36b23663e0
replace array designator C99 ( #1694 )
...
* replace array designator C99
* Update texture_functions.h
Highlight valid and invalid values in texFormatToSize
Co-authored-by: Maneesh Gupta <maneesh.gupta@amd.com >
[ROCm/clr commit: 32625638da
2019-12-23 19:10:24 +05:30
47725baf8c
static inline in a header, just like excess sugar in a diet, causes bloat (#1692 )
...
[ROCm/clr commit: cf526f5484
2019-12-23 19:09:38 +05:30
4bb4b5e291
Merge pull request #1734 from vsytch/win10-hipMemcpy-fixes
...
Fix OOM issues with hipMemcpy test on Windows
[ROCm/clr commit: 3c6203c814
2019-12-21 22:02:34 +03:00
35e4141eab
Update texture_functions.h
...
Highlight valid and invalid values in texFormatToSize
[ROCm/clr commit: 6c7ce07577
2019-12-21 12:25:36 +05:30
57b69adc8b
bump version to 3.1 ( #1756 )
...
* Bump version to 3.1
* [ci] Enable tests on ROCm 3.0
[ROCm/clr commit: 86e590ad62
2019-12-20 12:10:25 -08:00
b41f6ad42f
Merge pull request #1757 from emankov/master
...
[HIPIFY][doc] Update Readme.md: versioning
[ROCm/clr commit: 4a8b87a96c
2019-12-20 20:57:48 +03:00
919ef62fde
[HIPIFY][doc] Update Readme.md: versioning
...
[ROCm/clr commit: 65dd13c677
2019-12-20 20:55:19 +03:00
2916c02f7b
Fix for windows dtest build failure ( #1742 )
...
[ROCm/clr commit: 911d406bf1
2019-12-19 13:10:43 -08:00
313ab97c35
[hip] Add macro guarding the enum conversion for scalar accessor. ( #1748 )
...
- That's a high overhead part, which needs enabling ONLY if necessary.
[ROCm/clr commit: a0a50ac601
2019-12-19 10:08:37 -08:00
0bd42da319
Merge pull request #1739 from emankov/hipify-clang
...
[HIP] Unify hipError_t (Step 1)
[ROCm/clr commit: 08f1c8edf3
2019-12-19 07:57:02 +03:00
ad045983bd
Merge pull request #1743 from emankov/master
...
[HIPIFY] Revise HIP runtime
[ROCm/clr commit: b1aa111446
2019-12-16 20:11:19 +03:00
abcc1db917
[HIPIFY] Revise HIP runtime
...
+ hipMemAllocHost is deprecated, use hipHostMalloc instead
+ hipMemAllocManaged -> hipMallocManaged
+ hipFreeHost -> hipHostFree
+ cudaMemset3D(Async) -> hipMemset3D(Async)
+ update docs and hipify-perl accordingly
[ROCm/clr commit: 1108ef06e1
2019-12-16 20:09:55 +03:00
dc12f8017e
Add comments explaining choice on the cap for system memory on Windows.
...
[ROCm/clr commit: a91bafca40
2019-12-13 21:38:27 -05:00
014f068e2f
Add explicit cast when computing the available amount of system memory.
...
[ROCm/clr commit: 4c21eed68b
2019-12-13 21:37:30 -05:00
6c0241f539
[HIP] Unify hipError_t (Step 1)
...
Step 1. Set the same values for RT error codes as for analogous Driver's
[Reason] RT's and Driver's error codes unification in CUDA 10.2
[ROCm/clr commit: 21b0f64f3a
2019-12-13 19:40:16 +03:00
21ae8c53fe
Merge pull request #1738 from emankov/hipify-clang
...
[HIPIFY][build] Fix building with LLVM trunk
[ROCm/clr commit: 207d7d7475
2019-12-13 15:43:24 +03:00
6194dd0d3f
[HIPIFY][build] Fix building with LLVM trunk
...
[ROCm/clr commit: 1377db639f
2019-12-13 15:42:27 +03:00
7531ad39f9
Merge pull request #1736 from emankov/hipify
...
[HIPIFY][BLAS] Sync with cuBlas/rocblas/hipBlas
[ROCm/clr commit: dadbdad99d
2019-12-12 19:56:08 +03:00
7480c5aace
[HIPIFY][BLAS] Sync with cuBlas/rocblas/hipBlas
...
[ROCm/clr commit: c7bea1f865
2019-12-12 19:52:18 +03:00
d2be30f0e0
Reduce the amount of free host memory to 40% of what is reported on Windows, otherwise we can run into OOM situations.
...
[ROCm/clr commit: 282367ed6d
2019-12-11 20:21:12 -05:00
b967f1210d
Only add hipExtLaunchMultiKernelMultiDevice for non-HCC compilers. ( #1729 )
...
[ROCm/clr commit: 539370d308
2019-12-10 10:32:25 -08:00
2984be48d2
fix linking of vector types with gcc ( #1690 )
...
* fix linking of vector types when linking hipcc objects with gcc
* use __atribute__((vector_size)) with both clang and gcc
and reinstate nonaligned n=3 vector type
* use implicit conversion to value and ext_vector_type when available
* Alternate formulation for GCC compatibility
* Built-in arrays don't mix well with placement new
* Fix typo
* Add conversions to enum
* Fix Scalar_accessor assignment.
* Update hip_vector_types.h
* stir up the underlying_type hideous mess
This fixes the HIP build issue "error: only enumeration types have underlying types".
[ROCm/clr commit: 280a8ebadf
2019-12-10 09:40:15 +05:30
b71a6726f4
Merge pull request #1725 from emankov/hipify-clang
...
[HIPIFY] Add cleanupHipifyOptions functionality
[ROCm/clr commit: 8223a48683
2019-12-08 15:15:57 +03:00
6400f2f967
[HIPIFY] Add cleanupHipifyOptions functionality
...
Needed for sorting input files to avoid clang errors on unsupported options
[ROCm/clr commit: 486ec185e7
2019-12-08 15:13:28 +03:00
d851e115cd
Merge pull request #1723 from emankov/hipify-clang
...
[HIPIFY] main source file refactoring
[ROCm/clr commit: 62df9e5732
2019-12-07 14:07:23 +03:00
e374b3d356
[HIPIFY] main source file refactoring
...
[ROCm/clr commit: 72b1d16115
2019-12-07 13:50:13 +03:00
ef2c67b888
Merge pull request #1722 from emankov/hipify-clang
...
[HIPIFY][fix][#1246 ][#1655 ] Sort input files based on their dependency graph
[ROCm/clr commit: 97537d0818
2019-12-07 11:23:01 +03:00
15d0c4b58b
[HIPIFY][fix][ #1246 ][ #1655 ] Sort input files based on their dependency graph
...
+ Implemented by using clang Driver infrastructure
[ToDo] Add tests for in-place hipification of multiple files
[ROCm/clr commit: ef405daaee
2019-12-07 11:20:17 +03:00
6643c80e8e
Fix hipExtLaunchMultiKernelMultiDevice refactoring. ( #1714 )
...
- Use the correct condition for HIP VDI runtime.
[ROCm/clr commit: 209f31525f
2019-12-06 09:49:17 -08:00
48faf7423a
Revert - Changes related to hipMemcpyWithStream ( #1718 )
...
Reverting #1673 , #1697 and #1707 .
Support for hipMemcpyWithStream and memcpy optimizations, will be brought in again once issues seen with these are resolved independently.
[ROCm/clr commit: a369bd4418
2019-12-06 09:51:53 +05:30
d56471b77b
Merge pull request #1712 from emankov/hipify
...
[HIPIFY] CUDA 10.2 (functions only)
[ROCm/clr commit: c3076a2a07
2019-12-04 18:39:10 +03:00
8477ab5d86
[HIPIFY] CUDA 10.2 (functions only)
...
+ Both Driver API and RT API are supported and synced with each other
+ Update *.md docs and hipify-perl accordingly
+ Add new conversion type "virtual_memory", introduced in Dirver API
[ROCm/clr commit: 43441eb99c
2019-12-04 18:37:31 +03:00
03250ea190
Merge pull request #1710 from emankov/hipify
...
[HIPIFY] CUDA 10.2 fp_16 support
[ROCm/clr commit: 187f28e26b
2019-12-04 13:28:43 +03:00
f6da983c96
[HIPIFY] CUDA 10.2 fp_16 support
...
[ROCm/clr commit: 721a8540c0
2019-12-04 13:27:46 +03:00
ecbc77d921
[packaging] Rename hip packages to conform to debian package renaming rules ( #1708 )
...
[ROCm/clr commit: 59b3bd0048
2019-12-04 15:55:37 +05:30
9976235fb4
Merge pull request #1709 from emankov/hipify
...
[HIPIFY] CUDA 10.2 support (data types only)
[ROCm/clr commit: 89703e4ea9
2019-12-04 12:26:51 +03:00
deaf478ced
[HIPIFY] CUDA 10.2 support (data types only)
...
+ Both Driver API and RT API are supported and synced with each other
+ Update *.md docs and hipify-perl accordingly
[ROCm/clr commit: 33f81effb1
2019-12-04 12:25:20 +03:00
6c7a338471
Fix hipMemcpyWithStream ( #1707 )
...
* Fix hipMemcpyWithStream
* Add a dtest for the same.
[ROCm/clr commit: 261a04580f
2019-12-04 11:56:23 +05:30
2471c1950c
[hip] refactoring hipExtLaunchMultiKernelMultiDevice API ( #1698 )
...
[Background] it was found that if lazy linking used for a library that calls hipExtLaunchMultiKernelMultiDevice API then this API can get the wrong program_state object for looking up device kernels leading to a "No device code available" error in this API.
To fix this issue, the API was refactored to be inline and get and pass the correct program_state to an internal hip API to request a multi-device kernel launch.
[ROCm/clr commit: 8eaea4d114
2019-12-04 11:50:51 +05:30
9e9d49c891
Revert "[packaging] Conform to debian package naming rules ( #1647 )" ( #1688 )
...
This reverts commit 71be7baece38e971b645
2019-12-01 14:47:43 +05:30
2c2a726726
Merge pull request #1702 from emankov/hipify
...
[HIPIFY] Sync cudaError/CUresult between CUDA Driver API/CUDA RT API/HIP
[ROCm/clr commit: da468e181f
2019-11-29 19:31:28 +03:00
5317d618d1
[HIPIFY] Sync cudaError/CUresult between CUDA Driver API/CUDA RT API/HIP
...
+ Update *.md docs and hipify-perl accordingly
[Reason]
Starting with CUDA 10.1 all error codes are merged between Driver and RT APIs
[ToDo]
Do the same merge in HIP API as there is no need in distinguishing return codes by API
[ROCm/clr commit: 636108aaa9
2019-11-29 19:29:11 +03:00
97850ae9a8
Revert changes for atomic FADD support when address is in LDS ( #1701 )
...
This reverts PR #1591 and follow-on PR #1695
[ROCm/clr commit: 4c92bd50c4
2019-11-29 11:58:12 +05:30
b26aa2378e
Uniform is_shared query. ( #1695 )
...
[ROCm/clr commit: 17a4780dc6
2019-11-28 13:39:05 +05:30
22c07e2331
fix hipMempcy precondition, test for zero size first ( #1697 )
...
[ROCm/clr commit: 6df73e1f12
2019-11-25 19:03:03 -08:00
aec13e0fb3
replace array designator C99
...
[ROCm/clr commit: bc79c5636b
2019-11-25 16:51:49 +05:30
82b55e6502
Unary operators were too restrictive in the type of their argument. ( #1683 )
...
[ROCm/clr commit: 306d50291e
2019-11-22 07:54:53 +05:30
84953da127
[HIPIFY] fix typo ( #1681 )
...
[ROCm/clr commit: 562336e733
2019-11-22 07:54:37 +05:30
0ece8d0f9b
Fix rocBLAS compilation failure ( #1677 )
...
SWDEV-212749:
o Recent changes to “add support for extended launch” require hip_runtime.h to be include in hip_ext.h
o Order in which external applications include hip_hcc.h/hip_runtime.h causes compilation failure
[ROCm/clr commit: e5fc5aa41c
2019-11-22 07:54:17 +05:30
e1dca187c9
Use native support for atomic FADD when address is in LDS ( #1591 )
...
[ROCm/clr commit: 2ed3a0873c
2019-11-22 07:53:48 +05:30
Maneesh Gupta
Alex Voicu
Evgeny Mankov
Sarbojit2019
mhbliao
Vladislav Sytchenko
jglaser
Rahul Garg
Aryan Salmanpour
Jeff Daily
Anusha Godavarthy Surya
Shyam Saladi
ansurya