Version 5-preview.49

What's Changed

• Allow messages for error_todo by @zluda-violet in #415

Full Changelog: v5-preview.48...v5-preview.49

Version 5-preview.48

What's Changed

• Fix grammar errors in README.md by @brlin-tw in #414

Full Changelog: v5-preview.47...v5-preview.48

Version 5-preview.47

What's Changed

• Fix typo in README.md(self-contained) by @brlin-tw in #413

New Contributors

• @brlin-tw made their first contribution in #413

Full Changelog: v5-preview.46...v5-preview.47

Version 5-preview.46

What's Changed

• Handle WARP_SZ by @zluda-violet in #412

Full Changelog: v5-preview.45...v5-preview.46

Version 5-preview.45

What's Changed

• More descriptive message for unknown symbol by @zluda-violet in #411

Full Changelog: v5-preview.44...v5-preview.45

Version 5-preview.44

What's Changed

• Remove duplicate call to linker by @zluda-violet in #410

Full Changelog: v5-preview.43...v5-preview.44

Version 5-preview.43

What's Changed

• Update README by @vosen in #315
• Fix test zluda_dump by @JoelleJS in #316
• feat: enable LTO and codegen-units = 1 optimization by @zamazan4ik in #318
• fix: missing inherits in a release-lto profile by @zamazan4ik in #319
• Improve build system by @vosen in #329
• LLVM unit tests by @JoelleJS in #324
• Implement mode tracking for AMD GPU by @vosen in #342
• Implement mul24 by @JoelleJS in #351
• Explicitly fail compilation on ROCm 6.4 by @vosen in #361
• Create infrastructure for performance libraries by @vosen in #363
• Fix ROCm 6.4 failures by @vosen in #364
• Work around broken AMD Adrenalin 25.5.1 driver by @vosen in #366
• Redo logging to better log dark API and performance libraries by @vosen in #372
• Fix mad.wide, replace external CUDA library in tests with our own by @vosen in #376
• Implement cuGetProcAddress and cuGetProcAddress_v2 by @zluda-violet in #377
• Implement runtime_callback_hooks_fn2 by @zluda-violet in #380
• Implement cuModuleGetLoadingMode by @zluda-violet in #381
• Implement cudart_interface_fn2 by @zluda-violet in #382
• Add automated builds by @vosen in #358
• Handle new attributes in cuDeviceGetAttribute by @zluda-violet in #383
• Implement runtime_callback_hooks_fn6 by @zluda-violet in #386
• Add fp saturation, fix various bugs in cvt instruction exposed by ptx_tests by @vosen in #379
• Use integrity_check implementation by @zluda-violet in #387
• Implement cuLibraryLoadData by @zluda-violet in #388
• Fix bug in get_payload by @zluda-violet in #389
• Remove trailing zeroes from end of ptx by @zluda-violet in #390
• Error instead of infinite loop in derive_parser! by @zluda-violet in #391
• Bump dependencies by @vosen in #392
• Check LLVM IR for test_ptx! with no input/output by @zluda-violet in #394
• Unified fatbin versions behind a single iterator. by @aiwhskruht in #398
• Make derive_parser work with all optional arguments by @zluda-violet in #397
• Read test files at runtime for development ergonomics by @zluda-violet in #395
• Fix floating point min/max by @vosen in #399
• Add warp-wide tests by @zluda-violet in #400
• Add support for bar.red.and.pred by @zluda-violet in #402
• Run unit tests on every commit by @vosen in #401
• Add initialized check to protect zluda from calls during shutdown by @aiwhskruht in #404
• Implement more CUDA driver API to enable simple cuda-samples by @aiwhskruht in #405
• [WIP] Start working on PhysX 32bit by @vosen in #374
• Update README.md by @zluda-violet in #407
• Add support for multiple return arguments by @zluda-violet in #406
• Enable sccache in Rust builds, publish prerelease builds by @vosen in #408

New Contributors

• @zamazan4ik made their first contribution in #318
• @zluda-violet made their first contribution in #377
• @aiwhskruht made their first contribution in #398

Full Changelog: v4...v5-preview.43

Version 4

This is the first release post-rollback and is very limited: only Geekbench is supported

Version 3

Nobody expects the Red Team

Too many changes to list, but broadly:

• Remove Intel GPU support from the compiler
• Add AMD GPU support to the compiler
• Remove Intel GPU host code
• Add AMD GPU host code
• More device instructions. From 40 to 68
• More host functions. From 48 to 184
• Add proof of concept implementation of OptiX framework
• Add minimal support of cuDNN, cuBLAS, cuSPARSE, cuFFT, NCCL, NVML
• Improve ZLUDA launcher for Windows

Version 2

The goal of version 2 has been to fix end to end execution of GeekBench and improve Windows support:

• Several new host-side functions are supported now (e.g. cuModuleLoadDataEx)
• Several bugs have been fixed on the kernel side (e.g. threadIdx/blockIdx is now handled correctly)
• Minor improvement in generated code brought better I/O performance when reading/writing vector objects. This improved performance by several percentage points in select GeekBench benchmarks
• ZLUDA now ships its own injector (with_zluda.exe) which should make running ZLUDA on Windows much easier
• Additionally, we have gained ability to easily create traces of CUDA kernel execution, making enabling new workloads much easier
• ZLUDA now has a CI, which produces binaries on every pull request and commit

Special thanks to @take-cheeze, @nilsmartel and @ritschwumm for contributing to this release