }
 fast_image_resize = { version = "5.0.0", features = [] }
-pic-scale = { path = "..", features = ["half"] }
+pic-scale = { path = "..", features = ["nightly_f16", "rdm"], default-features = false }
 
 [[bench]]
 name = "resize_rgb"

app/accelerate/src/lib.rs

@@ -42,6 +42,15 @@ mod accelerate {
             flags: libc::c_uint,
         ) -> libc::c_int;
 
+        #[allow(non_camel_case_types)]
+        #[allow(non_snake_case)]
+        pub fn vImageScale_XRGB2101010W(
+            src: *const vImage_Buffer,
+            dest: *mut vImage_Buffer,
+            temp_buffer: *mut libc::c_void,
+            flags: libc::c_uint,
+        ) -> libc::c_int;
+
         #[allow(non_camel_case_types)]
         #[allow(non_snake_case)]
         pub fn vImageScale_ARGBFFFF(

app/benches/resize_rgb/main.rs

@@ -5,7 +5,7 @@ use fast_image_resize::{CpuExtensions, PixelType, ResizeAlg, ResizeOptions, Resi
 use image::{EncodableLayout, GenericImageView, ImageReader};
 use pic_scale::{
     ImageStore, ImageStoreMut, ResamplingFunction, Scaler, Scaling, ScalingF32, ScalingU16,
-    ThreadingPolicy,
+    ThreadingPolicy, WorkloadStrategy,
 };
 
 pub fn criterion_benchmark(c: &mut Criterion) {
@@ -25,6 +25,22 @@ pub fn criterion_benchmark(c: &mut Criterion) {
         b.iter(|| {
             let mut scaler = Scaler::new(ResamplingFunction::Lanczos3);
             scaler.set_threading_policy(ThreadingPolicy::Single);
+            scaler.set_workload_strategy(WorkloadStrategy::PreferSpeed);
+            let mut target =
+                ImageStoreMut::alloc(dimensions.0 as usize / 4, dimensions.1 as usize / 4);
+            scaler.resize_rgb(&store, &mut target).unwrap();
+        })
+    });
+
+    c.bench_function("Pic scale RGB: Lanczos 3/Quality", |b| {
+        let copied: Vec<u8> = Vec::from(src_bytes);
+        let store =
+            ImageStore::<u8, 3>::from_slice(&copied, dimensions.0 as usize, dimensions.1 as usize)
+                .unwrap();
+        b.iter(|| {
+            let mut scaler = Scaler::new(ResamplingFunction::Lanczos3);
+            scaler.set_threading_policy(ThreadingPolicy::Single);
+            scaler.set_workload_strategy(WorkloadStrategy::PreferQuality);
             let mut target =
                 ImageStoreMut::alloc(dimensions.0 as usize / 4, dimensions.1 as usize / 4);
             scaler.resize_rgb(&store, &mut target).unwrap();