v8.6.5

GPUArrays v8.6.5

Diff since v8.6.4

Merged pull requests:

• Revert "Avoid cartesian iteration where possible." (#463) (@maleadt)

v8.6.4

GPUArrays v8.6.4

Diff since v8.6.3

Merged pull requests:

• Avoid cartesian iteration where possible. (#454) (@maleadt)
• Add a test for reducing larger array sizes (#459) (@maxwindiff)
• change 1.8.0-beta3 to 1.8 in CI (#460) (@evelyne-ringoot)
• Add previously missing @test macro to math/power tests (#461) (@christiangnrd)

v8.6.3

GPUArrays v8.6.3

Diff since v8.6.2

Merged pull requests:

• Support for adjoint and transpose of strided inputs (non-contiguous views) (#452) (@evelyne-ringoot)
• Test number broadcast. (#455) (@maleadt)

v8.6.2

GPUArrays v8.6.2

Diff since v8.6.1

Merged pull requests:

• Use while loop instead of for-range loop in broadcast kernel (#448) (@maxwindiff)
• find: Avoid redefining functions on 1.6. (#449) (@maleadt)

v8.6.1

GPUArrays v8.6.1

Diff since v8.6.0

Closed issues:

• adjoint! does not work with vector inputs (#445)

Merged pull requests:

• Support transpose/adjoint on mixed vector/matrix inputs. (#447) (@maleadt)

v8.6.0

Bump GPUArrays version.

v8.5.0

GPUArrays v8.5.0

Diff since v8.4.2

Closed issues:

• Error from CuMatrix * Diagonal on Julia 1.8 (#426)

Merged pull requests:

• Add an iszero and isone methodfor AbstractGPUMatrix (#419) (@GVigne)
• move backend to GPUArraysCore.jl (#420) (@N5N3)
• Specialize mul! for mat * diag (#425) (@chengchingwen)

v8.4.2

GPUArrays v8.4.2

Diff since v8.4.1

Merged pull requests:

• Move AbstractGPUArrayStyle to GPUArraysCore? (#417) (@mcabbott)
• Move JLArray to a subpackage (#418) (@mcabbott)

v8.4.1

GPUArrays v8.4.1

Diff since v8.4.0

Closed issues:

• repmat and repeat (#93)
• Kernel for repeat and its derivative (#125)
• Implement copy! (#267)

Merged pull requests:

• Add n-dimensional repeat (#400) (@awadell1)
• Specialize Base.copy! for AbstractGPUArrays (#414) (@vpuri3)
• Specialize LinearAlgebra.mul! on Diagonal for GPUArrays (#416) (@vpuri3)

v8.4.0

GPUArrays v8.4.0

Diff since v8.3.2

Closed issues:

• opnorm wanted (#396)
• Broadcast overloading regression (#404)
• GPUArraysCore? (#409)

Merged pull requests:

• Implement opnorm (#403) (@danielwe)
• Implement isequal (#405) (@danielwe)
• Set :ScalarIndexing to the proper value in @allowscalar (#406) (@danielwe)
• Avoid use of doubles in varm implementation. (#407) (@maleadt)
• Fix flaky tests (#408) (@maleadt)
• Add size check for isequal and == (#410) (@JeffFessler)
• Move core functionality to GPUArraysCore. (#411) (@maleadt)
• CompatHelper: add new compat entry for "Reexport" at version "1" (#412) (@github-actions[bot])
• add conjugate option to copytri! (#413) (@Red-Portal)