Skip to content

Latest commit

 

History

History
38 lines (31 loc) · 2.14 KB

NEWS.md

File metadata and controls

38 lines (31 loc) · 2.14 KB
Raw

OpenCL.jl release notes

In development: OpenCL.jl v1.0

This version is a major rewrite of the package, with a focus on unifying the package with the rest of the JuliaGPU ecosystem.

Breaking changes:

  • Context, device and queue arguments have been removed from most APIs, and are now stored in task-local storage. These values can be queried (cl.platform(), cl.device(), etc) and set (cl.platform!(platform), cl.device!(device), etc) as needed.
  • As part of the above change, questionable APIs like cl.create_some_context() and cl.devices() have been removed.
  • The Buffer API has been completely reworked. It now only provides low-level functionality, such as unsafe_copyto! or unsafe_map!, while high-level functionality like copy! is implemented for the CLArray type.
  • The cl.info method, and the getindex overloading to access properties of OpenCL objects, have been replaced by getproperty overloading on the objects themselves (e.g., cl.info(dev, :name) and dev[:name] are now simply dev.name).
  • The blocking cl.launch has been replaced by a nonblocking cl.call, while also removing the getindex-overloading shorthand. However, it's recommended to use the newly-added cl.clcall function, which takes an additional tuple type argument and performs automatic conversions of arguments to those types. This makes it possible to pass a CLArray to an OpenCL C function expecting Buffer-backed pointers, for example.
  • Argument conversion has been removed; the user should make sure Julia arguments passed to kernels match the OpenCL argument types (i.e., no empty types, 4-element tuples for a 3-element float3 arguments).
  • The to_host function has been replaced by simply calling Array on the CLArray.
  • Queue and execution capabilities of a device are now to be queried using dedicated functions, cl.queue_properties and cl.exec_capabilities.

New features:

  • Loading SPIR-V IL programs is now supported: cl.Program(il=...).
  • Support for POCL has been added. Together with a rework of OpenCL_jll, this makes it possible to do using OpenCL, pocl_jll and have the POCL platform available.