merge metal features

.github/workflows/ci-macos.yml

@@ -0,0 +1,49 @@
+name: ci
+on:
+  pull_request:
+  push:
+    branches:
+      - macos
+jobs:
+  format:
+    name: cargo fmt
+    runs-on: macos-latest
+    steps:
+      - uses: actions/checkout@v5
+
+      - name: Setup toolchain
+        uses: actions-rust-lang/setup-rust-toolchain@v1
+
+      - name: Run fmt
+        run: make fmt
+
+  cargo-clippy:
+    name: cargo clippy
+    runs-on: macos-latest
+    steps:
+      - uses: actions/checkout@v5
+
+       # This action will configure Rust cache automatically:
+       # https://github.com/actions-rust-lang/setup-rust-toolchain?tab=readme-ov-file#inputs
+      - name: Setup toolchain
+        uses: actions-rust-lang/setup-rust-toolchain@v1
+
+      - run: make lint
+
+  build-and-test:
+    name: build and test
+    runs-on: macos-latest
+    steps:
+      - uses: actions/checkout@v5
+
+      - name: Setup toolchain
+        uses: actions-rust-lang/setup-rust-toolchain@v1
+
+      - name: Build
+        run: make build
+      
+      - name: Build with metal
+        run: cargo build --features metal
+
+#      - name: Run Unit Test
+#        run: make test

Cargo.lock

@@ -379,6 +379,12 @@ version = "2.6.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "6099cdc01846bc367c4e7dd630dc5966dccf36b652fae7a74e17b640411a91b2"
 
+[[package]]
+name = "block"
+version = "0.1.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0d8c1fef690941d3e7788d328517591fecc684c084084702d6ff1641e993699a"
+
 [[package]]
 name = "built"
 version = "0.7.7"
@@ -440,10 +446,12 @@ checksum = "a9f51e2ecf6efe9737af8f993433c839f956d2b6ed4fd2dd4a7c6d8b0fa667ff"
 dependencies = [
  "byteorder",
  "candle-kernels",
+ "candle-metal-kernels",
  "cudarc",
  "gemm 0.17.1",
  "half",
  "memmap2",
+ "metal 0.27.0",
  "num-traits",
  "num_cpus",
  "rand 0.9.2",
@@ -453,6 +461,7 @@ dependencies = [
  "thiserror 1.0.69",
  "ug",
  "ug-cuda",
+ "ug-metal",
  "yoke 0.7.5",
  "zip",
 ]
@@ -478,6 +487,19 @@ dependencies = [
  "bindgen_cuda",
 ]
 
+[[package]]
+name = "candle-metal-kernels"
+version = "0.9.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9a323ee9c813707f73b6e59300661b354a70410f31fe4135170c4eda8a061534"
+dependencies = [
+ "half",
+ "metal 0.27.0",
+ "once_cell",
+ "thiserror 1.0.69",
+ "tracing",
+]
+
 [[package]]
 name = "candle-nn"
 version = "0.9.1"
@@ -485,7 +507,9 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "c1980d53280c8f9e2c6cbe1785855d7ff8010208b46e21252b978badf13ad69d"
 dependencies = [
  "candle-core",
+ "candle-metal-kernels",
  "half",
+ "metal 0.27.0",
  "num-traits",
  "rayon",
  "safetensors",
@@ -702,6 +726,17 @@ version = "0.8.7"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "773648b94d0e5d620f64f280777445740e61fe701025087ec8b57f45c791888b"
 
+[[package]]
+name = "core-graphics-types"
+version = "0.1.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "45390e6114f68f718cc7a830514a96f903cccd70d02a8f6d9f643ac4ba45afaf"
+dependencies = [
+ "bitflags 1.3.2",
+ "core-foundation",
+ "libc",
+]
+
 [[package]]
 name = "crc32fast"
 version = "1.5.0"
@@ -1161,7 +1196,28 @@ version = "0.3.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "f6f339eb8adc052cd2ca78910fda869aefa38d22d5cb648e6485e4d3fc06f3b1"
 dependencies = [
- "foreign-types-shared",
+ "foreign-types-shared 0.1.1",
+]
+
+[[package]]
+name = "foreign-types"
+version = "0.5.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d737d9aa519fb7b749cbc3b962edcf310a8dd1f4b67c91c4f83975dbdd17d965"
+dependencies = [
+ "foreign-types-macros",
+ "foreign-types-shared 0.3.1",
+]
+
+[[package]]
+name = "foreign-types-macros"
+version = "0.2.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1a5c6c585bc94aaf2c7b51dd4c2ba22680844aba4c687be581871a6f518c5742"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
 ]
 
 [[package]]
@@ -1170,6 +1226,12 @@ version = "0.1.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "00b0228411908ca8685dba7fc2cdd70ec9990a6e753e89b6ac91a84c40fbaf4b"
 
+[[package]]
+name = "foreign-types-shared"
+version = "0.3.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "aa9a19cbb55df58761df49b23516a86d432839add4af60fc256da840f66ed35b"
+
 [[package]]
 name = "form_urlencoded"
 version = "1.2.2"
@@ -2257,6 +2319,15 @@ version = "0.2.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "670fdfda89751bc4a84ac13eaa63e205cf0fd22b4c9a5fbfa085b63c1f1d3a30"
 
+[[package]]
+name = "malloc_buf"
+version = "0.0.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "62bb907fe88d54d8d9ce32a3cceab4218ed2f6b7d35617cafe9adf84e43919cb"
+dependencies = [
+ "libc",
+]
+
 [[package]]
 name = "matchers"
 version = "0.2.0"
@@ -2292,6 +2363,36 @@ dependencies = [
  "stable_deref_trait",
 ]
 
+[[package]]
+name = "metal"
+version = "0.27.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c43f73953f8cbe511f021b58f18c3ce1c3d1ae13fe953293e13345bf83217f25"
+dependencies = [
+ "bitflags 2.10.0",
+ "block",
+ "core-graphics-types",
+ "foreign-types 0.5.0",
+ "log",
+ "objc",
+ "paste",
+]
+
+[[package]]
+name = "metal"
+version = "0.29.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7ecfd3296f8c56b7c1f6fbac3c71cefa9d78ce009850c45000015f206dc7fa21"
+dependencies = [
+ "bitflags 2.10.0",
+ "block",
+ "core-graphics-types",
+ "foreign-types 0.5.0",
+ "log",
+ "objc",
+ "paste",
+]
+
 [[package]]
 name = "mime"
 version = "0.3.17"
@@ -2598,6 +2699,25 @@ version = "0.4.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "830b246a0e5f20af87141b25c173cd1b609bd7779a4617d6ec582abaf90870f3"
 
+[[package]]
+name = "objc"
+version = "0.2.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "915b1b472bc21c53464d6c8461c9d3af805ba1ef837e1cac254428f4a77177b1"
+dependencies = [
+ "malloc_buf",
+ "objc_exception",
+]
+
+[[package]]
+name = "objc_exception"
+version = "0.1.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ad970fb455818ad6cba4c122ad012fae53ae8b4795f86378bce65e4f6bab2ca4"
+dependencies = [
+ "cc",
+]
+
 [[package]]
 name = "object"
 version = "0.37.3"
@@ -2649,7 +2769,7 @@ checksum = "24ad14dd45412269e1a30f52ad8f0664f0f4f4a89ee8fe28c3b3527021ebb654"
 dependencies = [
  "bitflags 2.10.0",
  "cfg-if",
- "foreign-types",
+ "foreign-types 0.3.2",
  "libc",
  "once_cell",
  "openssl-macros",
@@ -4438,6 +4558,20 @@ dependencies = [
  "ug",
 ]
 
+[[package]]
+name = "ug-metal"
+version = "0.4.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "76daec3c7a32a1b4a0e3307b6b057fa067aa64e750713987410a2c402e5cd731"
+dependencies = [
+ "half",
+ "metal 0.29.0",
+ "objc",
+ "serde",
+ "thiserror 1.0.69",
+ "ug",
+]
+
 [[package]]
 name = "uncased"
 version = "0.9.10"

Cargo.toml

@@ -40,6 +40,7 @@ symphonia = { version = "0.5.5", features = ["mp3", "wav"] }
 [features]
 flash-attn=["candle-flash-attn"]
 cuda=["candle-nn/cuda", "candle-core/cuda", "candle-transformers/cuda"]
+metal=["candle-nn/metal", "candle-core/metal", "candle-transformers/metal"]
 ffmpeg=["ffmpeg-next"]
 
 [lints.clippy]

src/utils/mod.rs

@@ -27,7 +27,11 @@ pub fn get_device(device: Option<&Device>) -> Device {
             {
                 Device::new_cuda(0).unwrap_or(Device::Cpu)
             }
-            #[cfg(not(feature = "cuda"))]
+            #[cfg(all(not(feature = "cuda"), feature = "metal"))]
+            {
+                Device::new_metal(0).unwrap_or(Device::Cpu)
+            }
+            #[cfg(all(not(feature = "cuda"), not(feature = "metal")))]
             {
                 Device::Cpu
             }