Compute Library v24.02

ARM-software · Feb 14, 2024 · bc89a0b · bc89a0b
1 parent afb5a98
commit bc89a0b
Show file tree

Hide file tree

Showing 145 changed files with 5,491 additions and 3,994 deletions.
diff --git a/Android.bp b/Android.bp
@@ -1,5 +1,5 @@
 //
-// Copyright © 2020-2023 Arm Ltd. All rights reserved.
+// Copyright © 2020-2024 Arm Ltd. All rights reserved.
 // SPDX-License-Identifier: MIT
 //
 
@@ -161,8 +161,7 @@ arm_compute_library_defaults {
             "-Wno-unused-parameter",
             "-DNO_DOT_IN_TOOLCHAIN",
             "-Wno-implicit-fallthrough",
-            "-fPIC",
-            "-DACL_INTERNAL_TEST_CKW_IN_DF"
+            "-fPIC"
     ],
     rtti: true,
 }
@@ -173,8 +172,6 @@ cc_library_static {
     proprietary: true,
     local_include_dirs: ["build/android-arm64v8a/src/core",
                          "build/android-arm64v8a/src/core/CL",
-                         "compute_kernel_writer/prototype/include",
-                         "compute_kernel_writer/prototype",
                          "src/core/common",
                          "src/core/helpers",
                          "src/core/NEON/kernels/arm_gemm",
@@ -184,15 +181,6 @@ cc_library_static {
                          "src/cpu/kernels/assembly"],
     export_include_dirs: [".", "./include"],
     srcs: [
-        "compute_kernel_writer/prototype/src/Kernel.cpp",
-        "compute_kernel_writer/prototype/src/KernelArgument.cpp",
-        "compute_kernel_writer/prototype/src/KernelWriter.cpp",
-        "compute_kernel_writer/prototype/src/OperandBase.cpp",
-        "compute_kernel_writer/prototype/src/TensorInfo.cpp",
-        "compute_kernel_writer/prototype/src/TensorOperand.cpp",
-        "compute_kernel_writer/prototype/src/TensorTileSampler.cpp",
-        "compute_kernel_writer/prototype/src/TileInfo.cpp",
-        "compute_kernel_writer/prototype/src/TileOperand.cpp",
         "src/c/AclContext.cpp",
         "src/c/AclOperator.cpp",
         "src/c/AclQueue.cpp",
@@ -624,7 +612,6 @@ cc_library_static {
         "src/cpu/operators/internal/CpuGemmAssemblyDispatch.cpp",
         "src/dynamic_fusion/runtime/gpu/cl/ClKernelRuntime.cpp",
         "src/dynamic_fusion/runtime/gpu/cl/ClWorkloadRuntime.cpp",
-        "src/dynamic_fusion/runtime/gpu/cl/ckw_driver/GpuCkwKernelArgumentsHelpers.cpp",
         "src/dynamic_fusion/sketch/attributes/CastAttributes.cpp",
         "src/dynamic_fusion/sketch/attributes/ClampAttributes.cpp",
         "src/dynamic_fusion/sketch/attributes/Conv2dAttributes.cpp",
@@ -642,20 +629,6 @@ cc_library_static {
         "src/dynamic_fusion/sketch/gpu/GpuOperatorGroup.cpp",
         "src/dynamic_fusion/sketch/gpu/GpuWorkloadContext.cpp",
         "src/dynamic_fusion/sketch/gpu/GpuWorkloadSketch.cpp",
-        "src/dynamic_fusion/sketch/gpu/ckw_driver/GpuCkwComponentArgument.cpp",
-        "src/dynamic_fusion/sketch/gpu/ckw_driver/GpuCkwDriver.cpp",
-        "src/dynamic_fusion/sketch/gpu/ckw_driver/GpuCkwKernelWriter.cpp",
-        "src/dynamic_fusion/sketch/gpu/ckw_driver/GpuCkwScopedKernelWriter.cpp",
-        "src/dynamic_fusion/sketch/gpu/ckw_driver/GpuCkwVariableTable.cpp",
-        "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwActivation.cpp",
-        "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwCast.cpp",
-        "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwDepthwiseConv2d.cpp",
-        "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwDirectConv2d.cpp",
-        "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwElementwiseBinary.cpp",
-        "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwMatMul.cpp",
-        "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwPool2d.cpp",
-        "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwResize.cpp",
-        "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwStore.cpp",
         "src/dynamic_fusion/sketch/gpu/components/cl/ClComponentActivation.cpp",
         "src/dynamic_fusion/sketch/gpu/components/cl/ClComponentCast.cpp",
         "src/dynamic_fusion/sketch/gpu/components/cl/ClComponentDepthwiseConv2d.cpp",

diff --git a/CMakeLists.txt b/CMakeLists.txt
@@ -28,7 +28,7 @@ cmake_minimum_required(VERSION 3.13 FATAL_ERROR)
 list(APPEND CMAKE_MESSAGE_CONTEXT ArmCompute)
 project(
   ArmCompute
-  VERSION 34.0.0
+  VERSION 35.0.0
   DESCRIPTION
     "The Arm Compute Library is a collection of low-level machine learning functions optimized for Arm® Cortex®-A CPU and Arm® Mali™ GPU architectures"
   LANGUAGES C CXX ASM)
@@ -58,6 +58,7 @@ endif()
 # Configuration
 
 set(CMAKE_CXX_FLAGS_DEBUG "-O0 -g -gdwarf-2 -DARM_COMPUTE_ASSERTS_ENABLED")
+set(CMAKE_CXX_FLAGS_RELEASE "-O3")
 # Default to Release Build
 if(NOT CMAKE_BUILD_TYPE AND NOT CMAKE_CONFIGURATION_TYPES)
   set(CMAKE_BUILD_TYPE
@@ -106,8 +107,7 @@ set(COMMON_CXX_FLAGS
     -Wlogical-op
     -Wnoexcept
     -Wstrict-null-sentinel
-    -Wno-misleading-indentation
-    -O3)
+    -Wno-misleading-indentation)
 
 # Disable note popups on compiler ABI changes
 if(CMAKE_CXX_COMPILER_ID STREQUAL "GNU")

diff --git a/README.md b/README.md
@@ -4,18 +4,12 @@
 > future releases. The implication of this is that the user is expected to translate NCHW models into NHWC in
 > order to benefit from the optimizations.
 
-> **⚠ Important**
-> From release 23.02: The 23.02 release introduces a change to the default tensor extend padding behavior.
-> To remain compatible with previous behavior, users will need to set the new flag `ITensorInfo::lock_paddings()` on
-> tensors for which paddings should not be extended, such as the input and output of the model that need to be mapped to
-> a camera frame or frame buffer.
-
 <br>
 <div align="center">
  <img src="https://raw.githubusercontent.com/ARM-software/ComputeLibrary/gh-pages/ACL_logo.png"/><br><br>
 </div>
 
-# Compute Library ![](https://img.shields.io/badge/latest_release-24.01-green)
+# Compute Library ![](https://img.shields.io/badge/latest_release-24.02-green)
 
 
 The Compute Library is a collection of low-level machine learning functions optimized for Arm® Cortex®-A, Arm® Neoverse® and Arm® Mali™ GPUs architectures.<br>
@@ -43,7 +37,7 @@ Key Features:
 <br>
 
 ## Documentation
-[![Documentation](https://img.shields.io/badge/documentation-24.01-green)](https://arm-software.github.io/ComputeLibrary/latest)
+[![Documentation](https://img.shields.io/badge/documentation-24.02-green)](https://arm-software.github.io/ComputeLibrary/latest)
 
 > Note: The documentation includes the reference API, changelogs, build guide, contribution guide, errata, etc.
 
@@ -56,24 +50,24 @@ All the binaries can be downloaded from [here](https://github.com/ARM-software/C
 
 | Platform       | Operating System | Release archive (Download) |
 | -------------- | ---------------- | -------------------------- |
-| Raspberry Pi 4 | Linux® 32bit      | [![](https://img.shields.io/badge/build-neon-orange)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.01/arm_compute-v24.01-bin-linux-armv7a-neon.tar.gz) |
-| Raspberry Pi 4 | Linux® 64bit      | [![](https://img.shields.io/badge/build-neon-orange)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.01/arm_compute-v24.01-bin-linux-arm64-v8a-neon.tar.gz) |
-| Odroid N2      | Linux® 64bit      | [![](https://img.shields.io/badge/build-neon-orange)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.01/arm_compute-v24.01-bin-linux-arm64-v8a-neon.tar.gz) [![](https://img.shields.io/badge/build-opencl-blue)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.01/arm_compute-v24.01-bin-linux-arm64-v8a-cl.tar.gz) [![](https://img.shields.io/badge/build-neon+cl-yellowgreen)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.01/arm_compute-v24.01-bin-linux-arm64-v8a-neon-cl.tar.gz) |
-| HiKey960       | Linux® 64bit      | [![](https://img.shields.io/badge/build-neon-orange)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.01/arm_compute-v24.01-bin-linux-arm64-v8a-neon.tar.gz) [![](https://img.shields.io/badge/build-opencl-blue)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.01/arm_compute-v24.01-bin-linux-arm64-v8a-cl.tar.gz) [![](https://img.shields.io/badge/build-neon+cl-yellowgreen)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.01/arm_compute-v24.01-bin-linux-arm64-v8a-neon-cl.tar.gz) |
+| Raspberry Pi 4 | Linux® 32bit      | [![](https://img.shields.io/badge/build-neon-orange)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.02/arm_compute-v24.02-bin-linux-armv7a-neon.tar.gz) |
+| Raspberry Pi 4 | Linux® 64bit      | [![](https://img.shields.io/badge/build-neon-orange)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.02/arm_compute-v24.02-bin-linux-arm64-v8a-neon.tar.gz) |
+| Odroid N2      | Linux® 64bit      | [![](https://img.shields.io/badge/build-neon-orange)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.02/arm_compute-v24.02-bin-linux-arm64-v8a-neon.tar.gz) [![](https://img.shields.io/badge/build-opencl-blue)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.02/arm_compute-v24.02-bin-linux-arm64-v8a-cl.tar.gz) [![](https://img.shields.io/badge/build-neon+cl-yellowgreen)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.02/arm_compute-v24.02-bin-linux-arm64-v8a-neon-cl.tar.gz) |
+| HiKey960       | Linux® 64bit      | [![](https://img.shields.io/badge/build-neon-orange)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.02/arm_compute-v24.02-bin-linux-arm64-v8a-neon.tar.gz) [![](https://img.shields.io/badge/build-opencl-blue)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.02/arm_compute-v24.02-bin-linux-arm64-v8a-cl.tar.gz) [![](https://img.shields.io/badge/build-neon+cl-yellowgreen)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.02/arm_compute-v24.02-bin-linux-arm64-v8a-neon-cl.tar.gz) |
 
 <br>
 
 | Architecture | Operating System | Release archive (Download) |
 | ------------ | ---------------- | -------------------------- |
-| armv7        | Linux®            | [![](https://img.shields.io/badge/build-neon-orange)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.01/arm_compute-v24.01-bin-linux-armv7a-neon.tar.gz) [![](https://img.shields.io/badge/build-opencl-blue)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.01/arm_compute-v24.01-bin-linux-armv7a-cl.tar.gz) [![](https://img.shields.io/badge/build-neon+cl-yellowgreen)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.01/arm_compute-v24.01-bin-linux-armv7a-neon-cl.tar.gz) |
-| arm64-v8a    | Android™          | [![](https://img.shields.io/badge/build-neon-orange)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.01/arm_compute-v24.01-bin-android-arm64-v8a-neon.tar.gz) [![](https://img.shields.io/badge/build-opencl-blue)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.01/arm_compute-v24.01-bin-android-arm64-v8a-cl.tar.gz) [![](https://img.shields.io/badge/build-neon+cl-yellowgreen)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.01/arm_compute-v24.01-bin-android-arm64-v8a-neon-cl.tar.gz) |
-| arm64-v8a    | Linux®            | [![](https://img.shields.io/badge/build-neon-orange)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.01/arm_compute-v24.01-bin-linux-arm64-v8a-neon.tar.gz) [![](https://img.shields.io/badge/build-opencl-blue)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.01/arm_compute-v24.01-bin-linux-arm64-v8a-cl.tar.gz) [![](https://img.shields.io/badge/build-neon+cl-yellowgreen)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.01/arm_compute-v24.01-bin-linux-arm64-v8a-neon-cl.tar.gz) |
-| arm64-v8.2-a | Android™          | [![](https://img.shields.io/badge/build-neon-orange)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.01/arm_compute-v24.01-bin-android-arm64-v8.2-a-neon.tar.gz) [![](https://img.shields.io/badge/build-opencl-blue)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.01/arm_compute-v24.01-bin-android-arm64-v8.2-a-cl.tar.gz) [![](https://img.shields.io/badge/build-neon+cl-yellowgreen)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.01/arm_compute-v24.01-bin-android-arm64-v8.2-a-neon-cl.tar.gz) |
-| arm64-v8.2-a | Linux®            | [![](https://img.shields.io/badge/build-neon-orange)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.01/arm_compute-v24.01-bin-linux-arm64-v8.2-a-neon.tar.gz) [![](https://img.shields.io/badge/build-opencl-blue)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.01/arm_compute-v24.01-bin-linux-arm64-v8.2-a-cl.tar.gz) [![](https://img.shields.io/badge/build-neon+cl-yellowgreen)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.01/arm_compute-v24.01-bin-linux-arm64-v8.2-a-neon-cl.tar.gz) |
+| armv7        | Linux®            | [![](https://img.shields.io/badge/build-neon-orange)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.02/arm_compute-v24.02-bin-linux-armv7a-neon.tar.gz) [![](https://img.shields.io/badge/build-opencl-blue)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.02/arm_compute-v24.02-bin-linux-armv7a-cl.tar.gz) [![](https://img.shields.io/badge/build-neon+cl-yellowgreen)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.02/arm_compute-v24.02-bin-linux-armv7a-neon-cl.tar.gz) |
+| arm64-v8a    | Android™          | [![](https://img.shields.io/badge/build-neon-orange)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.02/arm_compute-v24.02-bin-android-arm64-v8a-neon.tar.gz) [![](https://img.shields.io/badge/build-opencl-blue)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.02/arm_compute-v24.02-bin-android-arm64-v8a-cl.tar.gz) [![](https://img.shields.io/badge/build-neon+cl-yellowgreen)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.02/arm_compute-v24.02-bin-android-arm64-v8a-neon-cl.tar.gz) |
+| arm64-v8a    | Linux®            | [![](https://img.shields.io/badge/build-neon-orange)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.02/arm_compute-v24.02-bin-linux-arm64-v8a-neon.tar.gz) [![](https://img.shields.io/badge/build-opencl-blue)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.02/arm_compute-v24.02-bin-linux-arm64-v8a-cl.tar.gz) [![](https://img.shields.io/badge/build-neon+cl-yellowgreen)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.02/arm_compute-v24.02-bin-linux-arm64-v8a-neon-cl.tar.gz) |
+| arm64-v8.2-a | Android™          | [![](https://img.shields.io/badge/build-neon-orange)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.02/arm_compute-v24.02-bin-android-arm64-v8.2-a-neon.tar.gz) [![](https://img.shields.io/badge/build-opencl-blue)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.02/arm_compute-v24.02-bin-android-arm64-v8.2-a-cl.tar.gz) [![](https://img.shields.io/badge/build-neon+cl-yellowgreen)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.02/arm_compute-v24.02-bin-android-arm64-v8.2-a-neon-cl.tar.gz) |
+| arm64-v8.2-a | Linux®            | [![](https://img.shields.io/badge/build-neon-orange)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.02/arm_compute-v24.02-bin-linux-arm64-v8.2-a-neon.tar.gz) [![](https://img.shields.io/badge/build-opencl-blue)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.02/arm_compute-v24.02-bin-linux-arm64-v8.2-a-cl.tar.gz) [![](https://img.shields.io/badge/build-neon+cl-yellowgreen)](https://github.com/ARM-software/ComputeLibrary/releases/download/v24.02/arm_compute-v24.02-bin-linux-arm64-v8.2-a-neon-cl.tar.gz) |
 
 <br>
 
-Please refer to the following link for more pre-built binaries: [![](https://img.shields.io/badge/v24.01-bins-yellowgreen)](https://github.com/ARM-software/ComputeLibrary/releases/tag/v24.01)
+Please refer to the following link for more pre-built binaries: [![](https://img.shields.io/badge/v24.02-bins-yellowgreen)](https://github.com/ARM-software/ComputeLibrary/releases/tag/v24.02)
 
 Pre-build binaries are generated with the following security / good coding practices related flags:
 > -Wall, -Wextra, -Wformat=2, -Winit-self, -Wstrict-overflow=2, -Wswitch-default, -Woverloaded-virtual, -Wformat-security, -Wctor-dtor-privacy, -Wsign-promo, -Weffc++, -pedantic, -fstack-protector-strong

diff --git a/SConscript b/SConscript
@@ -31,8 +31,8 @@ import zlib
 import json
 import codecs
 
-VERSION = "v24.01"
-LIBRARY_VERSION_MAJOR = 34
+VERSION = "v24.02"
+LIBRARY_VERSION_MAJOR = 35
 LIBRARY_VERSION_MINOR =  0
 LIBRARY_VERSION_PATCH =  0
 SONAME_VERSION = str(LIBRARY_VERSION_MAJOR) + "." + str(LIBRARY_VERSION_MINOR) + "." + str(LIBRARY_VERSION_PATCH)
@@ -93,7 +93,7 @@ def build_multiisa_lib_objects():
                            'ARM_COMPUTE_ENABLE_I8MM', 'ARM_COMPUTE_ENABLE_SVEF32MM'])
 
     # Build all the common files for the base architecture
-    if env['arch'] == 'armv8a':
+    if env['arch'] == 'armv8a' or env['arch'] == 'arm64-v8a':
         lib_static_objs += build_obj_list(filedefs["armv8-a"], misa_lib_files, static=True)
         lib_shared_objs += build_obj_list(filedefs["armv8-a"], misa_lib_files, static=False)
     else:
@@ -137,7 +137,7 @@ def recursive_glob(root_dir, pattern):
 
 
 def get_ckw_obj_list():
-    cmake_obj_dir = os.path.abspath("prototype/CMakeFiles/ckw_prototype.dir/src")
+    cmake_obj_dir = os.path.abspath("CMakeFiles/ckw.dir/src")
     return recursive_glob(root_dir=cmake_obj_dir, pattern=".*.o$")
 
 
@@ -163,7 +163,7 @@ def build_library(name, build_env, sources, static=False, libs=[]):
     else:
         # Always statically link Compute Library against CKW
         if env['experimental_dynamic_fusion'] and name == "arm_compute":
-            libs.append('libckw_prototype.a')
+            libs.append('libckw.a')
 
         # Add shared library versioning
         if env['set_soname']:
@@ -532,8 +532,9 @@ arm_compute_env.Append(CPPDEFINES = [('ARM_COMPUTE_VERSION_MAJOR', LIBRARY_VERSI
 
 # Don't allow undefined references in the libraries:
 undefined_flag = '-Wl,-undefined,error' if 'macos' in arm_compute_env["os"] else '-Wl,--no-undefined'
-if not env['thread_sanitizer']:
+if not env['thread_sanitizer'] and not env['address_sanitizer'] and not env['undefined_sanitizer']:
     arm_compute_env.Append(LINKFLAGS=[undefined_flag])
+
 arm_compute_env.Append(CPPPATH =[Dir("./src/core/").path] )
 
 if env['os'] != 'openbsd':