|
1 | 1 | Change Log / Release Log for mallocMC |
2 | 2 | ================================================================ |
3 | 3 |
|
| 4 | +3.0.0-dev |
| 5 | +----- |
| 6 | + |
| 7 | +Date: not yet released |
| 8 | + |
| 9 | +This is a major release including breaking changes on the interface and |
| 10 | +build system level as well as many new features and bug fixes. |
| 11 | + |
| 12 | +Features |
| 13 | + |
| 14 | +- New algorithm FlatterScatter |
| 15 | +- New prototype GallatinCuda (CUDA-only wrapper of a recent proposal) |
| 16 | +- `mallocMC.cuh` header for convenient use in native CUDA applications |
| 17 | +- Modern CMake support and complete re-organisation of the file tree |
| 18 | +- Modern package management with CMake Package Manager (CPM) |
| 19 | +- New interface: Allocator takes a tag instead of an Acc |
| 20 | +- C++20 is required now. |
| 21 | + |
| 22 | +Bug fixes |
| 23 | + |
| 24 | +- Memfence before releasing a chunk in ScatterAlloc (and FlatterScatter) |
| 25 | +- Fix compilation error with multiple accelerators |
| 26 | + |
| 27 | +Maintenance: |
| 28 | + |
| 29 | +- Update alpaka to 1.2.0 + patches (latest develop branch) |
| 30 | +- Update tooling (clang-format, clang-tidy, ...) |
| 31 | +- Add pre-commit |
| 32 | +- Add rudimentary CI (additional coverage through PIConGPU) |
| 33 | +- Add decent test coverage for FlatterScatter and ScatterAlloc |
| 34 | + |
| 35 | +2.6.0 |
| 36 | +----- |
| 37 | + |
| 38 | +Date: 2024-08-28 |
| 39 | + |
| 40 | +This is mostly a maintenance release. |
| 41 | +Years of production use in PIConGPU have surfaced |
| 42 | +a number of bugs and inconsistencies that are addressed. |
| 43 | +We will drop the 'crp' suffix in our version numbers |
| 44 | +because the original version |
| 45 | +(that we distinguished from via 'crp') |
| 46 | +is no longer maintained. |
| 47 | + |
| 48 | +Features |
| 49 | + |
| 50 | +- Compatibility with C++20 |
| 51 | +- Use runtime selected number of access blocks |
| 52 | +- Full support for CPU accelerators |
| 53 | +- Scatter: optimize page search algorithm |
| 54 | +- Scatter: optimize hash and allocation size |
| 55 | +- Scatter: add more informations to `classname()` |
| 56 | + |
| 57 | +Bug fixes |
| 58 | + |
| 59 | +- Scatter: Various deadlocks, data races and inconsistencies |
| 60 | +- Compilation failure without `alpaka_ACC_GPU_CUDA_ONLY_MODE` set |
| 61 | +- Adjustments to the examples to compile and run cleanly again |
| 62 | +- Alignment violation in edge cases of small allocations fixed |
| 63 | +- Remove undefined behaviour in handling bit masks |
| 64 | + |
| 65 | +Maintenance |
| 66 | + |
| 67 | +- Update to alpaka 1.1.0 |
| 68 | +- Minor adjustments in the README |
| 69 | +- Scatter: Internal refactorings |
| 70 | +- CI fixes |
| 71 | + |
4 | 72 | 2.5.0crp |
5 | 73 | -------- |
6 | 74 | **Date:** 2021-02-18 |
|
0 commit comments