Commit Graph

332 Commits

Author SHA1 Message Date
Tom Stellard 6b195ece57 configure: Add amdgcn-mesa-mesa3d target
llvm-svn: 281793
2016-09-16 22:43:33 +00:00
Tom Stellard f19cf403c4 amdgcn-amdhsa: Add get_num_groups implementation
llvm-svn: 281792
2016-09-16 22:43:31 +00:00
Tom Stellard e7ad23bad3 amdgcn-amdhsa: Add get_global_size() implementation
llvm-svn: 281791
2016-09-16 22:43:29 +00:00
Aaron Watry af569547fa math: Implement tgamma
Signed-off-by: Aaron Watry <awatry@gmail.com>
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 281566
2016-09-15 00:17:34 +00:00
Aaron Watry e9009cdd21 math: Implement lgamma
Just use lgamma_r and ignore the value returned in the second argument

Signed-off-by: Aaron Watry <awatry@gmail.com>
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 281565
2016-09-15 00:17:31 +00:00
Aaron Watry 0ab07e1bde math: Implement lgamma_r
Ported from the amd-builtins branch, which is itself based on the
Sun Microsystems implementation.

Signed-off-by: Aaron Watry <awatry@gmail.com>
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 281564
2016-09-15 00:17:28 +00:00
Aaron Watry f969413a82 Add ADDR_SPACE parameter to _CLC_V_V_VP_VECTORIZE
This macro is currently unused, but I plan to use it shortly.

The previous form did casts of pointers without an address space, which
doesn't work so well for CL 1.x.

Signed-off-by: Aaron Watry <awatry@gmail.com>
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 281563
2016-09-15 00:17:22 +00:00
Matt Arsenault fbfd828d2a Replace nextafter implementation
This one passes conformance.

llvm-svn: 280961
2016-09-08 16:37:56 +00:00
Jan Vesely eade17271a Avoid ambiguity in calling atom_add functions.
clang (since r280553) allows pointer casts in function overloads,
so we need to disambiguate the second argument.

clang might be smarter about overloads in the future
see https://reviews.llvm.org/D24113, but let's be safe in libclc anyway.

llvm-svn: 280871
2016-09-07 22:11:02 +00:00
Niels Ole Salscheider 63f71057c0 configure.py: Add polaris10 and polaris11
llvm-svn: 280121
2016-08-30 18:00:41 +00:00
Matt Arsenault 958fce3192 amdgcn: Fix return type of get_num_groups
llvm-svn: 279723
2016-08-25 07:31:40 +00:00
Matt Arsenault 7ef7e6aacd Strip opencl.ocl.version metadata
This should be uniqued when linking, but right now it creates
a lot of metadata spam listing the same version. This should also
probably be reporting the compiled version of the user program,
which may differ from the library. Currently the library IR files report
1.0 while 1.1/1.2 are the default for user programs.

llvm-svn: 279692
2016-08-25 00:25:10 +00:00
Matt Arsenault d0a275228e amdgcn: Also correct get_local_size type for HSA
llvm-svn: 279656
2016-08-24 19:11:52 +00:00
Matt Arsenault 26d9c41ff6 amdgcn: Fix return type for get_global_size
llvm-svn: 279644
2016-08-24 17:52:04 +00:00
Matt Arsenault 314364cbd2 amdgpu: Fix default case value for get_local_size
llvm-svn: 279359
2016-08-20 04:17:17 +00:00
Matt Arsenault 220268d177 amdgcn: Fix get_local_size IR return type
llvm-svn: 279350
2016-08-20 00:01:21 +00:00
Matt Arsenault 2ce3d94a01 amdgcn: Correct return types to be size_t
llvm-svn: 279343
2016-08-19 22:49:39 +00:00
Jan Vesely ad8672727c Implement vstore_half{,n}
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 278962
2016-08-17 20:02:11 +00:00
Jan Vesely 4c59714a52 Make min follow the OCL 1.0 specs
OpenCL 1.0: "Returns y if y < x, otherwise it returns x. If x *and* y
are infinite or NaN, the return values are undefined."

OpenCL 1.1+: "Returns y if y < x, otherwise it returns x. If x *or* y
are infinite or NaN, the return values are undefined."

The 1.0 version is stricter so use that one.

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 276704
2016-07-25 22:36:22 +00:00
Tom Stellard d835b3f1af Implement cbrt builtin
This implementation was ported from the AMD builtin library
and has been tested with piglit, OpenCV, and the ocl conformance tests.

llvm-svn: 276497
2016-07-22 23:45:15 +00:00
Tom Stellard 9cb070f96a Implement cosh builtin
This implementation was ported from the AMD builtin library
and has been tested with piglit, OpenCV, and the ocl conformance tests.

llvm-svn: 276496
2016-07-22 23:45:13 +00:00
Tom Stellard ff13926a60 geometric/floatn.inc: Add vec8 and vec16 types
llvm-svn: 276495
2016-07-22 23:45:11 +00:00
Jan Vesely a82e080b57 AMDGPU: Implement get_global_offset builtin
Also fix get_global_id to consider offset
No idea how to add this for ptx, so they are stuck with the old get_global_id
implementation.

v2: split to a separate patch

v3: Switch R600 to use implictarg.ptr

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 276443
2016-07-22 17:24:24 +00:00
Jan Vesely 74f02db922 AMDGPU: Use clang intrinsics for workitem builtins
v2: split into 2 patches
    use clang builtins for other intrinsics as well

v3: Fix warnings
    Switch r600 to use implictarg.ptr

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 276442
2016-07-22 17:24:20 +00:00
Jan Vesely 7846c9b8f0 ptx: Fix builtin names after clang r274770
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Acked-By: Aaron Watry <awatry@gmail.com>
llvm-svn: 276423
2016-07-22 15:00:08 +00:00
Matt Arsenault 633d749da7 amdgpu: Use right builtn for rsq
The r600 path has never actually worked sinced double is not implemented
there.

llvm-svn: 276009
2016-07-19 19:02:01 +00:00
Matt Arsenault 1ab0d9c1ee R600: Use new barrier intrinsic
llvm-svn: 275874
2016-07-18 18:42:17 +00:00
Matt Arsenault b456c6dd56 Replace llvm.AMDGPU.ldexp with llvm.amdgcn.ldexp
It didn't really work on r600 to begin with, which should
get its own intrinsic.

llvm-svn: 275813
2016-07-18 16:42:50 +00:00
Jan Vesely e97deffb6a configure: Remove device specific defines
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-by: Tom Stellard <tom@stellard.net>
llvm-svn: 273044
2016-06-17 20:30:50 +00:00
Jan Vesely 5fd84d028d nvptx: Drop feature defines.
This is now handled by clang

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-by: Tom Stellard <tom@stellard.net>
llvm-svn: 273043
2016-06-17 20:30:49 +00:00
Jan Vesely 3317f253de 64 bit integers are legal in full profile without an extension
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-by: Tom Stellard <tom@stellard.net>
llvm-svn: 273042
2016-06-17 20:30:41 +00:00
Jan Vesely 973c1fa5f5 math: Use single precision fmax in sp path
Fixes fdim piglit on Turks

v2: use CL fmax instead of __builtin

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-by: Tom Stellard <tom.stellard@amd.com>
llvm-svn: 269807
2016-05-17 19:44:01 +00:00
Jan Vesely c374cb76f4 math: Add erf ported from amd-builtins
The scalar float/double function bodies are a direct copy/paste,
aside from the removed (optional) code in float function body that
requires subnormals.

reviewers: jvesely

Patch by: Vedran Miletić <rivanvx@gmail.com>

llvm-svn: 268766
2016-05-06 18:02:30 +00:00
Aaron Watry 55a8e0fd6d math: Add fdim implementation
Based on the amd-builtin, but explicitly vectorized for all sizes (not just
float4), and includes a vectorized double implementation.

Passes piglit (float) tests on pitcairn.

Signed-off-by: Aaron Watry <awatry@gmail.com>
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 268708
2016-05-06 03:34:45 +00:00
Tom Stellard 6cb18a09b1 prepare-builtins: Remove call to getGlobalContext()
This function has been removed from LLVM.

Patch By: Laurent Carlier

llvm-svn: 266430
2016-04-15 14:18:58 +00:00
Konstantin Zhuravlyov f8a81f869f [AMDGPU] Implement get_local_size for amdgcn--amdhsa triple
Differential Revision: http://reviews.llvm.org/D18284

llvm-svn: 265713
2016-04-07 19:54:19 +00:00
Paul Robinson 36a6eaeb34 Update copyright year to 2016.
llvm-svn: 264949
2016-03-30 22:39:03 +00:00
Aaron Watry 09f3c99a86 math: Fix ilogb(double) return type
Signed-off-by: Aaron Watry <awatry@gmail.com>
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 261714
2016-02-24 00:52:15 +00:00
Aaron Watry d6d0454231 math: Add ilogb ported from amd-builtins
The scalar float/double function bodies are a direct copy/paste
with usage of the CLC wrappers to vectorize them.

This commit also adds in the FP_ILOGB0 and FP_ILOGBNAN macros which are
equal to the results of ilogb(0.0f) and ilogb(float nan) respectively.

v2: Add FP_ILOGB0 and FP_ILOGBNAN definitions

Signed-off-by: Aaron Watry <awatry@gmail.com>
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>
v1 Reviewed-by: Tom Stellard <thomas.stellard@amd.com>

llvm-svn: 261639
2016-02-23 14:43:09 +00:00
Matt Arsenault 34766ca34b Add .gitignore for build directories
llvm-svn: 261043
2016-02-17 00:27:31 +00:00
Matt Arsenault 45e6eaaa05 amdgcn: Use new workitem intrinsics
llvm-svn: 261042
2016-02-17 00:27:27 +00:00
Matt Arsenault d4cd67ab9f Update page to list supported targets
llvm-svn: 260778
2016-02-13 01:02:06 +00:00
Matt Arsenault a48e15c6cb Split sources for amdgcn and r600
Most files remain in a common amdgpu directory.

Also switches barriers to to use convergent,
and use llvm.amdgcn.s.barrier.

This now requires 3.9/trunk to build amdgcn.

llvm-svn: 260777
2016-02-13 01:01:59 +00:00
Jan Vesely 6d870d2e84 configure: Remove llvm 3.6 defines
we require llvm 3.7

reviewer: tstellard
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 260304
2016-02-09 22:17:48 +00:00
Jan Vesely 46b7dd99e0 configure: Remove cl_khr_fp64 for device that don't support doubles
Also remove definitions if provided by clang (3.7+)
This halves the size of builtin.opt.{cedar,barts}.bc

reviewer: tstellard
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 260303
2016-02-09 22:17:46 +00:00
Jan Vesely 4f38142feb configure: Introduce per device defines
Make cl_khr_fp64 define per-device.
This patch does not change the generated Makefile (for llvm 3.6, 3.7)

v2: Make the device defines per LLVM version, 'all' for all versions

reviewer: tstellard
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 260302
2016-02-09 22:17:45 +00:00
Jan Vesely 7fbb96b907 math: Fix log2 vectorization on non-fp64 hw
reviewer: tstellard
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 260301
2016-02-09 22:17:42 +00:00
Aaron Watry 8872800eff math: Add frexp ported from amd-builtins
The float implementation is almost a direct port from the amd-builtins,
but instead of just having a scalar and float4 implementation, it has
a scalar and arbitrary width vector implementation.

The double scalar is also a direct port from AMD's builtin release.

The double vector implementation copies the logic in the float vector
implementation using the values from the double scalar version.

Both have been tested in piglit using tests sent to that project's
mailing list.

Signed-off-by: Aaron Watry <awatry@gmail.com>
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 260114
2016-02-08 17:07:21 +00:00
Tom Stellard 37d19875fa Implement modf math builtin
V2: use the reference implementation as suggested by Matt Arsenault

Patch By: Pavel Ondračka

llvm-svn: 258933
2016-01-27 14:52:10 +00:00
Tom Stellard a249f50970 Add _CLC_V_V_VP_VECTORIZE macro
Patch by: Pavel Ondračka

llvm-svn: 258932
2016-01-27 14:52:07 +00:00