6425 Коммитов

Автор SHA1 Сообщение Дата
Maneesh Gupta 182f8ff28f hipMemcpy returns success if sizeBytes is 0.
Fixes SWDEV-153754 & SWDEV-154178.

[ROCm/hip commit: 0180a82963]
2018-05-21 15:38:44 +05:30
Maneesh Gupta 1800081f27 Merge pull request #460 from mangupta/fix_nvcc_tests
Disable incomplete unit tests that don't work on nvcc path

[ROCm/hip commit: 8ea218cd1b]
2018-05-21 12:05:12 +05:30
Maneesh Gupta bb8918faf7 Disable incomplete unit tests that don't work on nvcc path
Change-Id: If5823ec96a3b2497a08c46ab802c5a0158271053


[ROCm/hip commit: 305592d622]
2018-05-21 11:35:03 +05:30
Maneesh Gupta 9cb5c4efd8 Add hipMalloc3D to nvcc detail
Change-Id: I8a5654066ed1504e3b05eddbbdebf05fd52aa149


[ROCm/hip commit: 661561eead]
2018-05-21 11:33:09 +05:30
Maneesh Gupta c72126e989 Merge pull request #456 from founta/master
defined hipPitchedPtr when using hipcc with nvcc

[ROCm/hip commit: 883241495d]
2018-05-21 10:53:39 +05:30
Maneesh Gupta d56eb91596 Merge pull request #455 from ROCm-Developer-Tools/magic
Change HIP fat binary magic number

[ROCm/hip commit: cac3f1c7cd]
2018-05-21 09:52:03 +05:30
Maneesh Gupta ecb874ef5a Merge pull request #454 from ROCm-Developer-Tools/hip-clang-hipcc
Let hipcc suport hip-clang

[ROCm/hip commit: 343083b807]
2018-05-21 09:51:42 +05:30
foreman 8634501c2b P4 to Git Change 1556942 by yaxunl@yaxunl-lc8 on 2018/05/18 14:25:12
SWDEV-145570 - [HIP] Change fat binary magic number and clang-offload-bundler target name to match clang

Affected files ...

... //depot/stg/opencl/drivers/opencl/api/hip/hip_platform.cpp#11 edit


[ROCm/hip commit: 303df5dd2e]
2018-05-18 14:34:14 -04:00
Alex Voicu 69a32877d7 Update hip_module.cpp
Typo.

[ROCm/hip commit: cd6c979c27]
2018-05-18 17:50:45 +01:00
Rahul Garg 14030c3f17 Fix for memcpy2DAsync for pinned host memory case
[ROCm/hip commit: afe62e7030]
2018-05-18 21:09:50 +05:30
founta 22c2164216 defined hipPitchedPtr
Added a define for hipPitchedPtr to resolve a compiler error

[ROCm/hip commit: 1a108ef7f3]
2018-05-18 09:11:50 -04:00
Maneesh Gupta 3d1d7ccf30 Merge pull request #433 from gargrahul/add_hipmemset3d
Added hipMemset3D

[ROCm/hip commit: 03ac8e6a92]
2018-05-18 14:54:15 +05:30
Maneesh Gupta b53c717615 Merge pull request #440 from yxsamliu/assert2
Add __assert_fail, __device_trap and hipErrorAssert for clang

[ROCm/hip commit: dcb4477f9f]
2018-05-18 14:13:27 +05:30
Maneesh Gupta 5fc95e7a30 Merge pull request #448 from 949f45ac/master
Provide correct __mul64hi and __umul64hi builtins, using code from ROCm-Device-Libs

[ROCm/hip commit: ac7713fa34]
2018-05-18 13:18:16 +05:30
Maneesh Gupta 3da5fac84b Merge pull request #444 from aaronenyeshi/vg20-initial
initial gfx906 support

[ROCm/hip commit: 67d45164fa]
2018-05-18 13:18:07 +05:30
Maneesh Gupta 183fce6294 Merge pull request #437 from scchan/mbcnt_intrinsics
add intrinsics mbcnt_lo, mbcnt_hi, lane_id

[ROCm/hip commit: 223bf5094c]
2018-05-18 13:17:57 +05:30
Yaxun (Sam) Liu 0a7d196cb5 Change HIP fat binary magic number
[ROCm/hip commit: d079463887]
2018-05-17 17:04:51 -04:00
Yaxun (Sam) Liu 67a2d9739e Let hipcc suport hip-clang
[ROCm/hip commit: f4d79a1615]
2018-05-17 14:40:15 -04:00
949f45ac c9db90b077 Reinstate accidentally deleted uchar2Holder
[ROCm/hip commit: 8303bfdffd]
2018-05-17 10:55:45 +02:00
Maneesh Gupta 0ddf49e931 Merge pull request #451 from gargrahul/fix_memcpy2d_for_1d_case
Fixed hipMemcpy2D to handle 1D memcpy case

[ROCm/hip commit: 0e44ca7d0f]
2018-05-17 07:42:47 +05:30
foreman 965ac24258 P4 to Git Change 1555866 by cpaquot@cpaquot-ocl-lc-lnx on 2018/05/16 16:27:00
SWDEV-145570 - [HIP] Store HIP mem flags inside amd::Buffer's flags
	Use the 16 upper bits of amd::Buffer's flags field instead of adding a new field.

Affected files ...

... //depot/stg/opencl/drivers/opencl/api/hip/hip_memory.cpp#26 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocdevice.cpp#86 edit


[ROCm/hip commit: b1ab722a25]
2018-05-16 16:35:53 -04:00
Rahul Garg a45f17fe46 Fix hipCommander Makefile
[ROCm/hip commit: dc4d305c25]
2018-05-16 15:01:32 +05:30
Rahul Garg 4c44cd4a88 Fixed hipMemcpy2D to handle 1D memcpy case
[ROCm/hip commit: 8f010ac68e]
2018-05-16 11:07:10 +05:30
foreman 958fe1930f P4 to Git Change 1555197 by cpaquot@cpaquot-ocl-lc-lnx on 2018/05/15 16:26:41
SWDEV-145570 - [HIP] Fixed a typo, hipStreamGetFlags test passes now

Affected files ...

... //depot/stg/opencl/drivers/opencl/api/hip/hip_stream.cpp#10 edit


[ROCm/hip commit: bb008fbf94]
2018-05-15 16:36:25 -04:00
foreman 134b6c2e26 P4 to Git Change 1555193 by cpaquot@cpaquot-ocl-lc-lnx on 2018/05/15 16:19:50
SWDEV-145570 - [HIP] Implemented events

Affected files ...

... //depot/stg/opencl/drivers/opencl/api/hip/hip_event.cpp#4 edit
... //depot/stg/opencl/drivers/opencl/api/hip/hip_event.hpp#2 edit
... //depot/stg/opencl/drivers/opencl/api/hip/hip_memory.cpp#25 edit
... //depot/stg/opencl/drivers/opencl/api/hip/hip_module.cpp#9 edit
... //depot/stg/opencl/drivers/opencl/api/hip/hip_stream.cpp#9 edit


[ROCm/hip commit: da00b9270a]
2018-05-15 16:26:16 -04:00
Alex Voicu 554aaee804 Update hip_module.cpp
[ROCm/hip commit: 5325b6535e]
2018-05-14 17:15:36 +01:00
Evgeny Mankov 3d792501c6 [HIPIFY][DNN] support of cuDNN 7.1.3 - continuation 2
- not finished yet.
- based on https://github.com/ROCmSoftwarePlatform/hipDNN.
- testing on https://github.com/baidu-research/DeepBench - almost pass, except cusparse (not supported yet).
- started testing of examples from libcudnn7-dev_7.1.3.16-1+cuda8.0_amd64 package.


[ROCm/hip commit: b0fd0c310d]
2018-05-14 16:23:59 +03:00
949f45ac 40681fef87 Provide correct __mul64hi and __umul64hi builtins, using code from ROCm-Device-Libs
[ROCm/hip commit: 79480d7cbd]
2018-05-14 08:34:56 +02:00
Alex Voicu 8712744ba0 Switch over to using native vector types, for better codegen. Remove noise.
[ROCm/hip commit: b8c98d2357]
2018-05-12 00:14:17 +01:00
foreman 72798b9f7b P4 to Git Change 1553709 by cpaquot@cpaquot-ocl-lc-lnx on 2018/05/11 11:24:30
SWDEV-145570 - [HIP] Make streamSet global and protect it
	By default from the spec, streamSet should be global and not per thread.
	There is a flag to make it per thread but we don't handle this yet. We
	would just add another variable that will be thread local and use it instead.

Affected files ...

... //depot/stg/opencl/drivers/opencl/api/hip/hip_context.cpp#12 edit
... //depot/stg/opencl/drivers/opencl/api/hip/hip_device_runtime.cpp#10 edit
... //depot/stg/opencl/drivers/opencl/api/hip/hip_internal.hpp#11 edit
... //depot/stg/opencl/drivers/opencl/api/hip/hip_stream.cpp#8 edit


[ROCm/hip commit: fbd86acc9f]
2018-05-11 11:34:01 -04:00
Aaron Enye Shi 9a591d6c26 Fix hipMathFunction for gfx906
[ROCm/hip commit: 848a24b524]
2018-05-11 10:53:07 -04:00
Alex Voicu 15e0bb5d15 Don't use magic constants, they're evil.
Also clarify that the register count cannot be queried at the moment.

[ROCm/hip commit: 1ba8a35dba]
2018-05-11 11:31:46 +01:00
Alex Voicu d5f965c44a Add support for the hipFuncGetAttributes interface.
[ROCm/hip commit: 13274ce559]
2018-05-11 03:35:10 +01:00
Siu Chi Chan c05e2e2261 initial gfx906 support
[ROCm/hip commit: b898049412]
2018-05-10 19:28:00 +00:00
Evgeny Mankov 3ff6d4c311 [HIPIFY][DNN] support of cuDNN 7.1.3 - continuation
- not finished yet.
- based on https://github.com/ROCmSoftwarePlatform/hipDNN.
- testing on https://github.com/baidu-research/DeepBench.


[ROCm/hip commit: dffe1802be]
2018-05-10 17:36:51 +03:00
foreman 24ffe8a333 P4 to Git Change 1552807 by skudchad@skudchad_test2_win_opencl on 2018/05/09 18:08:07
SWDEV-145570 - [HIP] - Release a stream first before taking it off from the set.
	- Queue::create() needs to be called before returning a valid queue.

	ReviewBoardURL = http://ocltc.amd.com/reviews/r/14830/diff/

Affected files ...

... //depot/stg/opencl/drivers/opencl/api/hip/hip_stream.cpp#7 edit


[ROCm/hip commit: ac011af36b]
2018-05-09 18:15:41 -04:00
foreman 3b885e17b6 P4 to Git Change 1552209 by cpaquot@cpaquot-ocl-lc-lnx on 2018/05/08 20:24:32
SWDEV-145570 - [HIP] Use the as_amd()->asHostQueue with streamSet

Affected files ...

... //depot/stg/opencl/drivers/opencl/api/hip/hip_stream.cpp#6 edit


[ROCm/hip commit: 50810d28b8]
2018-05-08 20:30:17 -04:00
foreman 925b3fbd17 P4 to Git Change 1552198 by skudchad@skudchad_rocm on 2018/05/08 18:57:32
SWDEV-145570 - [HIP] - Synchronize the legacy default stream with other blocking streams as per the spec.

	ReviewBoardURL = http://ocltc.amd.com/reviews/r/14796/diff/

Affected files ...

... //depot/stg/opencl/drivers/opencl/api/hip/hip_context.cpp#11 edit
... //depot/stg/opencl/drivers/opencl/api/hip/hip_internal.hpp#10 edit
... //depot/stg/opencl/drivers/opencl/api/hip/hip_stream.cpp#5 edit


[ROCm/hip commit: 7176a5fe81]
2018-05-08 19:04:35 -04:00
Yaxun (Sam) Liu bec6249f99 Fix warning about inlined function is not defined
[ROCm/hip commit: 19f3ed6f62]
2018-05-08 16:38:50 -04:00
foreman b4719ee8ae P4 to Git Change 1552011 by skudchad@skudchad_test2_win_opencl on 2018/05/08 14:48:45
SWDEV-145570 - [HIP] Fix offset calculation when getting a memory object. Also include case when destination VA may just be a CPU host VA and not nessarily device alloced.
	- Fix hipMemset* to write each byte and now a dword as per the spec

	ReviewBoardURL = http://ocltc.amd.com/reviews/r/14787/diff/

Affected files ...

... //depot/stg/opencl/drivers/opencl/api/hip/hip_memory.cpp#24 edit


[ROCm/hip commit: 2df122be02]
2018-05-08 15:43:13 -04:00
Yaxun (Sam) Liu cbbe58dcd5 Add __assert_fail, __device_trap and hipErrorAssert for clang
[ROCm/hip commit: 7672b44c79]
2018-05-08 15:42:27 -04:00
Siu Chi Chan 923d831d5e add intrinsics mbcnt_lo, mbcnt_hi, lane_id
[ROCm/hip commit: b285145966]
2018-05-08 13:43:53 -04:00
Evgeny Mankov 44007e29ab [HIPIFY][Blas] Sync with CUDA 9.1
[ROCm/hip commit: e5ba9668fc]
2018-05-08 20:42:30 +03:00
Evgeny Mankov ef30e4601e [HIPIFY] Sync with hipBLAS
[ROCm/hip commit: 7681775662]
2018-05-08 19:20:47 +03:00
Rahul Garg e2a2b5bdcf Added hipMemset3D
[ROCm/hip commit: da302c3e93]
2018-05-07 10:24:30 +05:30
Tom Maenan Read Cutting adfca5d8cf Add space between ### and Notes in hip_terms
Makes `Notes` an H3 heading.

[ROCm/hip commit: 9d76f5839e]
2018-05-05 13:30:11 +01:00
foreman 3cb2630ba2 P4 to Git Change 1550648 by skudchad@skudchad_test2_win_opencl on 2018/05/04 20:56:00
SWDEV-145570 - [HIP]
	- Implement hipMemcpyHtoD/DtoH/DtoD/ and their Async APIs
	- Combine logic for hipMemset/Memcpy/Memset2D/Memcpy2D that can be shared across multiple APIs

	ReviewBoardURL = http://ocltc.amd.com/reviews/r/14782/diff/

Affected files ...

... //depot/stg/opencl/drivers/opencl/api/hip/hip_memory.cpp#23 edit


[ROCm/hip commit: 2c5606ea44]
2018-05-05 00:34:05 -04:00
emankov 2ad9863d31 [HIPIFY][test] Undo commit "Apply .clangformat to all repo source files"
Commit broke tests due to code and comments formatting changes, thus FileCheck fails on checks, which are in comments.


[ROCm/hip commit: 21b79cd467]
2018-05-04 22:23:16 +03:00
foreman d6118d44ce P4 to Git Change 1550526 by cpaquot@cpaquot-ocl-lc-lnx on 2018/05/04 15:00:44
SWDEV-145570 - [HIP] Missing HIP_INIT_API macro in hipDeviceSynchronize.

Affected files ...

... //depot/stg/opencl/drivers/opencl/api/hip/hip_device_runtime.cpp#9 edit


[ROCm/hip commit: 65b557ecce]
2018-05-04 15:12:48 -04:00
foreman 2d0867f228 P4 to Git Change 1550486 by cpaquot@cpaquot-ocl-lc-lnx on 2018/05/04 14:00:33
SWDEV-145570 - [HIP] Fix multithread init
	Make the g_ihipInitialized variable per thread
	And make sure to assign a default g_context

Affected files ...

... //depot/stg/opencl/drivers/opencl/api/hip/hip_context.cpp#10 edit
... //depot/stg/opencl/drivers/opencl/api/hip/hip_internal.hpp#9 edit


[ROCm/hip commit: 9129aa17f6]
2018-05-04 14:13:13 -04:00