Changelog for
libclc-0.2.0+git.20181127-26.1.x86_64.rpm :
Sat Jan 5 13:00:00 2019 aaronpuchertAATTalice-dsl.net
- Update to version 0.2.0+git.20181127, which fixes issues with amdgcn:
* travis: Add cmake build
* Add cmake build system
* r600: Remove empty OVERRIDES file
* amdgcn: Consolidate atomic minmax helpers
* configure: Add target specific asm rule.
* configure: provide llvm_as helper variable
* r600: Add datalayout to image builtin implementation
* Remove redundant OVERRRIDES file
* configure: Provide symlink for amdgcn-mesa3d instead of configure hack
* travis: Check tahiti-amdgcn-mesa-mesa3d.bc
* amdgcn-amdhsa: Convert get_{global,local}_size to clc for all llvm versions
* amdgcn: Move __clc_amdgcn_s_waitcnt definition to clc file
* amdgcn: Convert get_num_groups to clc
* amdgcn: Convert get_global_size to clc
* amdgcn: Convert get_local_size to clc
* r600: Convert barrier to clc
* r600: Convert get_num_groups to clc
* r600: Convert get_global_size to clc
* r600: Convert get_local_size to clc
Fri Oct 12 14:00:00 2018 jimmyAATTboombatower.com
- Update to version 0.2.0+git.20180915:
* configure: Rework support for gfx9+ devices that were added post LLVM 3.9
* .travis: Add llvm-7 build
* .travis: Use source whitelist alias for llvm-6 repository
* amdgcn: Use __constant AS for amdgcn builtins.
* atom: Use volatile pointers for cl_khr_{global,local}_int32_{base,extended}_atomics
* atom: Consolidate cl_khr_{local,global}_int32_{base,extended}_atomics implementation
* atomic: Provide function implementation of atomic_{dec,inc}
* atom: Consolidate cl_khr_int64_{base,extended}_atomics declarations
* atom: Consolidate cl_khr_{local,global}_int32_{base,extended}_atomics declarations
* atomic: Cleanup atomic_cmpxchg header
* atomic: Move define cleanup to shared include
* Update copyright year to 2018.
* r600/fmin: Flush denormals before calling builtin.
* r600/fmax: Flush denormals before calling builtin.
* math/fma: Add fp32 software implementation
* Add initial support for half precision builtins
* rootn: Use denormal path only
* remquo: Flush denormals if not supported
* remquo: Port from amd builtins
* math: Add helper function to flush denormals if not supported.
* clc_sqrt: Reuse unary_decl.inc
* relational/select: Condition types for half are short/ushort, not char/uchar
* log10: Use sw implementation from amd builtins
* powr: Use denormal path only
* pown: Use denormal path only
* pow: Use denormal path only
* amdgcn/fmin: Fix typos that reduced precision
* exp10: Port from amd builtins
* hypot: Port from amd builtins
* select: simplify implementation and fix fp16
* fmod: Port from amd_builtins
* r600: Update datalayout after LLVM r328656
* amdgcn: Update datalayout after LLVM r328656
* remainder: Port from amd builtins
* nan: Implement
* travis: Add build using llvm-6
* amdgcn/fmax: fcanonicalize operands
* amdgcn/fmin: fcanonicalize operands
* amdgcn,popcount: Workaround broken llvm.ctpop intrinsic on some GCN ASICs
* integer/gentype: Add __CLC_VECSIZE macro
* popcount: Provide function implementation rather than intrinsic redirect
* lgamma_r: Move code from .inc to .cl file
* frexp: Reuse types provided by gentype.inc
* select: Add vector implementation
* minmag: Condition variable needs to be the same bitwidth as operands
* maxmag: Condition variable needs to be the same bitwidth as operands
* Move cl_khr_fp64 exntension enablement to gentype include lists
* utils: Adapt to llvm r325155
* amdgcn: Fix build after GDS/const AS swap in r325030
* amdgcn: Fix datalayout after addition of 32bit const AS in r324747
* r600: Fix datalayout after clang r324101
* amdgcn: Fix datalayout after clang r324101
* amdgpu/half_recip: Switch implementation to native_recip
* amdgpu/half_log2: Switch implementation to native_log2
* amdgpu/half_log10: Switch implementation to native_log10
* amdgpu/half_log: Switch implementation to native_log
* amdgpu/half_exp2: Switch implementation to native_exp2
* amdgpu/half_exp10: Switch implementation to native_exp10
* amdgpu/half_exp: Switch implementation to native_exp
* amdgpu/half_sqrt: Switch implementation to native_sqrt
* amdgpu/half_rsqrt: Switch implementation to native_rsqrt
* Add vstore_half_rte implementation
* Add vstore_half_rtp implementation
* Add vstore_half_rtn implementation
* Add vstore_half_rtz implementation
* vstore_half: Consolidate declarations
* vstore_half: Add support for custom rounding functions
* vstore_half: Make sure the helper function is always inline
* half_powr: Implement using powr
* math.h: Use logical operations instead of bit operations for readability
* math.h: Set HAVE_HW_FMA32 based on compiler provided macro
* tanpi: Port from amd_builtins
* tan: Port from amd_builtins
* half_divide: Implement using x/y
* half_tan: Implement using tan
* half_sin: Implement using sin
* half_recip: Implement using 1/x
* half_log2: Implement using log2
* half_log10: Implement using log10
* half_log: Implement using log
* half_exp10: Implement using exp10
* half_exp2: Implement using exp2
* half_exp: Implement using exp
* half_cos: Implement using cos
* half_sqrt: Cleanup implementation
* half_rsqrt: Cleanup implementation
* rootn: Port from amd_builtins
* powr: Port from amd_builtins
* pown: Port from amd_builtins
* pow: Port from amd_builtins
Sat Dec 23 13:00:00 2017 mpluskalAATTsuse.com
- Update to version 0.2.0+git.20171127:
* configure.py: Add gfx900 (Vega, Raven)
* math: Implement minmag
* math: Implement maxmag
* native_powr: Switch implementation to native_exp2 and native_log2
* native_divide: provide function implementation instead of macro
* native_recip: provide function implementation instead of macro
* native_rsqrt: Switch implementation to 1 / native_sqrt
* native_tan: Switch implementation to use native_sin/native_cos
* math: Use precomputed constant for log2(10.0)
* native_exp10: Switch implementation to llvm intrinsic
* native_sqrt: Switch implementation to llvm intrinsic
* native_sin: Switch implementation to llvm intrinsic
* native_cos: Switch implementation to llvm intrinsic
* native_exp2: Switch implementation to llvm intrinsic
* native_exp: Switch implementation to llvm intrinsic
* amdgpu: Add workaround for unimplemented llvm.exp intrinsic
* native_log10: Switch to generic native intrinsic inc file
* native_log: Switch to generic native intrinsic inc file
* native_log2: Switch to generic native intrinsic inc file
Tue Nov 7 13:00:00 2017 mpluskalAATTsuse.com
- Update to version 0.2.0+git.20171102:
* tgamma: Use unary_decl instead of custom inc file
* tanh: Use unary_decl instead of custom inc file
* tan: Use unary_decl instead of custom inc file
* sqrt: Use unary_decl instead of custom inc file
* sinpi: Use unary_decl instead of custom inc file
* sinh: Use unary_decl instead of custom inc file
* sin: Use unary_decl instead of custom inc file
* native_log: Use unary_decl instead of custom inc file
* native_log2: Use unary_decl instead of custom inc file
* native_log10: Use unary_decl instead of custom inc file
* log: Use unary_decl instead of custom inc file
* logb: Use unary_decl instead of custom inc file
* log2: Use unary_decl instead of custom inc file
* log1p: Use unary_decl instead of custom inc file
* lgamma: Use unary_decl instead of custom inc file
* exp2: Use unary_decl instead of custom inc file
* cospi: Use unary_decl instead of custom inc file
* cosh: Use unary_decl instead of custom inc file
* cos: Use unary_decl instead of custom inc file
* cbrt: Use unary_decl instead of custom inc file
* atanpi: Use unary_decl instead of custom inc file
* atanh: Use unary_decl instead of custom inc file
* atan: Use unary_decl instead of custom inc file
* asinpi: Use unary_decl instead of custom inc file
* asinh: Use unary_dec instead of custom inc file
* asin: Use unary_decl instead of custom inc file
* acospi: Use unary_decl instead of custom inc file
* acosh: Use unary_decl instead of custom inc file
* acos: Use unary_decl instead of custom inc file
* math: Implement native_log10
* amdgpu/math: Don\'t use llvm instrinsic for native_log
* shared: Implement aligned vector stores (vstorea_half)
* shared: Implement aligned vector loads (vloada_half)
* amdgcn: Add missing datalayout info to .ll files
* r600: Add missing datalayout to .ll files
* travis: enable checks of nvptx libraries
* travis: Enable external function call checks on llvm-{4,5}
* Make image builtins r600/llvm-3.9 only
* Implement mem_fence on ptx
* Make ptx barrier work irrespective of the cl_mem_fence_flags
* travis: Make sure we report failure even if only earlier checked files fail
* check_external_calls.sh: Print number of calls in tested file.
* ptx: Use __clc_nextafter to implement nextafter
* Do not include clc_nextafter header globally
* math/nextafter: Use custom declaration inc file
* math/binary_decl.inc: Do not declare mixed float/double functions
* ldexp: Fix double precision function return type
* configure: Fix handling of directories with compats only source lists
* Add vload_half helpers for ptx
* Add vstore_half helpers for ptx
* integer/sub_sat: Use clang builtin instead of llvm asm
* integer/add_sat: Use clang builtin instead of llvm asm
* integer/clz: Use clang builtin instead of llvm asm
* Let get_work_dim take exactly 0 arguments
* Do no circularly define NULL
* Fix amdgcn-amdhsa on llvm-3.9
* travis: Check built libraries on llvm-3.9
* Add script to check for unresolved function calls
* geometric: geometric functions are only supported for vector lengths <=4
* travis: add build using llvm-3.9
* Restore support for llvm-3.9
* Add missing HAVE_LLVM define to fix build with latest llvm
* Rework atomic ops to use clang builtins rather than llvm asm
* prepare_builtins: Fix compile breakage with older LLVM
* [Support] Rename tool_output_file to ToolOutputFile, NFC
- Use python3 for building
Thu Sep 21 14:00:00 2017 jimmyAATTboombatower.com
- Update to version 0.2.0+git.20170920:
* generic: add missing get_work_dim include
* add __kernel_exec macros
* configure.py: Make python3 friendly
* configure.py: Drop explicit import of int builtin
* amdgcn: Implement {read_,write_,}mem_fence builtin
* amdgcn: rewrite barrier() using fence and clang __builtin_amdgcn_s_barrier
* Add halfN types and enable fp16 when generating builtin declarations
* relational: Implement shuffle builtin
* relational: Implement shuffle2 builtin
* Fixup clc.h comment
* r600: Cleanup barrier implementation.
* amdgcn,waitcnt: Add datalayout info
* configure.py: Simplify compatibility sources
* vstore: Cleanup and add vstore(half)
* Implement vload_half{,n} and vload(half)
* integer: Add popcount implementation using ctpop intrinsic
* Add native_recip(x) as ((1)/(x))
* Add travis CI configuration file
* Implement cl_khr_int64_base_atomics builtins
* Implement cl_khr_int64_extended_atomics builtins
Wed Apr 12 14:00:00 2017 jimmyAATTboombatower.com
- Update rpmlintrc to include both lib dir .pc files.
Mon Apr 10 14:00:00 2017 jimmyAATTboombatower.com
- Update to version 0.2.0+git.20170225:
* Fix build since llvm r286566 and require at least llvm 4.0
* Fix build since r286752.
* math: Add expm1 builtin function
* math: Add logb builtin
* math: Add native_rsqrt builtin function
* Add the correct prefixes to the cl_khr_fp64 pragma
* Move BufferPtr into the block where it it being used
* math: Add native_tan as wrapper to tan
* .gitignore: Ignore amdgcn-mesa object directory
* math: Implement sinh function
Sun Sep 25 14:00:00 2016 mpluskalAATTsuse.com
- Update to version 0.2.0+git.20160921:
* Avoid ambiguity in calling atom_add functions.
* Replace nextafter implementation
* Add ADDR_SPACE parameter to _CLC_V_V_VP_VECTORIZE
* math: Implement lgamma_r
* math: Implement lgamma
* math: Implement tgamma
* amdgcn-amdhsa: Add get_global_size() implementation
* amdgcn-amdhsa: Add get_num_groups implementation
* configure: Add amdgcn-mesa-mesa3d target
* Provide vstore_half helper to workaround clc restrictions
Sun Jul 3 14:00:00 2016 mpluskalAATTsuse.com
- Update to version 0.2.0+git.20160209:
* integer: remove explicit casts from _MIN definitions
* AMDGPU: Add alias for tonga
* AMDGPU: Add aliases for all VI targets
* Add _CLC_V_V_VP_VECTORIZE macro
* Implement modf math builtin
* math: Add frexp ported from amd-builtins
* 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
Thu Dec 17 13:00:00 2015 cooloAATTsuse.com
- fix license according to legal team
Thu Dec 17 13:00:00 2015 sndirschAATTsuse.com
- modify license to \'BSD-3-Clause or MIT\'
- added LICENSE.TXT file to %doc
Wed Dec 9 13:00:00 2015 mpluskalAATTsuse.com
- Remove unnecessary ldconfig calls
- Rename rpmlintrc to libclc-rpmlintrc
- Minor spec file cleanup
Tue Dec 8 13:00:00 2015 sndirschAATTsuse.com
- used BSD-3-Clause instead of BSD-2-Clause in order to make our
legal team happy
Mon Dec 7 13:00:00 2015 sndirschAATTsuse.com
- added rpmlintrc as source to specfile
Wed Dec 2 13:00:00 2015 jimmyAATTboombatower.com
- Remove devel package in favor of main package since libclc is
unusable without the header files used to compile OpenCL
applications against.
Mon Nov 30 13:00:00 2015 jimmyAATTboombatower.com
- Merge home:X0F:HSF spec changes.
- Set _service file to static revisions in lieu of tags.
- Major cleanup of spec file.
- Apply spec-cleaner.
Fri Feb 3 13:00:00 2012 pontostroyAATTgmail.com
- initial package
