Halide Files

Major improvements

• Halide is now available for both C++ and Python usage via Pip. Try pip install halide today!
• The Vulkan backend has matured substantially.
• The HTML "conceptual statement" output now supports dark mode viewing.
• For developers, CMake 3.28 is now required and we no longer require an internet connection during the build.
• Thread pool improvements mean that workloads that do a small number of small tasks in parallel (e.g. a cheap operation applied to a small image) are up to 3x faster. If you have schedules that do not use parallelism for small inputs because you found it didn't provide any speedup, you may want to re-benchmark.
• You can now query properties of the compiled-for target as Exprs, simplifying helper code that wants to do different things depending on the target architecture. Example: f(x) = select(target_arch_is(Target::ARM), 3, 7). Helpers include target_arch_is, target_os_is, target_has_feature, target_bits, and target_natural_vector_size. These are resolved to constants at compile-time and simplified away. Use with care, as this (intentionally) results in different behavior on different platforms.

Breaking changes

• We now distribute libGenGen.a rather than GenGen.cpp.
• Downstream users should link to this library with /WHOLEARCHIVE: or -Wl,--whole-archive rather than build GenGen.cpp themselves.
• Users of the CMake package should be unaffected.
• In keeping with our LLVM support policy, support for LLVM 16 has been removed.
• We no longer use the le64/le32 generic targets for compiling runtime modules to LLVM. These targets were removed in LLVM upstream.

What's Changed

Apps and tests

• Reschedule the matrix multiply performance app by @abadams in https://github.com/halide/Halide/pull/8418
• Update lesson_22_jit_performance.cpp by @abadams in https://github.com/halide/Halide/pull/8438
• Add threadpool performance test by @abadams in https://github.com/halide/Halide/pull/8447
• Don't allow internal_error to pass an error test by @alexreinking in https://github.com/halide/Halide/pull/8458
• Get more consistent distributions in parallel scenarios test by @abadams in https://github.com/halide/Halide/pull/8451

Autoschedulers

• Consider all Exprs a func uses, not just the RHS, in Li2018 by @abadams in https://github.com/halide/Halide/pull/8326

Build system

• Python_bindings-test-as-installed by @LebedevRI in https://github.com/halide/Halide/pull/8355
• Bump Halide version to 19 in main branch by @steven-johnson in https://github.com/halide/Halide/pull/8357
• Remove warning for unsupported compilers by @alexreinking in https://github.com/halide/Halide/pull/8362
• Bump CMake minimum version to 3.28 by @alexreinking in https://github.com/halide/Halide/pull/8363
• Quick CMake fixes enabled by 3.28 by @alexreinking in https://github.com/halide/Halide/pull/8365
• Distribute GenGen as a static library by @alexreinking in https://github.com/halide/Halide/pull/8367
• Clean up serialization build code by @alexreinking in https://github.com/halide/Halide/pull/8369
• List headers with target_sources FILE_SETS by @alexreinking in https://github.com/halide/Halide/pull/8370
• Clean up autoscheduler dependencies by @alexreinking in https://github.com/halide/Halide/pull/8372
• Use a Find module for V8 by @alexreinking in https://github.com/halide/Halide/pull/8373
• Use a Find module for NodeJS by @alexreinking in https://github.com/halide/Halide/pull/8374
• Move dependencies/wasm to use sites by @alexreinking in https://github.com/halide/Halide/pull/8377
• Replace FetchContent with a custom dependency provider by @alexreinking in https://github.com/halide/Halide/pull/8378
• Two more build fixes by @LebedevRI in https://github.com/halide/Halide/pull/8371
• Rework LLVM into Find module and enact new component policy. by @alexreinking in https://github.com/halide/Halide/pull/8379
• Reflow src/CMakeLists.txt in logical groups by @alexreinking in https://github.com/halide/Halide/pull/8383
• Introduce HalideFeatures system for optional components by @alexreinking in https://github.com/halide/Halide/pull/8384
• Scan generated export files to determine dependencies. by @alexreinking in https://github.com/halide/Halide/pull/8385
• Rewrite bundle_static to be much more efficient. by @alexreinking in https://github.com/halide/Halide/pull/8386
• Support using vcpkg to build dependencies on all platforms by @alexreinking in https://github.com/halide/Halide/pull/8387
• Fix bundling error on buildbots by @alexreinking in https://github.com/halide/Halide/pull/8392
• Support CMAKE_OSX_ARCHITECTURES by @alexreinking in https://github.com/halide/Halide/pull/8390
• Fix Homebrew LLVM 19 by @alexreinking in https://github.com/halide/Halide/pull/8431
• Fix CPack package naming when cross-compiling by @alexreinking in https://github.com/halide/Halide/pull/8492
• Fix Apple libtool detection in bundle_static by @alexreinking in https://github.com/halide/Halide/pull/8495

CodeGen

• Select condition vector lanes must match the true and false value by @abadams in https://github.com/halide/Halide/pull/8465
• Emit vscale_range() fn attribute in correct syntax by @steven-johnson in https://github.com/halide/Halide/pull/8457
• Fix [#8455] (in combination with [#8457]) by @steven-johnson in https://github.com/halide/Halide/pull/8456
• Fix bonehead mistake in get_md_bool() by @steven-johnson in https://github.com/halide/Halide/pull/8469
• Propagate some facts about inequalities with min/max by @shoaibkamil in https://github.com/halide/Halide/pull/8475
• This fixed an issue where predicates in .specialize() directives weren't able to eliminate select() cases. [#8443]

Debugging

• Add LLDB pretty-printing by @alexreinking in https://github.com/halide/Halide/pull/8460
• Print constants in scientific precision by @antonysigma in https://github.com/halide/Halide/pull/8506
• Adaptive Dark colorscheme for Stmt HTML. Ability to programmatically export conceptual stmt files. by @mcourteaux in https://github.com/halide/Halide/pull/8327

Documentation

• Update README.md by @abadams in https://github.com/halide/Halide/pull/8404
• Big documentation update by @alexreinking in https://github.com/halide/Halide/pull/8410
• Document how to find Halide from a pip installation by @alexreinking in https://github.com/halide/Halide/pull/8411
• Link to PyPI from Doxygen index.html by @alexreinking in https://github.com/halide/Halide/pull/8415
• Include our Markdown documentation in the Doxygen site. by @alexreinking in https://github.com/halide/Halide/pull/8417
• Add missing backslash by @abadams in https://github.com/halide/Halide/pull/8419

Frontend

• Don't let users disguise RVars as Vars by @abadams in https://github.com/halide/Halide/pull/8441
• Add helper functions to query properties of the lowered Target (#8192) by @steven-johnson in https://github.com/halide/Halide/pull/8359

Hardware backends

• Fix injection of GPU buffers that do not go by a Func name (i.e. alloc groups). by @mcourteaux in https://github.com/halide/Halide/pull/8333
• Remove vestigial AMDGPU backend by @alexreinking in https://github.com/halide/Halide/pull/8382
• Add ARMv8.x feature flags by @steven-johnson in https://github.com/halide/Halide/pull/4489
• [vulkan] Fixes to address outstanding validation failures by @derek-gerstmann in https://github.com/halide/Halide/pull/8448
• [vulkan] Reduce descriptor sets, use official headers, improve allocator, remove module destructor by @derek-gerstmann in https://github.com/halide/Halide/pull/8452
• [vulkan] Skip async_copy_chain and gpu_allocation_cache correctness tests on Windows by @derek-gerstmann in https://github.com/halide/Halide/pull/8503

LLVM

• Don't use le32/le64 by @steven-johnson in https://github.com/halide/Halide/pull/8344
• Fix for the removed DataLayout constructor. by @mcourteaux in https://github.com/halide/Halide/pull/8391
• Drop support for LLVM 16 in main by @steven-johnson in https://github.com/halide/Halide/pull/8358
• Allow LLVM 20 by @steven-johnson in https://github.com/halide/Halide/pull/8352
• Fix for top-of-tree LLVM by @steven-johnson in https://github.com/halide/Halide/pull/8421
• Fix for top-of-tree LLVM by @steven-johnson in https://github.com/halide/Halide/pull/8425
• Fix for top-of-tree LLVM by @steven-johnson in https://github.com/halide/Halide/pull/8442
• Fix datalayout for osx-arm-64 by @abadams in https://github.com/halide/Halide/pull/8449
• Fix top of LLVM. by @mcourteaux in https://github.com/halide/Halide/pull/8454
• Replace all use of getPointerTo() with PointerType::get() by @steven-johnson in https://github.com/halide/Halide/pull/8473

Python

• Fix Numpy 2.0 compatibility bug in lesson 10 by @alexreinking in https://github.com/halide/Halide/pull/8381
• Pip packaging at last! by @alexreinking in https://github.com/halide/Halide/pull/8405
• Update pip package metadata by @alexreinking in https://github.com/halide/Halide/pull/8412
• Fix classifier spelling by @alexreinking in https://github.com/halide/Halide/pull/8413
• Upgrade LLVM to 19.1.0 in pip package by @alexreinking in https://github.com/halide/Halide/pull/8423
• Update PIP LLVM to 19.1.4 by @alexreinking in https://github.com/halide/Halide/pull/8488
• PythonExtensionGen: ~PyHalideBuffer should call device_free() (#8399) by @steven-johnson in https://github.com/halide/Halide/pull/8439

Runtime

• Fix profiler to report time spent on GPU kernels again instead of on 'wait for parallel tasks'. by @mcourteaux in https://github.com/halide/Halide/pull/8453
• Don't spin on the main mutex while waiting for new work by @abadams in https://github.com/halide/Halide/pull/8433

Minor bugfixes / other cleanup

• Remove remaining dregs of tuple_select (oops) by @steven-johnson in https://github.com/halide/Halide/pull/8329
• Fix incorrect output in Python tutorial, lesson 5 by @qqaatw in https://github.com/halide/Halide/pull/8331
• Make pybind11 minimum version check compatible with pybind11 v3. by @rwgk in https://github.com/halide/Halide/pull/8366
• Partially apply clang-tidy fixes we don't enforce yet by @abadams in https://github.com/halide/Halide/pull/8376
• Fix incorrect std::array sizes in Target.cpp by @steven-johnson in https://github.com/halide/Halide/pull/8396
• Fix _Float16 detection on ARM64 GCC<13 by @alexreinking in https://github.com/halide/Halide/pull/8401
• Make run-clang-tidy.sh work on macOS by @alexreinking in https://github.com/halide/Halide/pull/8416
• Some minor fixes for C++23 compilation errors. by @zvookin in https://github.com/halide/Halide/pull/8422
• Fix two warnings on GCC 14.2.1. by @mcourteaux in https://github.com/halide/Halide/pull/8430
• Fix typos by @alexreinking in https://github.com/halide/Halide/pull/8459
• Fix two trivial build errors by @steven-johnson in https://github.com/halide/Halide/pull/8467
• Fix [#8470]: fuzz_bounds should use select() rather than Select::make() by @steven-johnson in https://github.com/halide/Halide/pull/8471
• Add missing #include by @steven-johnson in https://github.com/halide/Halide/pull/8476
• Fix heap-use-after-free error in find_best_fit() by @steven-johnson in https://github.com/halide/Halide/pull/8483
• Fix typos in comments by @alexreinking in https://github.com/halide/Halide/pull/8485
• Remove unused is_update argument by @alexreinking in https://github.com/halide/Halide/pull/8487
• Backport reverse_view to clean up some code by @alexreinking in https://github.com/halide/Halide/pull/8486
• Remove type inspection helpers from ApplySplitResult and Split by @alexreinking in https://github.com/halide/Halide/pull/8489
• Use std::optional to clean up some code and prevent use-after-free bugs by @abadams in https://github.com/halide/Halide/pull/8484
• Fix comment for Buffer::copy() (Fixes [#8498]) by @steven-johnson in https://github.com/halide/Halide/pull/8500
• Make the default constructor for ConstantInterval inlinable by @abadams in https://github.com/halide/Halide/pull/8505
• Remove two unused functions by @steven-johnson in https://github.com/halide/Halide/pull/8501
• Move some large stack frames off recursive paths. by @abadams in https://github.com/halide/Halide/pull/8507

New Contributors

• @qqaatw made their first contribution in https://github.com/halide/Halide/pull/8331
• @rwgk made their first contribution in https://github.com/halide/Halide/pull/8366

Full Changelog: https://github.com/halide/Halide/compare/v18.0.0...v19.0.0