Skip to content

Release v1.0.0

Choose a tag to compare

@okozelsk okozelsk released this 15 Jun 10:16
· 1 commit to master since this release

This first release allows the user to safely utilize available GPU accelerators from a multi-threaded application environment. If multiple GPU accelerators are available, simple load-balancing is applied, which takes into account the number of currently submitted tasks on the accelerators.
The PPU4ILGPU library also implements a fast cache of compiled kernels and provides reusable AcceleratorStream objects.
PPU4ILGPU is designed for both singleton and dependency injection application scenarios.