ILGPU.Algorithms Release Versions

Version v0.9

  • Added support for the latest ILGPU release.
  • Fixed critical issues in RadixSort and Scan implementations.
  • Fixed critical issues in several XMath implementations.

Version v0.8

  • Added support for the latest ILGPU release.
  • Added significantly enhanced XMath functions, which MoFtZ built for you
  • Minor bug fixes.

Version v0.7.1

  • Added support for NVIDIA's cuBlas library.

Version v0.7

  • First release of the new ILGPU.Algorithms library (was built on top of the ILGPU.Lightning library)
  • Added support for ILGPU v0.7
  • Added support for new group and warp-based methods that can be used in the scope of kernels
  • Redesigned huge parts of the library to make it cross platform compatible

Version v0.3.0

  • ILGPU.Lightning release
  • Updated project to .Net Standard 2.0. (Community Request)
  • Changed native library content types to none.
  • Adapted RNG methods.
  • Removed obsolete functionality.

Version v0.2.0

  • ILGPU.Lightning release
  • Note that this version introduces several API changes. All functionality is still available for backwards compatibility. However, obsolete functionality will be removed in the next version. Refer to the upgrade guide to upgrade your project:
  • The LightningContext class is no longer required and is marked as obsolete. All functions are now implemented as extension methods to Accelerator objects.
  • Added backwards compatibility for obsolete lightning API.
  • Added several random-number generators.
  • Added support for Cuda 9.0.
  • Adapted lightning library to new ILGPU features.

Version v0.1.4

  • ILGPU.Lightning release
  • Disabled FastMath by default and added additional predefined flag configurations.

Version v0.1.3

  • ILGPU.Lightning release
  • Fixed too strict extent checks in ArrayViewExtensions.

Version v0.1.2

  • ILGPU.Lightning release
  • Fixed critical issues in segmented radix sort.
  • Fixed critical issues in reduction algorithms.
  • Added array-view extensions.
  • Added vector (System.Numerics.VectorX) and array extensions.
  • ILGPU.Lightning release
  • Added debug and release versions to NuGet package. Reason: Exceptions are not allowed in GPU code but debug assertions are allowed. Release builds do not contain assertions. Hence, debug assemblies are are required for proper error messages in GPU kernels during development.