v0.7.0 release

CHANGELOG.md

@@ -1,6 +1,76 @@
 # Changelog
 
-## [v0.6.0](https://github.com/EthicalML/vulkan-kompute/tree/v0.6.0)
+## [v0.7.0](https://github.com/EthicalML/vulkan-kompute/tree/v0.7.0)
+
+[Full Changelog](https://github.com/EthicalML/vulkan-kompute/compare/v0.6.0...v0.7.0)
+
+**Implemented enhancements:**
+
+- Extend non-spdlog print functions to use std::format [\#158](https://github.com/EthicalML/vulkan-kompute/issues/158)
+- Add code coverage reports with codecov [\#145](https://github.com/EthicalML/vulkan-kompute/issues/145)
+- Explore removing `std::vector mData;` completely from Tensor in favour of always storing data in hostVisible buffer memory \(TBC\) [\#144](https://github.com/EthicalML/vulkan-kompute/issues/144)
+- Update all examples to match breaking changes in 0.7.0 [\#141](https://github.com/EthicalML/vulkan-kompute/issues/141)
+- Avoid copy when returning python numpy / array [\#139](https://github.com/EthicalML/vulkan-kompute/issues/139)
+- Cover all Python & C++ tests in CI  [\#121](https://github.com/EthicalML/vulkan-kompute/issues/121)
+- Add C++ Test for Simple Work Groups Example [\#117](https://github.com/EthicalML/vulkan-kompute/issues/117)
+- Expose push constants in OpAlgo [\#54](https://github.com/EthicalML/vulkan-kompute/issues/54)
+- Expose ability to create barriers in OpTensor operations [\#45](https://github.com/EthicalML/vulkan-kompute/issues/45)
+- Create delete function in manager to free / destroy sequence [\#36](https://github.com/EthicalML/vulkan-kompute/issues/36)
+- Make specialisation data extensible [\#12](https://github.com/EthicalML/vulkan-kompute/issues/12)
+- Support multiple types for Kompute Tensors [\#2](https://github.com/EthicalML/vulkan-kompute/issues/2)
+- Added re-record sequence functionality and updated docs [\#171](https://github.com/EthicalML/vulkan-kompute/pull/171) ([axsaucedo](https://github.com/axsaucedo))
+- Extend non-spdlog print functions to use fmt::format / fmt::print [\#159](https://github.com/EthicalML/vulkan-kompute/pull/159) ([axsaucedo](https://github.com/axsaucedo))
+- Added support for custom SpecializedConstants and removed KomputeWorkgroup class [\#151](https://github.com/EthicalML/vulkan-kompute/pull/151) ([axsaucedo](https://github.com/axsaucedo))
+- Added destroy functions for tensors and sequences \(named and object\) [\#146](https://github.com/EthicalML/vulkan-kompute/pull/146) ([axsaucedo](https://github.com/axsaucedo))
+
+**Fixed bugs:**
+
+- push\_constant not working in my case? [\#168](https://github.com/EthicalML/vulkan-kompute/issues/168)
+- DescriptorPool set is not being freed [\#155](https://github.com/EthicalML/vulkan-kompute/issues/155)
+- Updated memory barriers to include staging buffers [\#182](https://github.com/EthicalML/vulkan-kompute/pull/182) ([axsaucedo](https://github.com/axsaucedo))
+- Adds push const ranges in pipelinelayout to fix \#168 [\#174](https://github.com/EthicalML/vulkan-kompute/pull/174) ([axsaucedo](https://github.com/axsaucedo))
+- Added destructor for staging tensors [\#134](https://github.com/EthicalML/vulkan-kompute/pull/134) ([axsaucedo](https://github.com/axsaucedo))
+
+**Closed issues:**
+
+- Update memory barriers to align with tensor staging/primary memory revamp [\#181](https://github.com/EthicalML/vulkan-kompute/issues/181)
+- Move shader defaultResource inside kp::Shader class [\#175](https://github.com/EthicalML/vulkan-kompute/issues/175)
+- Reach at least 90% code coverage on tests [\#170](https://github.com/EthicalML/vulkan-kompute/issues/170)
+- Add functionality to re-record sequence as now it's possible to update the underlying algorithm [\#169](https://github.com/EthicalML/vulkan-kompute/issues/169)
+- Use numpy arrays as default return value [\#166](https://github.com/EthicalML/vulkan-kompute/issues/166)
+- Update all shared\_ptr value passes to be by ref or const ref [\#161](https://github.com/EthicalML/vulkan-kompute/issues/161)
+- Amend memory hierarchy for kp::Operations so they can be created separately [\#160](https://github.com/EthicalML/vulkan-kompute/issues/160)
+- Customise theme of documentation [\#156](https://github.com/EthicalML/vulkan-kompute/issues/156)
+- Remove KomputeWorkgroup class in favour of std::array\<uint32\_t, 3\> [\#152](https://github.com/EthicalML/vulkan-kompute/issues/152)
+- Passing raw GLSL string to Shader Module depricated so remove this method from supported approach [\#150](https://github.com/EthicalML/vulkan-kompute/issues/150)
+- Add python backwards compatibility for eval\_tensor\_create\_def [\#147](https://github.com/EthicalML/vulkan-kompute/issues/147)
+- Document breaking changes for 0.7.0 [\#140](https://github.com/EthicalML/vulkan-kompute/issues/140)
+- Tensor memory management and memory hierarchy redesign [\#136](https://github.com/EthicalML/vulkan-kompute/issues/136)
+- Staging tensor GPU memory is not freed as part of OpCreateTensor removal [\#133](https://github.com/EthicalML/vulkan-kompute/issues/133)
+- eStorage Tensors are currently unusable as OpTensorCreate calls mapDataIntoHostMemory [\#132](https://github.com/EthicalML/vulkan-kompute/issues/132)
+- 0.6.0 Release [\#126](https://github.com/EthicalML/vulkan-kompute/issues/126)
+- java.lang.UnsatisfiedLinkError: dlopen failed: library "libkompute-jni.so" not found [\#125](https://github.com/EthicalML/vulkan-kompute/issues/125)
+- Initial exploration: Include explicit GLSL to SPIRV compilation [\#107](https://github.com/EthicalML/vulkan-kompute/issues/107)
+- Add support for push constants [\#106](https://github.com/EthicalML/vulkan-kompute/issues/106)
+
+**Merged pull requests:**
+
+- Resolve moving all functions from tensor HPP to CPP [\#186](https://github.com/EthicalML/vulkan-kompute/pull/186) ([axsaucedo](https://github.com/axsaucedo))
+- Device Properties [\#184](https://github.com/EthicalML/vulkan-kompute/pull/184) ([alexander-g](https://github.com/alexander-g))
+- Too many warnings [\#183](https://github.com/EthicalML/vulkan-kompute/pull/183) ([alexander-g](https://github.com/alexander-g))
+- Add support for bool, double, int32, uint32 and float32 on Tensors via TensorT [\#177](https://github.com/EthicalML/vulkan-kompute/pull/177) ([axsaucedo](https://github.com/axsaucedo))
+- Support for Timestamping [\#176](https://github.com/EthicalML/vulkan-kompute/pull/176) ([alexander-g](https://github.com/alexander-g))
+- Test for ShaderResources [\#165](https://github.com/EthicalML/vulkan-kompute/pull/165) ([aliPMPAINT](https://github.com/aliPMPAINT))
+- Amend memory hierarchy to enable for push constants and functional interface for more flexible operations [\#164](https://github.com/EthicalML/vulkan-kompute/pull/164) ([axsaucedo](https://github.com/axsaucedo))
+- made changes for include paths for complete installation [\#163](https://github.com/EthicalML/vulkan-kompute/pull/163) ([aliPMPAINT](https://github.com/aliPMPAINT))
+- Added dark mode on docs [\#157](https://github.com/EthicalML/vulkan-kompute/pull/157) ([axsaucedo](https://github.com/axsaucedo))
+- Glslang implementation for online shader compilation [\#154](https://github.com/EthicalML/vulkan-kompute/pull/154) ([axsaucedo](https://github.com/axsaucedo))
+- Adding test code coverage using gcov and lcov [\#149](https://github.com/EthicalML/vulkan-kompute/pull/149) ([axsaucedo](https://github.com/axsaucedo))
+- Added temporary backwards compatibility for eval\_tensor\_create\_def function [\#148](https://github.com/EthicalML/vulkan-kompute/pull/148) ([axsaucedo](https://github.com/axsaucedo))
+- Amend memory ownership hierarchy to have Tensor owned by Manager instead of OpCreateTensor / OpBase [\#138](https://github.com/EthicalML/vulkan-kompute/pull/138) ([axsaucedo](https://github.com/axsaucedo))
+- Removed Staging Tensors in favour of having two buffer & memory in a Tensor to minimise data transfer [\#137](https://github.com/EthicalML/vulkan-kompute/pull/137) ([axsaucedo](https://github.com/axsaucedo))
+
+## [v0.6.0](https://github.com/EthicalML/vulkan-kompute/tree/v0.6.0) (2021-01-31)
 
 [Full Changelog](https://github.com/EthicalML/vulkan-kompute/compare/v0.5.1...v0.6.0)
 
@@ -49,7 +119,6 @@
 - Remove the template params from OpAlgoBase for dispatch layout [\#57](https://github.com/EthicalML/vulkan-kompute/issues/57)
 - Enable layout to be configured dynamically within shaders [\#26](https://github.com/EthicalML/vulkan-kompute/issues/26)
 - replaced "static unsigned const" to "static const unsigned" to avoid SWIG parsing error. [\#95](https://github.com/EthicalML/vulkan-kompute/pull/95) ([0x0f0f0f](https://github.com/0x0f0f0f))
-- Added python bindings with kp as python module  [\#88](https://github.com/EthicalML/vulkan-kompute/pull/88) ([axsaucedo](https://github.com/axsaucedo))
 
 **Closed issues:**
 
@@ -69,6 +138,7 @@
 - Adding Python package for Kompute [\#87](https://github.com/EthicalML/vulkan-kompute/issues/87)
 - Python shader extension [\#91](https://github.com/EthicalML/vulkan-kompute/pull/91) ([axsaucedo](https://github.com/axsaucedo))
 - Enhanced python build [\#89](https://github.com/EthicalML/vulkan-kompute/pull/89) ([axsaucedo](https://github.com/axsaucedo))
+- Added python bindings with kp as python module  [\#88](https://github.com/EthicalML/vulkan-kompute/pull/88) ([axsaucedo](https://github.com/axsaucedo))
 
 **Closed issues:**
 

CMakeLists.txt

@@ -1,5 +1,5 @@
 cmake_minimum_required(VERSION 3.4.1)
-project(kompute VERSION 0.6.0)
+project(kompute VERSION 0.7.0)
 
 set(CMAKE_CXX_STANDARD 14)
 
@@ -20,6 +20,8 @@ option(KOMPUTE_OPT_REPO_SUBMODULE_BUILD, "Use the submodule repos instead of ext
 option(KOMPUTE_OPT_ANDOID_BUILD "Enable android compilation flags required" 0)
 option(KOMPUTE_OPT_DISABLE_VK_DEBUG_LAYERS "Explicitly disable debug layers even on debug" 0)
 option(KOMPUTE_OPT_DISABLE_SHADER_UTILS "Remove shader util code and dependencies including glslang" 0)
+option(KOMPUTE_OPT_DEPENDENCIES_SHARED_LIBS "Whether to use shared libraries for dependencies for install" 0)
+option(KOMPUTE_OPT_BUILD_AS_SHARED_LIB "Whether to build kompute as shared library" 0)
 # Build flags
 set(KOMPUTE_EXTRA_CXX_FLAGS "" CACHE STRING "Extra compile flags for Kompute, see docs for full list")
 
@@ -29,6 +31,10 @@ if(KOMPUTE_OPT_ENABLE_SPDLOG)
     if(KOMPUTE_OPT_INSTALL)
         # Enable install parameters for spdlog (overrides parameters passed)
         set(SPDLOG_INSTALL ON CACHE BOOL "Enables install of spdlot" FORCE)
+
+        if(KOMPUTE_OPT_DEPENDENCIES_SHARED_LIBS)
+            set(SPDLOG_BUILD_SHARED ON CACHE BOOL "Enables build of shared libraries" FORCE)
+        endif()
     endif()
 endif()
 
@@ -54,7 +60,11 @@ if(NOT KOMPUTE_OPT_DISABLE_SHADER_UTILS)
         # Enable install parameters for glslang (overrides parameters passed)
         # When install is enabled the glslang libraries become shared
         set(ENABLE_GLSLANG_INSTALL ON CACHE BOOL "Enables install of glslang" FORCE)
-        set(BUILD_SHARED_LIBS ON CACHE BOOL "Enables build of shared libraries" FORCE)
+
+        # By default we enable shared library based installation
+        if(KOMPUTE_OPT_DEPENDENCIES_SHARED_LIBS)
+            set(BUILD_SHARED_LIBS ON CACHE BOOL "Enables build of shared libraries" FORCE)
+        endif()
     endif()
 else()
     set(KOMPUTE_EXTRA_CXX_FLAGS "${KOMPUTE_EXTRA_CXX_FLAGS} -DKOMPUTE_DISABLE_SHADER_UTILS=1")

Makefile

@@ -199,4 +199,4 @@ format:
 build_changelog:
 	docker run --rm -it -v "$(PWD)":/usr/local/src/your-app -e CHANGELOG_GITHUB_TOKEN=${CHANGELOG_GITHUB_TOKEN} ferrarimarco/github-changelog-generator:1.15.2 -u EthicalML -p vulkan-kompute
 	chmod 664 CHANGELOG.md # (Read+Write, Read+Write, Read)
-	sed -i -e 's/\(HEAD\|Unreleased\)/v0.6.0/g' CHANGELOG.md # Replacing unreleased version with latest tag
+	sed -i -e 's/\(HEAD\|Unreleased\)/v${VERSION}/g' CHANGELOG.md # Replacing unreleased version with latest tag

VERSION

@@ -1 +1 @@
-0.6.0
+0.7.0

docs/conf.py

@@ -27,7 +27,7 @@
 author = 'Alejandro Saucedo'
 
 # The full version, including alpha/beta/rc tags
-release = '0.6.0'
+release = '0.7.0'
 
 
 # -- General configuration ---------------------------------------------------

docs/overview/build-system.rst

@@ -33,6 +33,10 @@ This by default configures without any of the extra build tasks (such as buildin
      - Disables the install step in the cmake file (useful for android build)
    * - -DKOMPUTE_OPT_ANDROID_BUILD=1
      - Enables android build which includes and excludes relevant libraries
+   * - -DKOMPUTE_OPT_DEPENDENCIES_SHARED_LIBS=1
+     - Ensures dependencies are referenced as shared libraries for kompute install
+   * - -DKOMPUTE_OPT_BUILD_AS_SHARED_LIB=1
+     - Whether to build Kompute as shared lib instead of static
 
 
 Compile Flags

docs/overview/ci-tests.rst

@@ -81,6 +81,7 @@ Performing Release
 In order to perform the release the following steps need to be carried out:
 
 * Build changelog
+    * Create branch called `v<VERSION>-release`
     * Generate latest changelog `make build_changelog`
     * Update latest tag in new CHANGELOG.md to be the vesion to release 
 * Python Release
@@ -98,7 +99,3 @@ In order to perform the release the following steps need to be carried out:
         * Ensure all tests pass in GPU and CPU: `python -m pytest`
 
 
-```
-```
-
-

python/test/test_tensor_types.py

@@ -82,7 +82,7 @@ def test_type_float_double_incorrect():
 
     assert np.all(tensor_out.data() != arr_in_a * arr_in_b)
 
-@pytest.mark.skipif("swiftshader" in os.environ.get("VK_ICD_FILENAMES"),
+@pytest.mark.skipif("swiftshader" in os.environ.get("VK_ICD_FILENAMES", ""),
                     reason="Swiftshader doesn't support double")
 def test_type_double():
 

setup.py

@@ -70,7 +70,7 @@ def build_extension(self, ext):
 
 setup(
     name='kp',
-    version='0.6.0',
+    version='0.7.0',
     author='Alejandro Saucedo',
     description='Vulkan Kompute: Blazing fast, mobile-enabled, asynchronous, and optimized for advanced GPU processing usecases.',
     long_description=long_description,

src/CMakeLists.txt

@@ -39,19 +39,25 @@ if(KOMPUTE_OPT_ANDOID_BUILD)
         ${PROJECT_SOURCE_DIR}/vk_ndk_wrapper_include/kompute_vk_ndk_wrapper.cpp)
 endif()
 
-add_library(
-    kompute STATIC
-    ${kompute_CPP})
+if(NOT KOMPUTE_OPT_BUILD_AS_SHARED_LIB)
+    add_library(
+        kompute STATIC
+        ${kompute_CPP})
+else()
+    add_library(
+        kompute SHARED
+        ${kompute_CPP})
+endif()
 
 target_include_directories(
     kompute PUBLIC
-    $<INSTALL_INTERFACE:include>
-    $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/include>
+    ${CMAKE_CURRENT_SOURCE_DIR}/include
+    ${PROJECT_SOURCE_DIR}/single_include
 )
 
 if(NOT KOMPUTE_OPT_ANDOID_BUILD)
     target_link_libraries(
-        kompute 
+        kompute
         Vulkan::Vulkan
     )
 else()

vcpkg.json.opt

@@ -1,6 +1,6 @@
 {
   "name": "example",
-  "version-string": "0.6.0",
+  "version-string": "0.7.0",
   "dependencies": [
     "fmt",
     "spdlog",