[libclc] branch master created (now 90c52d9)
Andreas Beckmann
anbe at moszumanska.debian.org
Tue Sep 8 10:53:26 UTC 2015
This is an automated email from the git hooks/post-receive script.
anbe pushed a change to branch master
in repository libclc.
at 90c52d9 add Vcs-* URLs
This branch includes the following new commits:
new 256deb0 Create libclc subdirectories.
new 6937ba9 Initial commit.
new 813bedf Update repository paths.
new 4600f26 Test web page update.
new c328076 Switch to BSD/MIT dual license.
new 5ea05cf Switch to the NVPTX backend.
new 9b95fc7 Explicit conversions.
new 96eafe4 Add some tests which had heretofore evaded 'git add'.
new 1171804 Add fabs builtin.
new 2f74bd8 Fix typo in double precision case.
new 39a6da0 Implement exp, exp2, log, log2, native_exp, native_exp2, native_log, native_log2. Patch by Joshua Cranmer!
new 33a274a Implement mad builtin.
new 0d9bead Add fma, hypot builtins.
new 54397b8 Define FLOAT in floatn.inc.
new f69df75 Add missing dot.h include.
new 65f1c28 Enable cl_khr_fp64 when building the library, and fix several bugs uncovered when doing so.
new 257a3ba Add pow builtin.
new a568017 configure.py: Add an install rule.
new 05edc47 Fix declarations of __clc_add_sat_*. Patch by Lei Mou!
new 3b230ff Implement sub_sat builtin. Patch by Lei Mou!
new a37e2c0 PTX: move implementations of work-item and synchronisation functions to lib, and add header files in generic. Incorporates a patch by Tom Stellard!
new 833b5f1 Do not use linkonce_odr linkage in .ll files. This prevented them from being linked into the library under lazy linkage.
new 1e8ef6c Add floor builtin. Patch by Cassie Epps!
new a965281 Add rsqrt builtin. Based on patch by Cassie Epps!
new 3290bdd Add barrier.cl to SOURCES, spotted by Jin Wang.
new 5c2bfa6 Add native_powr builtin. Patch by Tom Stellard!
new 14e5a4f Implement any() builtin. Patch by Tom Stellard!
new f07a899 Fix build against recent versions of Clang. Based on patch by Alastair Donaldson!
new c1a7808 Update the copyright coredits -- Happy new year 2013!
new 2b198dc r600: Initial support
new 98d740b r600: Fix get_global_id implementation
new 7601924 r600: Add get_global_size() implementation
new 6af4706 Move R600 headers into generic directory
new 0404dcc PTX: move implementations of work-item and synchronisation functions to lib, and add header files in generic. Incorporates a patch by Tom Stellard!
new 86951a5 R600: Replace cl implementations with LLVM IR implementation
new b234c94 Make libclc more Linux FHS conform.
new 513ac41 Allow targets to override generic implementations
new 655cb98 r600: Add overrides file
new ba36ac1 Fix typo in include/clc/geometric/length.inc
new 35e27a9 Use brackets around include files in length.cl and normalize.cl
new b08be4f Remove the static keyword from the _CLC_INLINE macro
new 0fd3463 Implement fmax() and fmin() builtins
new b77eb2d Implement ceil() builtin
new 58ccf6f Fix build with LLVM 3.3
new 9aba71d configure: fix out-of-source build
new 7dc9242 configure: Enable building separate libraries for target variants
new 485cf99 libclc: Add max() builtin function
new 56a70fa libclc: Fix abs_diff builtin integer function
new 213e719 libclc: Add clamp() builtin for integer/floating point
new 15ba684 libclc: Move max builtin to shared/
new f1bf4b8 libclc: implement rotate builtin
new 160ef13 Simplify rotate implementation a bit..
new a6c4ea3 Add a TODO note.
new 7a2f843 Add a another TODO note.
new 547d451 libclc: Rename [add|sub]_sat.ll to [add|sub]_sat_if.ll
new a8b6a9c libclc: implement initial version of min()
new b051410 libclc: Implement the min(vec, scalar) version of the min builtin.
new 7424a79 libclc: Add clamp(vec, scalar, scalar) and max(vec, scalar)
new 80f66b1 libclc: Implement clz() builtin
new 5c45e8a r600: Fix implementations of get_group_id.ll and get_local_size.ll
new 6b73fd4 libclc: Initial vload implementation
new 1a71238 libclc: Initial vstore implementation
new eadd80c libclc: Add assembly versions of vload for global int4/8/16
new 4ab9036 libclc: Add assembly versions of vstore for global [u]int4/8/16
new a3c59c9 Add bitselect() builtin
new efd6599 Implement barrier() builtin
new 5217211 Add __CLC_ prefix to all macro definitions in headers
new 045f1a8 Implement mad24() and mul24() builtins
new ce72a50 Add integer-gentype.inc: Missing file from r185839
new cfdac80 libclc: vload/vstore disable assembly and fix offset calculation
new 2b80a46 Fix and re-enable R600 vload/vstore assembly
new 2e8fa9f Fix build with LLVM 3.4
new 1489907 Implement generic upsample()
new 5c0b162 Added get_num_groups
new aa69869 Add missing integer min/max definitions
new c2bb465 configure: Fix build when clang is installed to a non-standard prefix
new 831732e Implement generic rint()
new 2676aff Add some missing convert_* functions
new 873135f Add vload* for addrspace(2) and use as constant load for R600
new d0438a7 Enable assembly vload3 int/uint constant/global for R600
new 227cc19 Add intN vloadN() implementations for address spaces 3 and 4
new f1a5c55 Add hadd builtin
new ebb0f60 Add rhadd builtin
new 14aec1e Add mul_hi implementation [v2]
new 2b5e8f2 Imported Debian patch 0~git20130819-1
new 795f6b0 Add atomic_inc and atomic_add builtins
new bfd2647 Remove unneeded semi-colons
new 8ddebf1 Place pkg-config file in $prefix/share/pkgconfig.
new d00ce85 Add atomic_sub and atomic_dec builtin functions
new 31238f4 Implement mad_hi built-in
new 89e5b14 Parenthesize arguments for mad_hi
new 21f2751 Imported Debian patch 0~git20130819-2
new bb01526 Add missing as_{float,double} functions
new 6c1a253 Implement isnan() builtin
new fbc19a7 Implement nextafter() builtin
new c002f62 Implement sign() builtin
new 9c215f7 Port pocl's gen_convert.py script to libclc
new ecdfb59 Make C++ compiler configurable
new 4c18120 Clean-up dependency files
new 6e12753 R600: Set the noduplicate attribute on barrier() intrinsics
new 499ed9d s/_CLC_DECL/_CLC_DEF/
new 65a950a Implement builtins for cl_khr_global_int32_base_atomics extension
new dc25692 Implement round builtin
new 9844fd6 R600: Add aliases for Sea Islands GPUs
new 8e0ea3a Fix a C&P error in r195021 (65a950abab3cb8435ccb2646ac4773986c995c81)
new 9173b5d Implement trunc builtin.
new b5e3c57 Add floating-point macro definitions v2
new 799014b Fix build with LLVM 3.5
new 5631a95 Pass -fno-builtin flag to clang to silence warnings
new 0e431f5 Update the copyright credits -- Happy new year 2014!
new 288d2da Merge tag 'upstream/0_git20140101'
new 51247bb Imported Debian patch 0~git20140101-1
new d9a6545 Fix build broken by LLVM commit r199279
new 5466ed1 Enforce python2 for systems that use python3 as their default.
new 7836a02 Fixed ninja build issues relating to use of $(DESTDIR)
new b5d4321 Fixed rules names so they are unique when aliases are present.
new 9f8cb14 Updated README.TXT with information about using DESTDIR and building with Ninja.
new 7376e31 Revert "Enforce python2 for systems that use python3 as their default."
new 5888f70 Add generic nvptx targets
new f5dec6f Fix build since r202052
new 5389458 Add cross for double3 and double4
new 1e278a7 Add sincos
new 0dd1f0b Add definition for M_PI_F v3
new 4341094 Replace tabs by spaces
new a112a9a Introduce M_LOG2E_F and M_LOG2E
new 9645a58 Remove clc/gentype.inc
new 9908d94 Fix build broken by LLVM commit r207593
new ae29487 Add definition for M_PI
new 539fe70 Fix _F definitions
new 616f124 Add more log related float constants
new 61dedf2 Fix build broken by LLVM commit r209103
new 8486107 Remove unused include which breaks build after r210803
new d24beb9 prepare-builtins: Use std:: prefix for error_code
new 6869202 Implementations for exp(float) and exp(double) v2
new 956f5b8 Add files forgotten in the previous commit
new b55fc87 Add intptr types
new 4cf021a clctypes.h: Don't rely on stddef.h for size_t and ptrdiff_t
new 644b73d Add all(igentype) builtin
new 7312d40 relational: Add isequal(floatN) builtin
new 1d6ae46 math: Implement mix builtin
new 21abf5e Revert "clctypes.h: Don't rely on stddef.h for size_t and ptrdiff_t"
new 74feee0 Add remaining float constants
new 850f357 Fix definition of INFINITY and add NAN/HUGE_VAL[F]
new 1573d8d Add missing undefs
new 22a202d Add pown
new a2daa48 Fix breakage after r211259
new e32fc55 Protect functions taking double by #ifdef cl_khr_fp64
new f6ad9d6 Move clcmacro.h to avoid cluttering user namespace v2
new a975af1 Add MAXFLOAT
new e06ebb8 Introduce CLC_VERSION macros v2
new 39185f3 Add half limits
new 86e401d Add exp10
new a8f5e35 relational: Implement signbit
new cce042c relational: Fix signbit
new 61127c5 OpenCL 1.1 does not define CL_VERSION_1_2 so use hardcoded number instead
new 0ec7437 prepare-builtins: Fix broken build due to recent LLVM API change
new 4c60824 relational: create re-usable macros for relational declarations
new ad77ffa Fix isnan definition for vector results
new 6b3b668 relational/signbit: Refactor to use relational macros
new 3ce0451 relational: Implement isgreater
new d5efed4 relational: Implement isgreaterequal
new 211d52c relational: Implement isnotequal
new 3b85337 Add several missing double constant definitions
new 1eda8fd Implement atan builtin
new cdbf566 Implement atan2 builtin
new 3c24d27 Implement cos builtin for float types
new a63df06 Implement sin builtin for float types
new 45ca8f6 relational: Add isless(floatN) builtin
new 5b48f17 relational: Add islessequal(floatN) builtin
new 7bd4852 vload/vstore: Use casts instead of scalarizing everything in CLC version
new bd22e6e R600: Add aliases for hainan and mullins
new 24a7d52 Implement prefetch builtin
new fd18b63 Add missing file from r216127
new 0492901 Include llvm-config.h instead of config.h
new 37216c7 Implement generic mad_sat
new ecdc105 Add int3/uint3 to integer-gentype.inc
new 78264ec Revert "Implement generic mad_sat"
new 2d5b894 Fix build against LLVM SVN >= r216393
new e822ae3 Fix build against LLVM SVN >= r216488
new b0fd7d5 Imported Debian patch 0~git20140101-2
new 6582457 configure: Add rpath to prepare-builtins util
new 7dc59ac Implement generic mad_sat
new c5564bc Fix implementation of copysign
new a0e8931 Implement isinf builtin
new ac633e1 add isfinite builtin
new c320ba5 add isnormal builtin
new 443e5ff add islessgreater builtin
new 64a3edf add isunordered builtin
new aba7f77 add isordered builtin
new 0e5bdb5 Imported Debian patch 0~git20140101-3
new 2b6cbed math: Add acos implementation
new 8aa236d math: Add asin implementation
new bdf9234 math: Add tan implementation
new e5b7ee4 atomic: define extension functions for existing atomic implementations
new 438e670 atomic: Add generic implementation of atom[ic]_max
new eed5b02 atomics: Add generic atom[ic]_and
new 407fe6f atomic: Add atom[ic]_or
new 16f0a63 atomic: Add generic atom[ic]_xor
new 17340b2 atomic: Add generic atomic_min implementation
new eb186e0 atomic: Implement generic atom[ic]_xchg
new fb4a04c atomic: Add generic atom[ic]_cmpxchg
new 4edaa87 R600: Map atomic_and address spaces
new be0f545 R600: Map address spaces for atomic_or
new 8cea45b R600: Map addr spaces and use atomic_max
new cd623d3 R600: Map address spaces for atomic_xor
new 2d453cc R600: Map address spaces for atomic_min
new d531877 R600: Map address spaces for atomic_xchg
new f33df1b R600: Map Address spaces for atomic_cmpxchg
new b659d9c Remove redundant semi-colons
new 82b919f atomic: undef macros that are included from atomic_decl.inc
new 8d5e696 Remove more redundant semi-colons
new 0677387 Implement wait_group_events builtin v2
new 7de7dcd Implement async_work_group_strided_copy builtin v2
new 7f6f5bf Implement async_work_group_copy builtin v3
new 254d4dc Implement fmod
new c8ca9c5 Implement log1p builtin
new c1553c3 Imported Debian patch 0~git20140101-4
new 2e96770 r600: Use llvm intrinsic to read work dimension information
new 287f862 r600: Fix get_work_dim range metadata
new ac887c4 Prune CRLF.
new f552032 Imported Debian patch 0~git20140101-5
new 2290645 Don't include <stddef.h>
new acd73c5 Remove wrong semi-colons
new 9a53500 Require LLVM 3.6 and bump version to 0.1.0
new 0bb35f1 r600: get_work_dim: Update metadata syntax for LLVM 3.6
new d73f33b Use amdgcn triple for SI+ GPUs
new 5cd2688 Implement log10
new 60b4217 libclc/math: Add cospi
new 3992fca Implement degrees builtin v2
new e0d14c1 Implement radians builtin v2
new 406b779 Implement smoothstep builtin v2
new c3ad657 Implement step builtin
new d474a13 Move mix from math to common
new 4aebf69 Fix bitselect for float/double types v2
new d0dae62 math: Implement erfc
new 8f0b94b Add __clc_ prefix to functions in sincos_helpers.cl
new d71e4a9 Fix implementation of length builtin v2
new fb3fd25 Implement distance builtin v2
new 6ce5735 Implement half_sqrt builtin v2
new c7c85a1 Implement fast_length builtin
new 0a2d161 Implement fast_distance builtin
new cbc4559 Implement fmin using __builtin_fmin
new b00c16b Implement fmax using __builtin_fmax
new 100d73f Implement acospi builtin
new 8db931c Implement asinh builtin
new 849d9e3 Implement asinpi builtin
new f1370e8 Implement atanpi builtin
new 40c4ddf Implement acosh builtin
new 69ebc81 Implement atanh builtin
new b096501 configure: Add --enable-runtime-subnormal option
new 360d372 Implement fract builtin
new fb4cee5 Fix compilation warnings without cl_khr_fp64
new 5a4891d Allow compilation depending to the LLVM version
new e7c2400 Fix implementation of normalize builtin
new 986df67 Implement ldexp for R600/SI
new 30b7949 math: Add ldexp implementation
new 301372a Implement sinpi builtin
new f2a34e6 Move ldexp soft implementation to a separate file
new a345263 r600: Use __clc_ldexp on asics that don't implement the intruction
new 9795997 Implement half_rsqrt builtin v3
new 9dcfd36 Implement fast_normalize builtin v4
new 8354ce1 Fix ldexp fp64 build error
new b25cbb3 geometric: Limit fast_{distance,length} functions to single precision
new 4b96333 math: limit half_sqrt to single precision
new 2ed6939 Implement atan2 for doubles
new 16b7abe Implement atan2pi builtin
new 6657f6d Implement cos for double types
new 538329d Implement sin for double types
new 6f8227e Implement exp2 using OpenCL C rather than using an intrinsic
new 4f09ca6 Use a more accurate implementation for exp
new 403ad25 prepare-builtins: Fix build with LLVM 3.7
new fe3a883 Properly initialize Module pointer
new eed524f prepare-builtins: Fix build with LLVM 3.6
new 2f3343a Fix implementation of sqrt v2
new dd1ea7f R600: Implement accurate double precision sqrt v2
new 3eeded3 Use llvm intrinsics for native_log and native_log2
new 2b6e032 Implement accurate log2 function
new 7958b02 Fix double implementation of log
new f97d9db Require LLVM >=3.7 and bump version to 0.2.0
new 63f5af3 Remove files accidentally not removed in r244310
new 8aa0613 Merge branch 'upstream'
new b58c397 update the changelog
new f1b85a9 control: Bump llvm build-dep to 3.7.
new 65441c8 Remove clang3.5.patch, refresh system-clang.patch.
new 911d7a0 further changes for llvm 3.7
new ef75ec6 libclc-amdgcn: Add a new subpackage for amdgcn. (Closes: #788219)
new 09e907f Remove system-clang.patch.
new 4493363 Add supported GPU families to description.
new 86896be Bump standards version to 3.9.6 (no changes needed).
new 182e323 Fix Lintian warning: invalid-short-name-in-dep5-copyright bsd.
new 90c52d9 add Vcs-* URLs
The 291 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