v0.8.2

KernelAbstractions v0.8.2

Diff since v0.8.1

Closed issues:

• Support atomics (#7)
• Support for atomics (#276)

Merged pull requests:

• adding atomic support with atomix (#299) (@leios)
• CompatHelper: add new compat entry for Atomix at version 0.1, (keep existing compat) (#301) (@github-actions[bot])
• Enzyme v0.10 supports GPU compilation again (#303) (@vchuravy)
• Add atomics support for ROCKernels (#304) (@jpsamaroo)
• Enable CPU atomics (#305) (@jpsamaroo)

v0.8.1

KernelAbstractions v0.8.1

Diff since v0.8.0

Closed issues:

• Separate Cassette context from CompilerMetadata (#231)

Merged pull requests:

• WIP: Make CUDA stream from cache match CUDA context (#294) (@christophernhill)
• using CPU as default for Scratchpad and SharedMemory (#300) (@leios)

v0.8.0

KernelAbstractions v0.8.0

Diff since v0.7.2

Closed issues:

• Adding a function to get device from array (type)? (#268)
• CUDA 3.6.3 broke KernelAbstactions. (#280)
• wait(kernel(...) hangs up on Julia v1.7 (#290)

Merged pull requests:

• Add function get_device (#269) (@oschulz)
• Synchronize SpecialFunctions compat (#279) (@charleskawczynski)
• [CUDAKernels] Avoid Cassette looking at device_functions (#281) (@vchuravy)
• Simplify CI and drop bors (#283) (@vchuravy)
• Excise Cassette (#288) (@vchuravy)
• Use Timer to avoid hangs due to error on the device (#291) (@vchuravy)
• don't test KernelGradients on nightly for now (#292) (@vchuravy)

v0.7.2

KernelAbstractions v0.7.2

Diff since v0.7.1

Merged pull requests:

• CompatHelper: bump compat for "SpecialFunctions" to "2.0" (#278) (@github-actions[bot])

v0.7.1

KernelAbstractions v0.7.1

Diff since v0.7.0

Closed issues:

• Compilation error with type-converting round functions, e.g., ceil(Int32, 1.2f0) (#254)
• invalid syntax on upcoming 1.7 (#260)
• ROCKernels run doesn't work for me. (#261)
• Convert related kernels errors with CUDAKernels (#265)
• Event(ROCDevice()) errors (#267)

Merged pull requests:

• add KernelGradients (#255) (@vchuravy)
• Add AMD section to a couple of examples (#266) (@ali-ramadhan)
• Fix Event for multiple event case (#270) (@vchuravy)
• fixing shmem line to match CUDA.jl (#272) (@leios)
• fix to allow for converting to Int (#273) (@leios)

v0.7.0

KernelAbstractions v0.7.0

Diff since v0.6.3

Merged pull requests:

• fix failures on 1.7 (#258) (@simeonschaub)

v0.6.3

KernelAbstractions v0.6.3

Diff since v0.6.2

Merged pull requests:

• nix overdub of _unsafe_getindex (#247) (@vchuravy)
• Stop overdubbing CUDA math functions (#249) (@ali-ramadhan)
• Backport #249 (#251) (@vchuravy)
• Check pow overdubbing (#253) (@vchuravy)

v0.6.2

KernelAbstractions v0.6.2

Diff since v0.6.1

Merged pull requests:

• Remove metadata from cassette context (#244) (@vchuravy)
• nix overdub of _unsafe_getindex (#246) (@vchuravy)

v0.6.1

KernelAbstractions v0.6.1

Diff since v0.6.0

Closed issues:

• [wrong repository] "Spills" from adjacent views of ROCVector (#237)
• Julia compat entry (#238)
• Auto-inbounds and nothing? (#240)

Merged pull requests:

• Use hostcall for wait and stream GC (#85) (@vchuravy)
• [CUDAKernels] add an implicit sync to kernels with no dependencies (#222) (@vchuravy)
• Add some docs for CUDAKernels/ROCKernels (#233) (@jpsamaroo)
• Add new overdub for unsafe_getindex to avoid allocating error message in Julia 1.6+ (#236) (@jakebolewski)
• update to CUDA 3.0 and use task-local stream (#242) (@vchuravy)

v0.6.0

KernelAbstractions v0.6.0

Diff since v0.5.5

Merged pull requests:

• Add ROCKernels backend (#209) (@jpsamaroo)
• Fix Bors (#230) (@DilumAluthge)