diff --git a/.github/workflows/simba-ci-build.yml b/.github/workflows/simba-ci-build.yml
index 6dc3ae1..a50892c 100644
--- a/.github/workflows/simba-ci-build.yml
+++ b/.github/workflows/simba-ci-build.yml
@@ -13,7 +13,7 @@ jobs:
   fmt:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@v4
       - name: Check formatting
         run: cargo fmt -- --check
   clippy:
@@ -21,7 +21,7 @@ jobs:
     env:
       RUSTFLAGS: -D warnings
     steps:
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@v4
       - name: Install latest nightly
         uses: actions-rs/toolchain@v1
         with:
@@ -33,7 +33,7 @@ jobs:
   build-native:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@v4
       - name: Install latest nightly
         uses: actions-rs/toolchain@v1
         with:
@@ -52,22 +52,24 @@ jobs:
   build-wasm:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@v4
       - run: rustup target add wasm32-unknown-unknown
       - name: build
         run: cargo build --verbose --target wasm32-unknown-unknown;
   build-no-std:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v2
-      - name: Install latest nightly
-        uses: actions-rs/toolchain@v1
+      - uses: actions/checkout@v4
+      - name: Install latest stable
+        uses: dtolnay/rust-toolchain@master
         with:
-          toolchain: nightly
-          override: true
-      - name: install xargo
-        run: cp .github/Xargo.toml .; rustup component add rust-src; cargo install -f xargo;
-      - name: build x86_64-unknown-linux-gnu
-        run: xargo build --verbose --no-default-features --target=x86_64-unknown-linux-gnu;
-      - name: build x86_64-unknown-linux-gnu --features libm
-        run: xargo build --verbose --no-default-features --features libm --target=x86_64-unknown-linux-gnu;
\ No newline at end of file
+          toolchain: stable
+          targets: "x86_64-unknown-none,thumbv7em-none-eabihf"
+      - name: build x86_64-unknown-none
+        run: cargo build --verbose --no-default-features --target=x86_64-unknown-none
+      - name: build x86_64-unknown-none --features libm
+        run: cargo build --verbose --no-default-features --features libm --target=x86_64-unknown-none
+      - name: build thumbv7em-none-eabihf
+        run: cargo build --verbose --no-default-features --target=thumbv7em-none-eabihf
+      - name: build thumbv7em-none-eabihf --features libm
+        run: cargo build --verbose --no-default-features --features libm --target=thumbv7em-none-eabihf
diff --git a/CHANGELOG b/CHANGELOG
index e5e007e..9c44e2f 100644
--- a/CHANGELOG
+++ b/CHANGELOG
@@ -1,3 +1,9 @@
+## Release v0.9.1 (05 Sept. 2025)
+- `into_arr`, `from_arr`, `map`, `zip_map` are now public on `WideF32x4`, `WideF32x8` and `WideF64x4`.
+- `from_arr` and `into_arr` are now public on `WideBoolF32x4`, `WideBoolF32x8`, `WideBoolF64x4`.
+- Use SIMD implementations of `WideBool*` methods `all`, `any`, and `none`.
+- Inline more `AutoSimd` methods.
+
 ## Release v0.9.0 (22 June 2023)
 - The `cuda` feature has been removed, as the toolchain it depends on is long abandoned.
 - The `packed_simd` feature has been removes as it has been incompatible with the `nightly` compiler for several months.
diff --git a/Cargo.toml b/Cargo.toml
index 164035e..01b2d5e 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "simba"
-version = "0.9.0"
+version = "0.9.1"
 authors = ["sebcrozet <developer@crozet.re>"]
 description = "SIMD algebra for Rust"
 keywords = ["algebra", "simd", "math"]
diff --git a/src/simd/auto_simd_impl.rs b/src/simd/auto_simd_impl.rs
index 9a48ce8..9526d08 100644
--- a/src/simd/auto_simd_impl.rs
+++ b/src/simd/auto_simd_impl.rs
@@ -68,6 +68,7 @@ macro_rules! impl_bool_simd (
             pub const ZERO: Self = AutoSimd([false; $lanes]);
             pub const ONE: Self = AutoSimd([true; $lanes]);
 
+            #[inline(always)]
             pub fn new($($i: bool),*) -> Self {
                 AutoSimd([$($i),*])
             }
@@ -83,7 +84,7 @@ macro_rules! impl_bool_simd (
         impl Not for AutoSimd<$t> {
             type Output = Self;
 
-            #[inline]
+            #[inline(always)]
             fn not(self) -> Self {
                 self.map(|x| !x)
             }
@@ -91,6 +92,8 @@ macro_rules! impl_bool_simd (
 
         impl BitAnd<AutoSimd<$t>> for AutoSimd<$t> {
             type Output = Self;
+
+            #[inline(always)]
             fn bitand(self, rhs: Self) -> Self {
                 self.zip_map(rhs, |x, y| x & y)
             }
@@ -98,6 +101,8 @@ macro_rules! impl_bool_simd (
 
         impl BitOr<AutoSimd<$t>> for AutoSimd<$t> {
             type Output = Self;
+
+            #[inline(always)]
             fn bitor(self, rhs: Self) -> Self {
                 self.zip_map(rhs, |x, y| x | y)
             }
@@ -105,6 +110,8 @@ macro_rules! impl_bool_simd (
 
         impl BitXor<AutoSimd<$t>> for AutoSimd<$t> {
             type Output = Self;
+
+            #[inline(always)]
             fn bitxor(self, rhs: Self) -> Self {
                 self.zip_map(rhs, |x, y| x ^ y)
             }
diff --git a/src/simd/wide_simd_impl.rs b/src/simd/wide_simd_impl.rs
index ead5e24..26ec21d 100644
--- a/src/simd/wide_simd_impl.rs
+++ b/src/simd/wide_simd_impl.rs
@@ -119,23 +119,23 @@ macro_rules! impl_wide_f32 (
             pub const ONE: Self = $WideF32xX(<wide::$f32xX>::ONE);
 
             #[inline(always)]
-            fn into_arr(self) -> [$f32; $lanes] {
+            pub fn into_arr(self) -> [$f32; $lanes] {
                 self.0.into()
             }
 
             #[inline(always)]
-            fn from_arr(arr: [$f32; $lanes]) -> Self {
+            pub fn from_arr(arr: [$f32; $lanes]) -> Self {
                 Self(arr.into())
             }
 
             #[inline(always)]
-            fn map(self, f: impl Fn($f32) -> $f32) -> Self {
+            pub fn map(self, f: impl Fn($f32) -> $f32) -> Self {
                 let arr = self.into_arr();
                 Self::from([f(arr[0]), $(f(arr[$ii])),+])
             }
 
             #[inline(always)]
-            fn zip_map(self, rhs: Self, f: impl Fn($f32, $f32) -> $f32) -> Self {
+            pub fn zip_map(self, rhs: Self, f: impl Fn($f32, $f32) -> $f32) -> Self {
                 let arr = self.into_arr();
                 let rhs = rhs.into_arr();
                 Self::from([
@@ -146,11 +146,13 @@ macro_rules! impl_wide_f32 (
         }
 
         impl $WideBoolF32xX {
-            fn from_arr(arr: [$f32; $lanes]) -> Self {
+            #[inline(always)]
+            pub fn from_arr(arr: [$f32; $lanes]) -> Self {
                 Self(arr.into())
             }
 
-            fn into_arr(self) -> [$f32; $lanes] {
+            #[inline(always)]
+            pub fn into_arr(self) -> [$f32; $lanes] {
                 self.0.into()
             }
         }
@@ -319,17 +321,17 @@ macro_rules! impl_wide_f32 (
 
             #[inline(always)]
             fn all(self) -> bool {
-                self == Self(!wide::$f32xX::ZERO)
+                self.0.all()
             }
 
             #[inline(always)]
             fn any(self) -> bool {
-                self != Self(wide::$f32xX::ZERO)
+                self.0.any()
             }
 
             #[inline(always)]
             fn none(self) -> bool {
-                self == Self(wide::$f32xX::ZERO)
+                self.0.none()
             }
 
             #[inline(always)]
@@ -1027,7 +1029,8 @@ macro_rules! impl_wide_f32 (
 
             #[inline(always)]
             fn simd_sin_cos(self) -> (Self, Self) {
-                (self.simd_sin(), self.simd_cos())
+                let (sin, cos) = self.0.sin_cos();
+                ($WideF32xX(sin), $WideF32xX(cos))
             }
 
             //            #[inline(always]