Releases: Libraries-Openly-Fused/FusedKernelLibrary
Releases · Libraries-Openly-Fused/FusedKernelLibrary
Beta-0.1.13-LTS
Increment project version revision to 13
Beta-0.1.12-LTS
What's Changed
- 137 use cxp lib functions instead of ops inside ops to reduce include mess by @morousg in #139
- 140 add test fot back fuser methods by @morousg in #141
- 136 implement executor for divergentbatchtransformdpp by @morousg in #144
- Fix Ptr Copy and Move constructors on VS2017 by @morousg in #146
- Removing unsused parameter that whas making linux compilation to fail by @morousg in #148
Full Changelog: Beta-0.1.11...Beta-0.1.12-LTS
Beta-0.1.11
Merge pull request #129 from Libraries-Openly-Fused/128-support-llvm-…
Beta-0.1.10
Changes in this version:
- Adding a more complete and C++ compliant implementation of the vector types operators
- Fixed VS2017 compilation issues
- Refactored the batch operations and their interaction with Read and ReadBack operations
- Introduced the IncompleteReadBackType, and adapted the corresponding code to take it into account.
- Simplified OperationData implementation
Beta-0.1.9
Merge pull request #88 from Libraries-Openly-Fused/copilot/fix-87 Convert all enums to enum class for type safety and fix CUDA compilation errors
Beta-0.1.8
Merge pull request #39 from morousg/38-fix-cmake-generation-stages Fix generation stage with cmake 4.0 and Ninja
Beta-0.1.7
Now PR's will build in a runner, and execute the tests (Linux ARM64)
There has been a major refactor of the code, in order to have different concepts in different files. I makes the code easier to follow and maintain.
fuseDF() now is fuse(), and it's capable to fuse back. That is to join a Read or ReadBack IOp with an incomplete instantiation of a ReadBack IOp. Previously this feature was only available from then().
Beta-0.1.6
Bump Version 2
Beta-0.1.5
Fixed a performance issue, because we where computing the maximum Widht and Height of all the images in an ReadBatch Operation on the GPU instead of the CPU.
Beta-0.1.4
Removing some warnings and improving Operation VectorReorderRT