Add initial support for half precision builtinsv2: fix fmax implementation use consistent checks for __CLC_FP_SIZE add missing TODOs fix whitespace in definitions.hv3: undef ZERO in mo
Add initial support for half precision builtinsv2: fix fmax implementation use consistent checks for __CLC_FP_SIZE add missing TODOs fix whitespace in definitions.hv3: undef ZERO in modf.incSigned-off-by: Jan Vesely <[email protected]>reviewer: Jeroen Ketema <[email protected]>Reviewed-by: Aaron Watry <[email protected]>Tested-by: Aaron Watry <[email protected]>llvm-svn: 332677
show more ...
Move cl_khr_fp64 exntension enablement to gentype include listsThis will make adding cl_khr_fp16 support easierReviewed-by: Aaron Watry <[email protected]>Signed-off-by: Jan Vesely <jan.vesely@ru
Move cl_khr_fp64 exntension enablement to gentype include listsThis will make adding cl_khr_fp16 support easierReviewed-by: Aaron Watry <[email protected]>Signed-off-by: Jan Vesely <[email protected]>llvm-svn: 326816
geometric: geometric functions are only supported for vector lengths <=4Signed-off-by: Jan Vesely <[email protected]>Reviewed-by: Aaron Watry <[email protected]>llvm-svn: 314545
geometric/floatn.inc: Add vec8 and vec16 typesllvm-svn: 276495
geometric: Limit fast_{distance,length} functions to single precisionSigned-off-by: Jan Vesely <[email protected]>Reviewed-by: Tom Stellard <[email protected]>llvm-svn: 236940
Implement fast_normalize builtin v4This implementation was ported from the AMD builtin libraryand has been tested with piglit, OpenCV, and the ocl conformance tests.v2: - Remove f suffix from
Implement fast_normalize builtin v4This implementation was ported from the AMD builtin libraryand has been tested with piglit, OpenCV, and the ocl conformance tests.v2: - Remove f suffix from constant in double implementations. - Consolidate implementations using the .cl/.inc approach.v3: - Use __CLC_FPSIZE instead of __CLC_FP{32,64}v4 (Jan Vesely): - Limit to single precision.llvm-svn: 236920
Implement fast_distance builtinThis implementation was ported from the AMD builtin libraryand has been tested with piglit, OpenCV, and the ocl conformance tests.llvm-svn: 232978
Implement fast_length builtinThis implementation was ported from the AMD builtin libraryand has been tested with piglit, OpenCV, and the ocl conformance tests.llvm-svn: 232977
Implement distance builtin v2This implementation was ported from the AMD builtin libraryand has been tested with piglit, OpenCV, and the ocl conformance tests.v2: - Remove unnecessary copyrigh
Implement distance builtin v2This implementation was ported from the AMD builtin libraryand has been tested with piglit, OpenCV, and the ocl conformance tests.v2: - Remove unnecessary copyright.llvm-svn: 232964
Add cross for double3 and double4Patch by: Jeroen KetemaReviewed-by: Tom Stellard <[email protected]>llvm-svn: 204477
Add __CLC_ prefix to all macro definitions in headerslibclc was defining and undefing GENTYPE and several other macros withcommon names in its header files. This was preventing applications from
Add __CLC_ prefix to all macro definitions in headerslibclc was defining and undefing GENTYPE and several other macros withcommon names in its header files. This was preventing applications fromdefining macros with identical names as command line arguments to thecompiler, because the definitions in the header files were masking themacros defined as compiler arguements.Reviewed-by: Aaron Watry <[email protected]>llvm-svn: 185838
Fix typo in include/clc/geometric/length.incllvm-svn: 184984
Add missing dot.h include.llvm-svn: 157615
Define FLOAT in floatn.inc.llvm-svn: 157614
Initial commit.llvm-svn: 147756