Merge branch 'google:main' into avxifma

.bazelrc

@@ -0,0 +1,35 @@
+###############################################################################
+# Common flags that apply to all configurations.
+# Use sparingly for things common to all compilers and platforms.
+###############################################################################
+
+# Enable Bzlmod for every Bazel command
+common --enable_bzlmod
+
+# Prevent invalid caching if input files are modified during a build.
+build --experimental_guard_against_concurrent_changes
+
+###############################################################################
+# Options for continuous integration.
+###############################################################################
+
+# Speedup bazel using a ramdisk.
+build:ci --sandbox_base=/dev/shm
+
+# Show as many errors as possible.
+build:ci --keep_going
+
+# Show subcommands when building
+build:ci --subcommands=true
+
+# Make sure we test for C99 compliance when building the library
+build:ci --conlyopt=-std=c99
+
+# Show test errors.
+test:ci --test_output=errors
+
+###############################################################################
+
+# The user.bazelrc file is not checked in but available for local mods.
+# Always keep this at the end of the file so that user flags override.
+try-import %workspace%/user.bazelrc

.github/workflows/amd64_macos_bazel.yml

@@ -1,3 +1,4 @@
+# ref: https://github.com/actions/runner-images
 name: amd64 MacOS Bazel
 
 on:
@@ -10,7 +11,7 @@ on:
 jobs:
   # Building using the github runner environement directly.
   bazel:
-    runs-on: macos-latest
+    runs-on: macos-13 # Using x86 processors, ref: https://github.com/actions/runner-images
     steps:
       - name: Check out repository code
         uses: actions/checkout@v3

.github/workflows/amd64_macos_cmake.yml

@@ -1,3 +1,4 @@
+# ref: https://github.com/actions/runner-images
 name: amd64 MacOS CMake
 
 on:
@@ -10,7 +11,7 @@ on:
 jobs:
   # Building using the github runner environement directly.
   xcode:
-    runs-on: macos-latest
+    runs-on: macos-13 # Using x86 processors, ref: https://github.com/actions/runner-images
     env:
       CTEST_OUTPUT_ON_FAILURE: 1
     steps:
@@ -26,7 +27,7 @@ jobs:
     - name: Install
       run: cmake --build build --config Release --target install -v
   make:
-    runs-on: macos-latest
+    runs-on: macos-13 # Using x86 processors, ref: https://github.com/actions/runner-images
     env:
       CTEST_OUTPUT_ON_FAILURE: 1
     steps:

.github/workflows/arm64_macos_bazel.yml

@@ -0,0 +1,36 @@
+# ref: https://github.com/actions/runner-images
+name: Arm64 MacOS Bazel
+
+on:
+  push:
+  pull_request:
+  schedule:
+    # min hours day(month) month day(week)
+    - cron: '0 0 7,22 * *'
+
+jobs:
+  # Building using the github runner environement directly.
+  bazel:
+    runs-on: macos-latest # Using M1 processors, ref: https://github.com/actions/runner-images
+    steps:
+      - name: Check out repository code
+        uses: actions/checkout@v3
+      - name: Install Bazel
+        run: |
+          brew update
+          brew unlink bazelisk
+          brew install bazel
+      - name: Check Bazel
+        run: bazel version
+      - name: Build
+        run: >
+          bazel build
+          -c opt
+          --subcommands=true
+          ...
+      - name: Test
+        run: >
+          bazel test
+          -c opt
+          --test_output=errors
+          ...

.github/workflows/arm64_macos_cmake.yml

@@ -0,0 +1,44 @@
+# ref: https://github.com/actions/runner-images
+name: Arm64 MacOS CMake
+
+on:
+  push:
+  pull_request:
+  schedule:
+    # min hours day(month) month day(week)
+    - cron: '0 0 7,22 * *'
+
+jobs:
+  # Building using the github runner environement directly.
+  xcode:
+    runs-on: macos-latest # Using M1 processors, ref: https://github.com/actions/runner-images
+    env:
+      CTEST_OUTPUT_ON_FAILURE: 1
+    steps:
+    - uses: actions/checkout@v2
+    - name: Check cmake
+      run: cmake --version
+    - name: Configure
+      run: cmake -S. -Bbuild -G "Xcode" -DCMAKE_CONFIGURATION_TYPES=Release -DCMAKE_INSTALL_PREFIX:PATH=destination
+    - name: Build
+      run: cmake --build build --config Release --target ALL_BUILD -v
+    - name: Test
+      run: cmake --build build --config Release --target RUN_TESTS -v
+    - name: Install
+      run: cmake --build build --config Release --target install -v
+  make:
+    runs-on: macos-latest # Using M1 processors, ref: https://github.com/actions/runner-images
+    env:
+      CTEST_OUTPUT_ON_FAILURE: 1
+    steps:
+    - uses: actions/checkout@v2
+    - name: Check cmake
+      run: cmake --version
+    - name: Configure
+      run: cmake -S. -Bbuild -DCMAKE_BUILD_TYPE=Release -DCMAKE_INSTALL_PREFIX:PATH=destination
+    - name: Build
+      run: cmake --build build --target all -v
+    - name: Test
+      run: cmake --build build --target test -v
+    - name: Install
+      run: cmake --build build --target install -v

.gitignore

@@ -14,6 +14,7 @@ out/
 
 # Bazel artifacts
 **/bazel-*
+MODULE.bazel.lock
 
 # Per-user bazelrc files
 user.bazelrc

BUILD.bazel

@@ -14,7 +14,6 @@ exports_files(["LICENSE"])
 INCLUDES = ["include"]
 
 C99_FLAGS = [
-    "-std=c99",
     "-Wall",
     "-Wextra",
     "-Wmissing-declarations",
@@ -56,7 +55,7 @@ cc_test(
     includes = INCLUDES,
     deps = [
         ":bit_utils",
-        "@com_google_googletest//:gtest_main",
+        "@googletest//:gtest_main",
     ],
 )
 
@@ -90,7 +89,7 @@ cc_test(
     includes = INCLUDES,
     deps = [
         ":string_view",
-        "@com_google_googletest//:gtest_main",
+        "@googletest//:gtest_main",
     ],
 )
 
@@ -148,7 +147,7 @@ cc_test(
         ":cpu_features_macros",
         ":filesystem_for_testing",
         ":string_view",
-        "@com_google_googletest//:gtest_main",
+        "@googletest//:gtest_main",
     ],
 )
 
@@ -312,7 +311,7 @@ cc_library(
         PLATFORM_CPU_ARM: ["include/cpuinfo_arm.h"],
         PLATFORM_CPU_ARM64: [
             "include/cpuinfo_aarch64.h",
-            "include/internal/cpuid_aarch64.h"
+            "include/internal/cpuid_aarch64.h",
         ],
         PLATFORM_CPU_MIPS: ["include/cpuinfo_mips.h"],
         PLATFORM_CPU_PPC: ["include/cpuinfo_ppc.h"],
@@ -322,6 +321,10 @@ cc_library(
     copts = C99_FLAGS,
     defines = selects.with_or({
         PLATFORM_CPU_X86_64: ["CPU_FEATURES_MOCK_CPUID_X86"],
+        PLATFORM_CPU_ARM64: [
+            "CPU_FEATURES_MOCK_CPUID_AARCH64",
+            "CPU_FEATURES_MOCK_SYSCTL_AARCH64",
+        ],
         "//conditions:default": [],
     }) + selects.with_or({
         PLATFORM_OS_MACOS: ["HAVE_SYSCTLBYNAME"],
@@ -365,7 +368,7 @@ cc_test(
         ":filesystem_for_testing",
         ":hwcaps_for_testing",
         ":string_view",
-        "@com_google_googletest//:gtest_main",
+        "@googletest//:gtest_main",
     ],
 )
 

MODULE.bazel

@@ -0,0 +1,17 @@
+###############################################################################
+# Bazel now uses Bzlmod by default to manage external dependencies.
+# Please consider migrating your external dependencies from WORKSPACE to MODULE.bazel.
+#
+# For more details, please check https://github.com/bazelbuild/bazel/issues/18958
+###############################################################################
+
+CPU_FEATURES_VERSION = "0.9.0"
+
+module(
+    name = "cpu_features",
+    repo_name = "com_google_cpufeatures",
+    version = CPU_FEATURES_VERSION,
+)
+
+bazel_dep(name = "bazel_skylib", version = "1.7.1")
+bazel_dep(name = "googletest", version = "1.15.2")

README.md

@@ -9,7 +9,7 @@ instructions) at runtime.
 
 |  | Linux | FreeBSD | MacOS | Windows |
 | :-- | --: | --: | --: | --: |
-| amd64 | [![CMake][i1a0]][l1a0]<br/>[![Bazel][i1a1]][l1a1] | [![CMake][i2a0]][l2a0]<br/>![Bazel][d1] | [![CMake][i3a0]][l3a0]<br/>[![Bazel][i3a1]][l3a1] | [![CMake][i4a0]][l4a0]<br/>![Bazel][d1] |
+| amd64 | [![CMake][i1a0]][l1a0]<br/>[![Bazel][i1a1]][l1a1] | ![CMake][d0]<br/>![Bazel][d1] | [![CMake][i3a0]][l3a0]<br/>[![Bazel][i3a1]][l3a1] | [![CMake][i4a0]][l4a0]<br/>![Bazel][d1] |
 | AArch64 | [![CMake][i1b0]][l1b0]<br/>[![Bazel][i1b1]][l1b1] | ![CMake][d0]<br/>![Bazel][d1] | ![CMake][d0]<br/>![Bazel][d1] | ![CMake][d0]<br/>![Bazel][d1] |
 | ARM | [![CMake][i1c0]][l1c0]<br/>![Bazel][d1] | ![CMake][d0]<br/>![Bazel][d1] | ![CMake][d0]<br/>![Bazel][d1] | ![CMake][d0]<br/>![Bazel][d1] |
 | MIPS | [![CMake][i1d0]][l1d0]<br/>![Bazel][d1] | ![CMake][d0]<br/>![Bazel][d1] | ![CMake][d0]<br/>![Bazel][d1] | ![CMake][d0]<br/>![Bazel][d1] |
@@ -29,7 +29,6 @@ instructions) at runtime.
 [i1e0]: https://img.shields.io/github/actions/workflow/status/google/cpu_features/power_linux_cmake.yml?branch=main&event=push&label=&logo=cmake
 [i1f0]: https://img.shields.io/github/actions/workflow/status/google/cpu_features/riscv_linux_cmake.yml?branch=main&event=push&label=&logo=cmake
 [i1h0]: https://img.shields.io/github/actions/workflow/status/google/cpu_features/s390x_linux_cmake.yml?branch=main&event=push&label=&logo=cmake
-[i2a0]: https://img.shields.io/github/actions/workflow/status/google/cpu_features/amd64_freebsd_cmake.yml?branch=main&event=push&label=&logo=cmake
 [i3a0]: https://img.shields.io/github/actions/workflow/status/google/cpu_features/amd64_macos_cmake.yml?branch=main&event=push&label=&logo=cmake
 [i3a1]: https://img.shields.io/github/actions/workflow/status/google/cpu_features/amd64_macos_bazel.yml?branch=main&event=push&label=&logo=data:image/svg%2bxml;base64,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
 [i4a0]: https://img.shields.io/github/actions/workflow/status/google/cpu_features/amd64_windows_cmake.yml?branch=main&event=push&label=&logo=cmake
@@ -42,7 +41,6 @@ instructions) at runtime.
 [l1e0]: https://github.com/google/cpu_features/actions/workflows/power_linux_cmake.yml
 [l1f0]: https://github.com/google/cpu_features/actions/workflows/riscv_linux_cmake.yml
 [l1h0]: https://github.com/google/cpu_features/actions/workflows/s390x_linux_cmake.yml
-[l2a0]: https://github.com/google/cpu_features/actions/workflows/amd64_freebsd_cmake.yml
 [l3a0]: https://github.com/google/cpu_features/actions/workflows/amd64_macos_cmake.yml
 [l3a1]: https://github.com/google/cpu_features/actions/workflows/amd64_macos_bazel.yml
 [l4a0]: https://github.com/google/cpu_features/actions/workflows/amd64_windows_cmake.yml

WORKSPACE

@@ -1,19 +1,4 @@
-workspace(name = "com_google_cpufeatures")
-
-load("@bazel_tools//tools/build_defs/repo:git.bzl", "git_repository")
-
-git_repository(
-    name = "com_google_googletest",
-    tag = "release-1.11.0",
-    remote = "https://github.com/google/googletest.git",
-)
+# This file marks the root of the Bazel workspace.
+# See MODULE.bazel for external dependencies setup.
 
-git_repository(
-    name = "bazel_skylib",
-    tag = "1.2.0",
-    remote = "https://github.com/bazelbuild/bazel-skylib.git",
-)
-
-load("@bazel_skylib//:workspace.bzl", "bazel_skylib_workspace")
-
-bazel_skylib_workspace()
+workspace(name = "com_google_cpufeatures")

bazel/ci/docker/Dockerfile

@@ -25,13 +25,7 @@ COPY . .
 
 FROM devel AS build
 RUN bazel version
-RUN bazel build \
- -c opt \
- --subcommands=true \
- ...
+RUN bazel build --config=ci ...
 
 FROM build AS test
-RUN bazel test \
- -c opt \
- --test_output=errors \
- ...
+RUN bazel test  --config=ci ...

cmake/ci/docker/amd64/Dockerfile

@@ -6,7 +6,7 @@ LABEL maintainer="corentinl@google.com"
 ENV PATH=/usr/local/bin:$PATH
 RUN apt-get update -qq \
 && DEBIAN_FRONTEND=noninteractive apt-get install -yq git wget libssl-dev build-essential \
- ninja-build python3 pkgconf libglib2.0-dev \
+ ninja-build python3 python3-venv pkgconf libglib2.0-dev \
 && apt-get clean \
 && rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/*
 ENTRYPOINT ["/usr/bin/bash", "-c"]

cmake/ci/docker/toolchain/Dockerfile

@@ -6,7 +6,7 @@ LABEL maintainer="corentinl@google.com"
 ENV PATH=/usr/local/bin:$PATH
 RUN apt-get update -qq \
 && DEBIAN_FRONTEND=noninteractive apt-get install -yq git wget libssl-dev build-essential \
- ninja-build python3 pkgconf libglib2.0-dev \
+ ninja-build python3 python3-venv pkgconf libglib2.0-dev \
 && apt-get clean \
 && rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/*
 ENTRYPOINT ["/usr/bin/bash", "-c"]

include/cpu_features_macros.h

@@ -235,11 +235,13 @@
 #endif  // defined(CPU_FEATURES_ARCH_X86)
 
 #if defined(CPU_FEATURES_ARCH_ANY_ARM)
-#if defined(__ARM_NEON__)
+// Note: MSVC targeting ARM does not define `__ARM_NEON` but Windows on ARM
+// requires it. In that case we force NEON detection.
+#if defined(__ARM_NEON) || defined(CPU_FEATURES_COMPILER_MSC)
 #define CPU_FEATURES_COMPILED_ANY_ARM_NEON 1
 #else
 #define CPU_FEATURES_COMPILED_ANY_ARM_NEON 0
-#endif  //  defined(__ARM_NEON__)
+#endif  //  defined(__ARM_NEON) || defined(CPU_FEATURES_COMPILER_MSC)
 #endif  //  defined(CPU_FEATURES_ARCH_ANY_ARM)
 
 #if defined(CPU_FEATURES_ARCH_MIPS)

include/cpuinfo_aarch64.h

@@ -192,6 +192,31 @@ typedef struct {
   int smef16f16 : 1;   // FP16 to FP16 outer product.
   int mops : 1;        // Standardized memory operations.
   int hbc : 1;         // Hinted conditional branches.
+  int sveb16b16 : 1;   // Non-widening BFloat16 to BFloat16 arithmetic for SVE2
+                       // and SME2.
+  int lrcpc3 : 1;      // Load-Acquire RCpc instructions version 3.
+  int lse128 : 1;      // 128-bit Atomics.
+  int fpmr : 1;        // Floating-point Mode Register.
+  int lut : 1;        // Lookup table instructions with 2-bit and 4-bit indices.
+  int faminmax : 1;   // Maximum and minimum absolute value instructions.
+  int f8cvt : 1;      // FP scaling instructions and FP8 convert instructions.
+  int f8fma : 1;      // FP8 to single-precision and half-precision
+                      // multiply-accumulate instructions.
+  int f8dp4 : 1;      // FP8 to single-precision 4-way dot product FDOT (4-way)
+                      // instructions.
+  int f8dp2 : 1;      // FP8 to half-precision 2-way dot product FDOT (2-way)
+                      // instructions.
+  int f8e4m3 : 1;     // Arm FP8 E4M3 format.
+  int f8e5m2 : 1;     // Arm FP8 E5M2 format.
+  int smelutv2 : 1;   // SME2 lookup table LUTI4 and MOVT instructions.
+  int smef8f16 : 1;   // SME2 F8F16 instructions.
+  int smef8f32 : 1;   // SME2 F8F32 instructions.
+  int smesf8fma : 1;  // SVE2 FP8 to single-precision and half-precision
+                      // multiply-accumulate instructions.
+  int smesf8dp4 : 1;  // SVE2 FP8 to single-precision 4-way dot product FDOT
+                      // (4-way) instructions.
+  int smesf8dp2 : 1;  // SVE2 FP8 to half-precision 2-way dot product FDOT
+                      // (2-way) instructions.
 
   // Make sure to update Aarch64FeaturesEnum below if you add a field here.
 } Aarch64Features;
@@ -288,6 +313,24 @@ typedef enum {
   AARCH64_SME_F16F16,
   AARCH64_MOPS,
   AARCH64_HBC,
+  AARCH64_SVE_B16B16,
+  AARCH64_LRCPC3,
+  AARCH64_LSE128,
+  AARCH64_FPMR,
+  AARCH64_LUT,
+  AARCH64_FAMINMAX,
+  AARCH64_F8CVT,
+  AARCH64_F8FMA,
+  AARCH64_F8DP4,
+  AARCH64_F8DP2,
+  AARCH64_F8E4M3,
+  AARCH64_F8E5M2,
+  AARCH64_SME_LUTV2,
+  AARCH64_SME_F8F16,
+  AARCH64_SME_F8F32,
+  AARCH64_SME_SF8FMA,
+  AARCH64_SME_SF8DP4,
+  AARCH64_SME_SF8DP2,
   AARCH64_LAST_,
 } Aarch64FeaturesEnum;