[libclc] annotated tag upstream/0.2.0+git20160907 created (now 935d861)
Andreas Boll
aboll-guest at moszumanska.debian.org
Thu Oct 6 08:16:39 UTC 2016
This is an automated email from the git hooks/post-receive script.
aboll-guest pushed a change to annotated tag upstream/0.2.0+git20160907
in repository libclc.
at 935d861 (tag)
tagging 88b82a6f70012a903b10dfc1e2304d3ef2e76dbc (commit)
replaces upstream/0.2.0+git20150813
tagged by Andreas Boll
on Wed Sep 7 15:26:07 2016 +0200
- Log -----------------------------------------------------------------
upstream 0.2.0+git20160907
Aaron Watry (6):
integer: Update integer limits to comply with spec
integer: remove explicit casts from _MIN definitions
math: Add frexp ported from amd-builtins
math: Add ilogb ported from amd-builtins
math: Fix ilogb(double) return type
math: Add fdim implementation
Jan Vesely (14):
math: Fix log2 vectorization on non-fp64 hw
configure: Introduce per device defines
configure: Remove cl_khr_fp64 for device that don't support doubles
configure: Remove llvm 3.6 defines
math: Add erf ported from amd-builtins
math: Use single precision fmax in sp path
64 bit integers are legal in full profile without an extension
nvptx: Drop feature defines.
configure: Remove device specific defines
ptx: Fix builtin names after clang r274770
AMDGPU: Use clang intrinsics for workitem builtins
AMDGPU: Implement get_global_offset builtin
Make min follow the OCL 1.0 specs
Implement vstore_half{,n}
Konstantin Zhuravlyov (1):
[AMDGPU] Implement get_local_size for amdgcn--amdhsa triple
Matt Arsenault (14):
Split sources for amdgcn and r600
Update page to list supported targets
amdgcn: Use new workitem intrinsics
Add .gitignore for build directories
Replace llvm.AMDGPU.ldexp with llvm.amdgcn.ldexp
R600: Use new barrier intrinsic
amdgpu: Use right builtn for rsq
amdgcn: Correct return types to be size_t
amdgcn: Fix get_local_size IR return type
amdgpu: Fix default case value for get_local_size
amdgcn: Fix return type for get_global_size
amdgcn: Also correct get_local_size type for HSA
Strip opencl.ocl.version metadata
amdgcn: Fix return type of get_num_groups
Niels Ole Salscheider (2):
Implement tanh builtin
configure.py: Add polaris10 and polaris11
Paul Robinson (1):
Update copyright year to 2016.
Peter Collingbourne (1):
Update mailing list reference.
Tom Stellard (13):
Add image attribute getter builtins
r600: Add image reading builtins.
r600: Add image writing builtins.
Add image attribute defines.
Add sampler defines.
AMDGPU: Add alias for tonga
AMDGPU: Add aliases for all VI targets
Add _CLC_V_V_VP_VECTORIZE macro
Implement modf math builtin
prepare-builtins: Remove call to getGlobalContext()
geometric/floatn.inc: Add vec8 and vec16 types
Implement cosh builtin
Implement cbrt builtin
-----------------------------------------------------------------------
This annotated tag includes the following new commits:
new 4346c30 integer: Update integer limits to comply with spec
new a559ddd Add image attribute getter builtins
new 125bc99 r600: Add image reading builtins.
new 9168d34 r600: Add image writing builtins.
new e0232de Add image attribute defines.
new 49df8a0 Add sampler defines.
new a93fb25 Implement tanh builtin
new 6d78407 integer: remove explicit casts from _MIN definitions
new 1b1b5a8 AMDGPU: Add alias for tonga
new b858eb3 AMDGPU: Add aliases for all VI targets
new eaab502 Add _CLC_V_V_VP_VECTORIZE macro
new dc330a3 Implement modf math builtin
new bd61e82 math: Add frexp ported from amd-builtins
new c59e018 math: Fix log2 vectorization on non-fp64 hw
new b72f405 configure: Introduce per device defines
new 79830c2 configure: Remove cl_khr_fp64 for device that don't support doubles
new b518692 configure: Remove llvm 3.6 defines
new ba01128 Split sources for amdgcn and r600
new 03d52f1 Update page to list supported targets
new ba9858c amdgcn: Use new workitem intrinsics
new 3f4daf0 Add .gitignore for build directories
new f08376f math: Add ilogb ported from amd-builtins
new d2b49fa math: Fix ilogb(double) return type
new d5dbee6 Update copyright year to 2016.
new 20d977a [AMDGPU] Implement get_local_size for amdgcn--amdhsa triple
new fea4078 prepare-builtins: Remove call to getGlobalContext()
new 9fee821 math: Add fdim implementation
new 78dd721 math: Add erf ported from amd-builtins
new 8243e97 math: Use single precision fmax in sp path
new ca880ed 64 bit integers are legal in full profile without an extension
new ca7993b nvptx: Drop feature defines.
new 802a4a1 configure: Remove device specific defines
new ffb74a4 Replace llvm.AMDGPU.ldexp with llvm.amdgcn.ldexp
new d6e912e R600: Use new barrier intrinsic
new a3d6cec amdgpu: Use right builtn for rsq
new 3d39fb5 ptx: Fix builtin names after clang r274770
new ff45572 AMDGPU: Use clang intrinsics for workitem builtins
new e3cd5bd AMDGPU: Implement get_global_offset builtin
new 838e0fa geometric/floatn.inc: Add vec8 and vec16 types
new 99539f4 Implement cosh builtin
new 785bfd3 Implement cbrt builtin
new 693817b Make min follow the OCL 1.0 specs
new 1b4b553 Implement vstore_half{,n}
new a28ef48 amdgcn: Correct return types to be size_t
new f5b81cf amdgcn: Fix get_local_size IR return type
new 3aa4eea amdgpu: Fix default case value for get_local_size
new 0f2ddaf amdgcn: Fix return type for get_global_size
new b674069 amdgcn: Also correct get_local_size type for HSA
new 1b2f5e0 Strip opencl.ocl.version metadata
new 6c3f6de amdgcn: Fix return type of get_num_groups
new 88b82a6 configure.py: Add polaris10 and polaris11
The 51 revisions listed above as "new" are entirely new to this
repository and will be described in separate emails. The revisions
listed as "adds" were already present in the repository and have only
been added to this reference.
--
Alioth's /usr/local/bin/git-commit-notice on /srv/git.debian.org/git/pkg-opencl/libclc.git
More information about the Pkg-opencl-commits
mailing list