llvm-project.git/libclc, branch users/makslevental/ptr-dialectpython Unnamed repository; edit this file 'description' to name the repository. [NFC][libclc] Rename clc_cbrt.inc to clc_cbrt.h (#166330) 2025-11-05T00:20:29+00:00 Wenju He wenju.he@intel.com 2025-11-05T00:20:29+00:00 8648beff758db89e2eb816bd01d9b3c37e6aa3f9 __clc_cbrt functions are declared in clc_cbrt.inc. Rename to .h for consistency with other headers. 
__clc_cbrt functions are declared in clc_cbrt.inc. Rename to .h for
consistency with other headers.
 [libclc] Implement integer __clc_abs using __builtin_elementwise_abs (#164957) 2025-10-27T04:58:14+00:00 Wenju He wenju.he@intel.com 2025-10-27T04:58:14+00:00 a61e016565409b6a6620422dd915e591ea75f998 Previous implementation was cmp, select and @llvm.smax sequence in LLVM IR. __CLC_GEN_U/__CLC_GEN_S is upstreamed from intel/llvm repo. 
Previous implementation was cmp, select and @llvm.smax sequence in LLVM IR.
__CLC_GEN_U/__CLC_GEN_S is upstreamed from intel/llvm repo.
 [NFC][libclc] Improve empty builtins error: include ARCH_SUFFIX in message (#164527) 2025-10-22T04:47:44+00:00 Wenju He wenju.he@intel.com 2025-10-22T04:47:44+00:00 d6b68545aa6249bf4445e0ef96c0134d69a98a26 To clarify which builtin set has no bytecode files. 
To clarify which builtin set has no bytecode files.
 [libclc] Fix double NAN_MASK in __clc_nan (#163522) 2025-10-21T02:43:28+00:00 Wenju He wenju.he@intel.com 2025-10-21T02:43:28+00:00 7a5435351d7cc26e6e3d7d66c3b5317e217521bf 0x7ff0000000000000 is +inf. Change it to quiet nan 0x7ff8000000000000. 
0x7ff0000000000000 is +inf. Change it to quiet nan 0x7ff8000000000000.
 [NFC][libclc] Simplify degrees, radians and smoothstep macros (#164203) 2025-10-20T05:51:18+00:00 Wenju He wenju.he@intel.com 2025-10-20T05:51:18+00:00 54eaf22269ebd9433196602acc152152928f9a0d Implementation doesn't change. 
Implementation doesn't change.
 [libclc] Move functions definition from header clc_sincos_piby4.inc into clc_sincos_helpers.cl (#164028) 2025-10-20T02:08:35+00:00 Wenju He wenju.he@intel.com 2025-10-20T02:08:35+00:00 377148fd60df55e486dc72b26248006ae092725a inline functions defined in clc_sincos_piby4.inc miss static specifier and are deleted by EliminateAvailableExternallyPass when not inlined. This PR fix the problem by removing inline and moving function definition into clc/lib/generic/math/clc_sincos_helpers.cl. It makes sense to put all sin/cos helpers definitions in one file clc_sincos_helpers.cl. 
inline functions defined in clc_sincos_piby4.inc miss static specifier
and are deleted by EliminateAvailableExternallyPass when not inlined.

This PR fix the problem by removing inline and moving function
definition
into clc/lib/generic/math/clc_sincos_helpers.cl. It makes sense to put
all sin/cos helpers definitions in one file clc_sincos_helpers.cl.
 [libclc] Change libclc install dir to ${clang_resource_dir}/lib/libclc in in-tree build (#163896) 2025-10-20T01:47:00+00:00 Wenju He wenju.he@intel.com 2025-10-20T01:47:00+00:00 b9e2f7ac0489b2045aa78475e9663c2d3e5bedf7 Commit df7473673214b placed libclc libraries into clang resource dir <resource-dir>/lib/libclc at build stage. This PR does it at install stage as well. Note that in standalone (not in-tree) build, libclc is still installed to old ${CMAKE_INSTALL_DATADIR}/clc dir. 
Commit df7473673214b placed libclc libraries into clang resource dir
<resource-dir>/lib/libclc at build stage.
This PR does it at install stage as well.
Note that in standalone (not in-tree) build, libclc is still installed
to old ${CMAKE_INSTALL_DATADIR}/clc dir.
 [NFC][libclc] Add missing evaluation for variable ${tool}_target (#163540) 2025-10-16T01:40:09+00:00 Wenju He wenju.he@intel.com 2025-10-16T01:40:09+00:00 754ebc6ebb9fb9fbee7aef33478c74ea74949853 The variabls should be evaluated before checking for empty. 
The variabls should be evaluated before checking for empty.
 [libclc] Remove -fno-builtin from compile options (#162075) 2025-10-07T00:14:54+00:00 Wenju He wenju.he@intel.com 2025-10-07T00:14:54+00:00 64574d3564fc2b4d3ec812a57a414f67558c26e4 The flag was added in 8ef48d07efa3 to suppress build warning and is no longer needed. It adds "no-builtins" attribute, which prevents libclc functions from being inlined into caller that don't have the attribute. The flag is meant to prevent folding standard library calls into optimized implementations. For libclc device targets, however, such target‑driven folding is desirable. llvm-diff shows no change to amdgcn--amdhsa.bc and nvptx--nvidiacl.bc. Co-authored-by: Mészáros Gergely <gergely.meszaros@intel.com> 
The flag was added in 8ef48d07efa3 to suppress build warning and is no
longer needed.

It adds "no-builtins" attribute, which prevents libclc functions from
being inlined into caller that don't have the attribute.

The flag is meant to prevent folding standard library calls into
optimized implementations. For libclc device targets, however, such
target‑driven folding is desirable.

llvm-diff shows no change to amdgcn--amdhsa.bc and nvptx--nvidiacl.bc.

Co-authored-by: Mészáros Gergely <gergely.meszaros@intel.com>
 libclc: Add gfx1250 and gfx1251 to amdgpu target list (#162034) 2025-10-06T01:56:17+00:00 Matt Arsenault Matthew.Arsenault@amd.com 2025-10-06T01:56:17+00:00 6f3d765d04041412e9801187eb261253c3ceb2a1