커밋 그래프

495 커밋

작성자 SHA1 메시지 날짜
Yaxun Sam Liu fc228c7ea6 Fix hipLaunchKernelGGL for hip-clang
Do not decay function pointer type of the kernel argument passed to hipLaunchKernelGGL
and hipLaunchKernel, otherwise some type information is lost which may cause
type inference failure for the template.

This issue caused compilation error of FeatureLPPooling in Caffe2/PyTorch and this patch
fixes that.
2018-09-17 11:20:41 -04:00
Maneesh Gupta 66f863d1f3 Merge branch 'master' into support-malloc 2018-09-17 10:17:25 +05:30
Maneesh Gupta cb348421d7 Merge pull request #650 from ROCm-Developer-Tools/hip-clang-new
Support placement new in hip-clang
2018-09-15 11:21:01 +05:30
Maneesh Gupta 8fe4e22b19 Merge pull request #665 from aaronenyeshi/fix-min-funcs
Use templates for min to prevent ambiguity
2018-09-14 13:21:38 +05:30
Aaron Enye Shi 6b811ca6d1 Fix Tensorflow ambiguous min issue 2018-09-13 23:16:20 +00:00
Maneesh Gupta aed5ad31ba Merge pull request #669 from ROCm-Developer-Tools/feature_automatic_cast
Remove potential for mismatch between runtime passed actuals and defined formals
2018-09-13 07:54:22 +05:30
Maneesh Gupta 411e53a665 Merge pull request #661 from yxsamliu/add-empty-printf
Add empty printf for hip-clang
2018-09-13 07:54:03 +05:30
Aaron Enye Shi 894cbdd749 Avoid AMP-retrict call to CPU-restrict 2018-09-12 14:54:31 +00:00
Alex Voicu cdfea3ef7b Remove potential for mismatch between runtime passed actuals and defined formals. 2018-09-12 10:30:48 +01:00
Maneesh Gupta 8249cf037b Merge pull request #664 from lcskrishna/master
added __host__ to float2half and half2float functions.
2018-09-12 14:50:01 +05:30
Maneesh Gupta 133d665a88 Merge pull request #663 from yxsamliu/fix-launch
Use template for hipLaunchKernelGGL for hip-clang
2018-09-12 14:49:38 +05:30
carlushuang d577f27d1a fix __longlong_as_double() problem, return the double value
previous version return a long long valus *as* double, hence we may get the wrong result.
this also affect atomicAdd(double * ...), which use long long pointer to mimic double pointer.

Signed-off-by: carlushuang <carlus.huang@amd.com>
2018-09-12 13:25:00 +08:00
Aaron Enye Shi ffd89dde9c Avoid host min func conflict with gcc min 2018-09-11 18:48:31 +00:00
Aaron Enye Shi 0121ec13aa Use templates for min to prevent ambiguity 2018-09-11 18:21:54 +00:00
Yaxun Sam Liu 9e9a93e10a Use template for hipLaunchKernelGGL for hip-clang 2018-09-07 16:20:00 -04:00
Yaxun Sam Liu 3e1833ca52 Add empty printf for hip-clang 2018-09-07 16:15:08 -04:00
lcskrishna 2984c020b8 added host for float2half and hlaf2float 2018-09-06 15:02:52 -07:00
Yaxun (Sam) Liu 338eaefa84 Support placement new in hip-clang 2018-09-04 16:53:46 -04:00
Maneesh Gupta 52b8879f12 Merge pull request #654 from ROCm-Developer-Tools/fix-force-inline
__forceinline__ needs inline and always_inline
2018-09-03 16:35:34 +05:30
Maneesh Gupta a3a93a68d4 Merge pull request #657 from yxsamliu/fix-clz
Fix ambiguity of __clz and __clzll
2018-09-03 16:34:42 +05:30
Maneesh Gupta 6b21d77a86 Merge pull request #656 from searlmc1/msearles-fixes
Add keyword 'explicit' to avoid ambiguity
2018-09-03 16:33:41 +05:30
Maneesh Gupta 7459defec1 Merge pull request #651 from yxsamliu/hip-clang-clock
Add clock() and clock64()
2018-09-03 16:32:30 +05:30
Yaxun Sam Liu 11c91c2fdd Fix ambiguity of __clz and __clzll 2018-08-31 18:07:57 -04:00
Mark Searles aa7ce2ef44 Add keyword 'explicit' to avoid ambiguity
Older gcc, e.g., 5.x, see an ambiguity in some calls. Example error
output as seen with gcc 5.5:

In file included from benchmark_wino.cpp:1:
In file included from ./miopen.hpp:13:
In file included from /usr/lib/gcc/x86_64-linux-gnu/5.5.0/../../../../include/c++/5.5.0/regex:38:
In file included from /usr/lib/gcc/x86_64-linux-gnu/5.5.0/../../../../include/c++/5.5.0/algorithm:62:
In file included from /usr/lib/gcc/x86_64-linux-gnu/5.5.0/../../../../include/c++/5.5.0/bits/stl_algo.h:66:
In file included from /usr/lib/gcc/x86_64-linux-gnu/5.5.0/../../../../include/c++/5.5.0/random:51:
/usr/lib/gcc/x86_64-linux-gnu/5.5.0/../../../../include/c++/5.5.0/bits/random.tcc:1324:27: error: call to 'abs' is ambiguous
                    const double __y = -std::abs(__n) * __param._M_sm - 1;
                                        ^~~~~~~~
/opt/rocm/hip/include/hip/hcc_detail/hip_complex.h:345:31: note: candidate function
__DEFINE_HIP_COMPLEX_REAL_FUN(abs, hipCabs)
2018-08-31 11:10:58 -07:00
Mark Searles f0183dfaac Add keyword 'explicit' to avoid ambiguity
Older gcc, e.g., 5.x, see an ambiguity in some calls. Example error
output as seen with gcc 5.5:

In file included from benchmark_wino.cpp:1:
In file included from ./miopen.hpp:13:
In file included from /usr/lib/gcc/x86_64-linux-gnu/5.5.0/../../../../include/c++/5.5.0/regex:38:
In file included from /usr/lib/gcc/x86_64-linux-gnu/5.5.0/../../../../include/c++/5.5.0/algorithm:62:
In file included from /usr/lib/gcc/x86_64-linux-gnu/5.5.0/../../../../include/c++/5.5.0/bits/stl_algo.h:66:
In file included from /usr/lib/gcc/x86_64-linux-gnu/5.5.0/../../../../include/c++/5.5.0/random:51:
/usr/lib/gcc/x86_64-linux-gnu/5.5.0/../../../../include/c++/5.5.0/bits/random.tcc:1324:27: error: call to 'abs' is ambiguous
                    const double __y = -std::abs(__n) * __param._M_sm - 1;
                                        ^~~~~~~~
/opt/rocm/hip/include/hip/hcc_detail/hip_complex.h:345:31: note: candidate function
__DEFINE_HIP_COMPLEX_REAL_FUN(abs, hipCabs)
2018-08-31 10:36:05 -07:00
Aaron Enye Shi e7352a1f4a __forceinline__ needs inline and always_inline
Related to SWDEV-163084
2018-08-29 18:33:35 +00:00
Yaxun Sam Liu 1f74a66bea Add clock() and clock64() 2018-08-29 10:51:00 -04:00
Aaron Enye Shi ba9fc6f355 Add bitinsert64 to device functions 2018-08-28 20:01:14 +00:00
Aaron Enye Shi b939a3c6ed Clean up and add meaningful variables to bit funcs 2018-08-28 18:19:48 +00:00
Aaron Enye Shi b445e45007 Add bitextract and bitinsert functions 2018-08-20 16:46:12 +00:00
Maneesh Gupta d8d45a906c Merge pull request #623 from ROCm-Developer-Tools/tf18
Fix declaration conflict when hip/math_functions.h is included first
2018-08-10 09:36:27 +05:30
Maneesh Gupta 11cccaa5d0 Merge pull request #613 from ROCm-Developer-Tools/hcc-compat
Add HCC compatibility mode for hip-clang
2018-08-10 09:32:31 +05:30
Yaxun Sam Liu b8a0c14171 Fix __HIP_ARCH_* not defined after including math_functions.h
hcc_detail/math_functions.h used to include hcc_detail/hip_runtime.h.

Removing it has caused regression in TensorFlow 1.8.

Put it back for backward compatibiliity.
2018-08-08 08:55:28 -04:00
Yaxun Sam Liu 450ea49df0 Fix __HIP_DEVICE_COMPILE__ not defined when hip/math_functions.h is included
This fixes build failure in TensorFlow 1.8 for HCC
2018-08-07 17:14:39 -04:00
Yaxun Sam Liu dacb18414e Fix declaration conflict when hip/math_functions.h is included first
This fixes build failure in TensorFlow 1.8 for HCC
2018-08-07 15:44:59 -04:00
Wen-Heng (Jack) Chung 2604f33930 Revert "HIP program state re-initialization logic"
This reverts commit 379b7a2241.
2018-08-03 17:03:04 -05:00
Maneesh Gupta 78ca12d583 Merge pull request #603 from b-sumner/master
Drop use of NVCC_COMPAT since it is always set
2018-08-03 08:45:18 +05:30
Maneesh Gupta d0cdc69676 Merge pull request #610 from aaronenyeshi/enable-hcc-module-launch
Enable hipHccModuleLaunchKernel on HIP-Clang
2018-08-02 08:08:46 +05:30
Yaxun Sam Liu ef1d1d57f7 Add hc_get_workitem_absolute_id for hip-clang HCC compatibility mode 2018-08-01 15:14:38 -04:00
Yaxun Sam Liu 43bb490947 Add HCC compatibility mode 2018-08-01 15:14:38 -04:00
Aaron Enye Shi d1149a9b64 Enable hipHccModuleLaunchKernel on HIP-Clang 2018-07-31 20:38:20 +00:00
Aaron Enye Shi 99e6496898 Protect macro TYPEDEF_VECTOR_VALUE_TYPE and prefix __hip_ for vector_value_types 2018-07-30 20:41:24 +00:00
Aaron Enye Shi e0e785a45d Ensure correct vector value type used in texture
Implement the vector value types in terms of SCALAR_TYPE##SIZE_vector_value_type for float, int, uint and sizes 2, 3, 4, 8, and 16.
2018-07-30 19:43:12 +00:00
Yaxun Sam Liu 03320890de Fix thread index calculation in __hip_malloc 2018-07-27 16:24:51 -04:00
Yaxun Sam Liu 02d0e93601 Support malloc/free for hip-clang 2018-07-27 16:24:51 -04:00
Brian Sumner 71f96388ed Drop use of NVCC_COMPAT since it is always set 2018-07-27 09:29:29 -07:00
Aaron Enye Shi 8c50bcfdab Enable HIP texture functions with HIP vector types
Include texture_functions.h on HIP-Clang path. Use HIP vector types instead of hc::short_vectors for texture_functions, and remove HCC headers.
2018-07-25 19:52:19 +00:00
Maneesh Gupta 06db076b10 Merge pull request #584 from ROCm-Developer-Tools/fix-vector-explicit-cast
Fix explicit cast required
2018-07-23 10:13:05 +05:30
Aaron Enye Shi 4ddb81424f Fix explicit cast required
This is required for PyTorch which runs into an issue with narrowing types. Requires an explicit cast.
2018-07-20 16:34:15 -04:00
Aaron Enye Shi ffefae77d7 [HIPClang] Switch ordering on math functions
Header math_functions.h should be included after including __clang_cuda_math_forward_declares.h to avoid warning: attribute declaration must precede definition.
2018-07-20 13:25:04 -04:00