Skip to content
This repository has been archived by the owner on May 27, 2021. It is now read-only.

Releases: JuliaGPU/CUDAnative.jl

v2.9.0

30 Jan 14:57
v2.9.0
d6d81d7
Compare
Choose a tag to compare

v2.9.0 (2020-01-30)

Diff since v2.8.1

Closed issues:

  • ERROR: CUDA toolkit at doesn't contain ptxas (#530)

Merged pull requests:

v2.8.1

17 Jan 11:46
v2.8.1
bb0923e
Compare
Choose a tag to compare

v2.8.1 (2020-01-17)

Diff since v2.8.0

Merged pull requests:

v2.8.0

16 Jan 08:45
v2.8.0
Compare
Choose a tag to compare

v2.8.0 (2020-01-16)

Diff since v2.7.0

Closed issues:

  • "recursion is currently not supported" in non-recursive code (#526)
  • test/utils.jl:julia_script should forward --startup-file=no (#511)
  • Co-existence with pre-initialized CUDA (#455)
  • Cannot convert the element type of a CuArray (#408)
  • performance gap with clang (#398)
  • How to explicitly provide CUDA install location? (#375)
  • Use extern llvmcall (#364)

Merged pull requests:

  • Minor exception display changes. (#524) (maleadt)
  • Add a system-wide thread fence to the exception flag setter. (#523) (maleadt)
  • Fixes for Julia 1.4: disable DI on LLVM 8.0. (#522) (maleadt)
  • Limit deptch of nonisbits-explanation tree and work around Julia issue. (#521) (maleadt)
  • Pass additional arguments that aren't captured by julia_cmd(). (#520) (maleadt)
  • Use ccall extern (#519) (maleadt)
  • Prevent stack overflow during init, and use preexisting device. (#518) (maleadt)
  • fix CuArrays.jl #543 - complex exp type instability (#516) (PhilipVinc)

v2.7.0

20 Dec 15:45
v2.7.0
Compare
Choose a tag to compare

v2.7.0 (2019-12-20)

Diff since v2.6.0

Closed issues:

  • blockDim() crashes Julia 1.3 (#512)

Merged pull requests:

v2.6.0

01 Dec 09:49
v2.6.0
1ba404a
Compare
Choose a tag to compare
See github.com/JuliaGPU/CUDAnative.jl/releases/tag/v2.6.0 for release…

v2.5.5

14 Nov 10:48
v2.5.5
Compare
Choose a tag to compare

v2.5.5 (2019-11-14)

Diff since v2.5.4

Closed issues:

  • Sort paths from CUDAapi.find_toolkit() in the descending order? (#498)
  • CUDA version warning throws MethodError: no method matching version() (#497)
  • 'MethodError' message doesn't show 'Closest candidates' (#493)

Merged pull requests:

  • Print a message if the package fails to load. (#499) (maleadt)
  • Use CUDAapi.at-runtime_ccall for calling into potentially missing libraries (#496) (maleadt)

v2.5.4

07 Nov 10:50
v2.5.4
5e73381
Compare
Choose a tag to compare

Improve conditional use of the package.

v2.5.3

06 Nov 12:48
v2.5.3
3e71b47
Compare
Choose a tag to compare

Hotfix for conditional use of Base.inferencebarrier on older versions of Julia.

v2.5.2

06 Nov 10:47
v2.5.2
43a542d
Compare
Choose a tag to compare

v2.5.2 (2019-11-06)

Diff since v2.5.1

Merged pull requests:

  • Reinstate 1.0 compatibility. (#489) (maleadt)
  • Use an inference barrier to delay initialization. (#488) (maleadt)
  • Remove unnecessary joinpath includes. (#487) (maleadt)
  • Make it possible to precompile the runtime library. (#486) (maleadt)

v2.5.1

05 Nov 13:47
v2.5.1
f7e0925
Compare
Choose a tag to compare

v2.5.1 (2019-11-05)

Diff since v2.5.0

Merged pull requests: