AI/rocm-systems
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
5,050 Commits 3 Branches 0 Tags
446c8685e2fbc90645f03fdb9c48030969d6cb0c
Commit Graph

4345 Commits

Author SHA1 Message Date
Alex Voicu 9ec697c620 Re-sync with upstream. 2018-10-10 11:43:49 +01:00
Maneesh Gupta 07ee1f07d8 Implementation for stream priority 
- Requires ROCm 1.9.x or higher
- Requires HCC with PR#886 merged

Change-Id: Id7c95ea091ee610e80c9ad815f1cb989cba570ca
 2018-10-05 16:27:46 +05:30
Aaron Enye Shi 03822afaa9 Replace IRIF fences with atomic_work_item_fence 2018-10-04 21:47:28 +00:00
Maneesh Gupta e7d3f3860e Merge pull request #697 from yxsamliu/dev-lib-path 
Let hipcc add --hip-device-lib-path by default for hip-clang
 2018-10-04 07:40:37 +05:30
Aaron Enye Shi 6a71cf17a1 Fix hip_vector_types.h for long long vectors 
There was a missing long in the declaration for [u]longlongN types.
 2018-10-03 13:57:52 -04:00
Evgeny Mankov cb4eb94174 [HIPIFY] CUDA 10.0 Driver API initial support 2018-10-03 20:29:22 +03:00
Rahul Garg a78d68d639 Corrected the width calculation logic to accomodate multi channels 2018-10-03 12:07:38 +05:30
Yaxun Sam Liu 88768895bc Let hipcc add --hip-device-lib-path by default for hip-clang 
hip-clang by default assumes -fno-gpu-rdc, therefore requires
--hip-device-lib-path by default.
 2018-10-01 15:14:54 -04:00
Yaxun Sam Liu 2c361906fa Add HIPCC_COMPILE_FLAGS_APPEND 2018-10-01 14:51:29 -04:00
Rahul Garg 167265c279 Fixed image width for linear resource type texture 2018-10-01 15:28:34 +05:30
Evgeny Mankov 4e7b330037 [HIPIFY][cmake] CUDA 10.0 is not supported. 2018-09-27 19:05:22 +03:00
Wen-Heng (Jack) Chung e257de95f3 Keep the map which tracks GPU kernel symbols to grow monotonically 2018-09-26 19:49:02 +00:00
Wen-Heng (Jack) Chung 060b3c0bf8 Improve performance of re-initialization logic 
Keep track of shared libaries already discovered. Do not build HSA executables
for them.
 2018-09-26 19:48:56 +00:00
Wen-Heng (Jack) Chung 319f007bf1 HIP program state re-initialization logic 
This commit is to support kernels dynamically loaded thru means such as
dlopen() after HIP runtime initializes.
 2018-09-26 19:48:47 +00:00
Evgeny Mankov 6056686796 [HIPIFY][docs] Fix typos in Readme.md 2018-09-26 17:26:25 +03:00
Evgeny Mankov e879118d87 Merge pull request #690 from emankov/hipBLAS 
[HIPIFY][doc] Update README.md due to new LLVM 7.0.0 and CUDA 10.0 re…
 2018-09-26 17:05:47 +03:00
Evgeny Mankov a7e75c72ee [HIPIFY][doc] Update README.md due to new LLVM 7.0.0 and CUDA 10.0 releases. 2018-09-26 17:01:59 +03:00
Rahul Garg bd27310127 Return hipSuccess when sizeBytes=0 in hipMemset 2018-09-26 12:47:36 +05:30
Maneesh Gupta 802520f5f1 Merge pull request #685 from ROCm-Developer-Tools/hip-trig-return 
Improve hip_trig test case
 2018-09-26 09:50:48 +05:30
Aaron Enye Shi eeb6c11050 Use sinf and cosf from ocml device libs 
Using llvm_amdgcn builtin fails to produce accurate values, we should move to using the ocml device library versions.
 2018-09-25 19:31:39 +00:00
foreman 8d912e037a P4 to Git Change 1610499 by skudchad@skudchad-lnx on 2018/09/25 14:43:35 
SWDEV-145570 - [HIP] - Map hipinfo sample from HIP mirror so that we have the same hipinfo

Affected files ...

... //depot/stg/opencl/drivers/opencl/api/hip/hip_peer.cpp#3 edit
... //depot/stg/opencl/drivers/opencl/tools/hipinfo/CMakeLists.txt#2 edit
... //depot/stg/opencl/drivers/opencl/tools/hipinfo/build/Makefile.hipinfo#2 edit
... //depot/stg/opencl/drivers/opencl/tools/hipinfo/hipinfo.cpp#2 delete
 2018-09-25 15:01:11 -04:00
Aaron Enye Shi 04ed44f074 Use trig functions from ocml instead 2018-09-25 15:58:36 +00:00
Evgeny Mankov d12c1f7eff Merge pull request #687 from emankov/hipBLAS 
[HIPIFY][BLAS] Add support of hipblasGemmEx and corresponding types
 2018-09-25 18:48:06 +03:00
Evgeny Mankov 65035fa485 [HIPIFY][BLAS] Add support of hipblasGemmEx and corresponding types 
TODO (hipBLAS/HIP): rename hipblasDatatype_t to hipDataType_t and move it from hipBLAS to HIP, as Data types are used not only in BLAS library.
 2018-09-25 18:46:23 +03:00
Evgeny Mankov f627d9a9a2 [HIPIFY][docs] Update CUDNN_API_supported_by_HIP.md 2018-09-25 16:50:40 +03:00
Aaron Enye Shi bc5de031a5 Fix missing HIP_ASSERT in hip_trig 2018-09-24 15:01:24 +00:00
Maneesh Gupta ad008c5e3f Merge pull request #683 from ROCm-Developer-Tools/hip-float-rounded 
Disable non-default-rounded functions
 2018-09-24 09:32:52 +05:30
Aaron Enye Shi afef6a3ece Improve hip_trig test case 
Hip_trig is failing but returning a passing value on HIP-Clang.  Also update with debug output and free arrays used.
 2018-09-21 15:28:10 -04:00
Maneesh Gupta 5468e01d04 Merge pull request #684 from ROCm-Developer-Tools/mangupta-patch-2 
[ci] Renable nvcc testing
 2018-09-21 11:05:45 +05:30
Maneesh Gupta a4bdbe0aaf Update hipTestFMA.cpp 2018-09-21 11:00:08 +05:30
Maneesh Gupta cc4e8c7e8b [ci] Renable nvcc testing 
Change-Id: I7d720b41a3ddc99453fee8b9be30494bfec3a808
 2018-09-21 10:11:31 +05:30
Aaron Enye Shi a3dea28c9d Disable non-default-rounded functions 
Device library has removed the non-default-rounded functions, so hipFloatMath will fail to build. These include the removal of __ocml_sqrt_rte, __ocml_sqrt_rtn, __ocml_sqrt_rtp, and __ocml_sqrt_rtz. As seen here: https://github.com/RadeonOpenCompute/ROCm-Device-Libs/commit/2fc04e10e1354edee331ce700f98a60f8255effb . Disable these function tests for now, until they are re-enabled, or deleted completely.
 2018-09-20 16:33:32 -04:00
Maneesh Gupta 7cec85e1ae Merge pull request #614 from ROCm-Developer-Tools/fma 
Add overloading resolution functions for fma
 2018-09-20 13:38:03 +05:30
Maneesh Gupta e614aac0da Update hipTestFMA.cpp 
Fix the test so that it works on nvcc path as well.
 2018-09-20 11:23:51 +05:30
Maneesh Gupta ff8bdd0375 Merge pull request #682 from aaronenyeshi/hip-config-files 
Create separate config files for hcc and clang
 2018-09-20 11:21:31 +05:30
Aaron Enye Shi 73a5ea0f9f Create separate config files for hcc and clang 
The state of HIP_COMPILER is not passing into config files, therefore config files cannot use if statements to determine dependency. For HIP-Clang, we should remove find_dependency(hcc), so we create separate config files depending on compiler path. This fixes issue in apps that uses hip-config.cmake.
 2018-09-19 18:29:56 +00:00
Yaxun Sam Liu cfa71293ce Silent warnings about duplicate static keyword 
static is already in __DEVICE__, so should be removed.
 2018-09-19 10:39:45 -04:00
Yaxun Sam Liu a8bed200b7 Add fma function with float and _Float16 arguments 2018-09-19 09:59:33 -04:00
Maneesh Gupta f074207858 Update Jenkinsfile 
[ci] Update list of disable tests for automation
 2018-09-19 15:06:22 +05:30
Maneesh Gupta 50228bf5fc Merge pull request #635 from ROCm-Developer-Tools/fix-include 
Fix build failure of hipTestHalf and hipTestIncludeMath for hip-clang
 2018-09-19 09:04:06 +05:30
Yaxun Sam Liu 781f0d97b5 Fix build failure of hipTestHalf and hipTestIncludeMath for hip-clang 2018-09-18 21:00:15 -04:00
Rahul Garg c878d9cde4 Fixed get alignment offset test for NVCC 
- Removed dtest hipGetAlgntoffset2D.cpp
- hipGetTextureAlignmentOffset is being tested now in hipBindTexRef1DFetch.cpp
 2018-09-18 14:10:03 +05:30
Maneesh Gupta d44f7401f0 Merge pull request #672 from iotamudelta/fp16_fix 
Only LLVM6 and higher contain the necessary intrinsics.
 2018-09-18 08:43:33 +05:30
Maneesh Gupta 77ed6e2c46 Merge pull request #674 from mangupta/fix_dtests_on_nvcc 
[dtests] Fix hipTestClock, hipTestNew, hipTestGlobalVariable, hipSimpleAtomicsTest & hipTestIncludeMath tests on nvcc path
 2018-09-18 07:50:52 +05:30
Maneesh Gupta dd0373dc02 Merge pull request #677 from yxsamliu/fix-launch-decay 
Fix hipLaunchKernelGGL for hip-clang
 2018-09-18 07:50:37 +05:30
Maneesh Gupta ff1163143f Merge pull request #676 from VincentSC/patch-1 
Update install.sh
 2018-09-18 07:50:28 +05:30
Yaxun Sam Liu b23db4e452 Disable device code for gcc in hip_memory.h 
These device code should only be seen by HCC or hip-clang. They causd build failure
for HIP-VDI runtime and should be disabled for gcc.
 2018-09-17 16:50:42 -04:00
Yaxun Sam Liu 0471ce2da6 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
VincentSC fcc9aa3412 Update install.sh 
'rename' added as a Linux prerequisite.
 2018-09-17 13:25:12 +02:00
Maneesh Gupta f63ffaf6e5 [dests] Fix hipTestClock, hipTestNew & hipTestGlobalVariable tests for nvcc 
nvcc does not support global kernels in struct/class

Change-Id: I2d7297e0c3725564215e20dbdd31c0bb8d7a07de
 2018-09-17 15:32:05 +05:30