AI/rocm-systems
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
3,499 Commits 3 Branches 0 Tags
ea991e9e5d49921c8cd6a30743dc2a48d4f06801
Commit Graph

3499 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Jenkins ea991e9e5d Merge 'master' into 'amd-master' 
Change-Id: I36a294a054f029038c3ae5a9376695e8790897ff
 2019-03-07 04:09:48 -06:00
eshcherb 88b4621dbd roctracer-proto header find_path (#884) 2019-03-06 17:36:34 +05:30
Maneesh Gupta 352b17346c Merge pull request #949 from gargrahul/single_stream_concurrent_kernels 
Add extension for kernel concurrency on same stream
 2019-03-06 17:34:54 +05:30
Maneesh Gupta 0c4a40efcc Merge pull request #932 from ROCm-Developer-Tools/feature_maybe_dlopen_test 
Add one test case for complex dynamic loading behavior
 2019-03-06 17:32:23 +05:30
Alex Voicu ea0fcf3e61 dlopen() fixes (#929) 
* Initial attempt to switch over to internally linked state.

* Add missing CMake update.

* hipLaunchKernelGGLImpl must be inline as well. Ensure internal linkage.

* Ensure global retrieval uses internally linked state.

* Hide HC in the implementation. Minimise ADL woes.

* Strange software exists, and must be catered to.

* Use a less spammy mechanism for ensuring internal linkage / non-export.

* Remove leftover internal detail.
 2019-03-06 17:31:44 +05:30
Jenkins 086cf48fec Merge 'master' into 'amd-master' 
Change-Id: Ia02ee3c7c354a61f2c3027090ba2ee8d2803e37a
 2019-03-06 04:09:50 -06:00
Rahul Garg 59081c69fc Add extension for kernel concurrency on same stream 2019-03-06 12:55:39 +05:30
Maneesh Gupta 2d67bc5e9c Merge pull request #936 from mangupta/swdev-174923 
[hipconfig] Update HIP_PLATFORM detection logic
 2019-03-06 06:08:11 +05:30
Evgeny Mankov 23de66bc7e Merge pull request #948 from emankov/master 
[HIPIFY] Change CUDA Driver's functions' cuMemsetD32(Async) mapping
 2019-03-05 18:18:39 +03:00
Evgeny Mankov a87d798d90 [HIPIFY] Change CUDA Driver's functions' cuMemsetD32(Async) mapping 
cuMemsetD32(Async) -> hipMemsetD32(Async) (was hipMemset(Async))

based on:
[#933] https://github.com/ROCm-Developer-Tools/HIP/pull/933
 2019-03-05 18:13:18 +03:00
Maneesh Gupta 2ed31e9f6c Merge pull request #933 from ROCm-Developer-Tools/fix_hipmemset 
Add HIP memset APIs to cope with non-zero initial values of integer types
 2019-03-05 14:31:38 +05:30
Maneesh Gupta 51f29b9cee Update hipMemset.cpp 
Address build issues on nvcc path.
 2019-03-05 12:11:11 +05:30
Maneesh Gupta 8db717c769 Update hip_runtime_api.h 
Use hipCUResultTohipError instead of hipCUDAErrorTohipError in hipMemsetD32 & hipMemsetD32Async.
 2019-03-05 12:10:01 +05:30
Wen-Heng (Jack) Chung 5cbd28f29b Address code review comments to use hipDeviceptr_t 2019-03-05 05:51:05 +00:00
Wen-Heng (Jack) Chung 0337b686ef Add implementation for NVCC path 2019-03-04 20:11:12 -08:00
Wen-Heng (Jack) Chung bfde8a7fab Add direct test for hipMemsetD32 and hipMemsetD32Async 2019-03-04 17:20:32 +00:00
Wen-Heng (Jack) Chung 7ebbbd3525 Add hipMemsetD32 and hipMemsetD32Async 
Add 2 extra memset functions which fills memory with integer-typed data

Also change the parameters of ihipMemset to better explain the semantic
 2019-03-04 17:00:33 +00:00
Maneesh Gupta de4a9b8446 Merge pull request #939 from gargrahul/update_hipmemset_test 
[dtest] Update hipMemset test
 2019-03-03 20:29:55 +05:30
Rahul Garg 131c0ea326 Fix review comments 2019-03-02 23:38:37 +05:30
Maneesh Gupta d28c13ed69 Merge pull request #945 from wkwchau/hipMemset3D_fix 
Fix hipMemset3D test
 2019-03-01 21:18:12 +05:30
Maneesh Gupta b051d44d24 Merge pull request #942 from yxsamliu/v3 
revert hipcc changes about code object v3
 2019-03-01 21:17:10 +05:30
Wilkin Chau 8d92d1ebd7 Fix hipMemset3D test 
Calculate the allocated size based on the width, height and depth.
 2019-02-28 22:42:46 +00:00
Rahul Garg 77c6d5451f Fix hipMemset test for HIP/NVCC 2019-03-01 03:46:57 +05:30
Yaxun Sam Liu e8556388f9 Revert "hipcc should consume -mcode-object-v3 flag" 
This reverts commit eb33507774.
 2019-02-28 11:21:47 -05:00
Yaxun Sam Liu 1bd8d24278 Revert "Change code-object flag to only HIP-Clang" 
This reverts commit 1c6365fbe7.
 2019-02-28 11:20:04 -05:00
Yaxun Sam Liu 737972d31c Revert "Consume the code obj args to prevent duplicates" 
This reverts commit 2cd34f3c01.
 2019-02-28 11:19:35 -05:00
Jenkins 9965ca4b55 Merge 'master' into 'amd-master' 
Change-Id: Iba89017b25273c611bee5a15fa619c7ddf5ae0aa
 2019-02-28 04:09:59 -06:00
Maneesh Gupta b93666bdde Merge pull request #938 from gargrahul/fix_hipBusBW_p2p_bidir 
Fix hipBusBW sample for P2P bidirectional test
 2019-02-28 07:14:38 +05:30
Maneesh Gupta 07b0e9574c Merge pull request #937 from yxsamliu/nan2 
Fix nan for windows
 2019-02-28 07:14:27 +05:30
Maneesh Gupta 2da14e97f2 Merge pull request #935 from gargrahul/fix_hipbusbw_beatsoverflow 
Fix hipBusBW overflow with setting beats/iterations
 2019-02-28 07:14:16 +05:30
Maneesh Gupta 2619b11efc Merge pull request #934 from gargrahul/fix_forceinline_non_hcc 
Fix forceinline for non HCC compilation
 2019-02-28 07:14:05 +05:30
Rahul Garg 026dc9457e Update hipMemset test 2019-02-28 06:54:49 +05:30
Rahul Garg 70278f1777 Fix hipBusBW sample for P2P bidirectional test 2019-02-28 00:56:07 +05:30
Yaxun Sam Liu 278007218a Fix nan for windows 2019-02-27 12:33:26 -05:00
Maneesh Gupta 8b25f7b4cf [hipcofig] Update HIP_PLATFORM detection logic 
HIP_PLATFORM detection logic relied on finding a working KFD. If it was
found, the platform was set as hcc else as nvcc.
However this logic is flawed since it is possible for the development
system to only have the user mode bits to build HIP application code.
Hence the better logic is to rely on finding a suitable compiler.
The new logic is as follows:
- look for a working HCC. If found, platform is set as hcc.
- else look for a working NVCC. If found, platform is set as nvcc.
- else the platform defaults to hcc for now.

Change-Id: Ifcc42c29a19f722153d5c23c55f1a8765dceaf6b
 2019-02-27 14:10:21 +05:30
Rahul Garg 673ecd02fe Fix hipBusBW overflow with setting beats/iterations 2019-02-27 00:18:52 +05:30
Rahul Garg 55145feae6 Fix forceinline for non HCC compilation 2019-02-26 07:50:09 +05:30
Wen-Heng (Jack) Chung 8e5bb3c57a Add one test case for complex dynamic loading behavior 
Existing HIT syntax doesn't seem to support the expected build and run steps
for this test.
 2019-02-25 17:03:31 +00:00
Evgeny Mankov e8ca9cbf95 Merge pull request #930 from emankov/master 
[HIPIFY][doc] Update README.md
 2019-02-25 18:29:26 +03:00
Evgeny Mankov 391aa7221b [HIPIFY][doc] Update README.md 
+ Populate Dependencies section with upcoming LLVM versions
+ Add clang bugs for not working configs LLVM+CUDA
+ Update Testing section
 2019-02-25 18:26:25 +03:00
Evgeny Mankov 70dcb13852 Merge pull request #928 from emankov/master 
[HIPIFY][tests] caffe2 test fix
 2019-02-25 17:39:06 +03:00
Evgeny Mankov 3a8ef9c8a2 [HIPIFY][tests] caffe2 test fix 2019-02-25 17:12:32 +03:00
Evgeny Mankov 284bf085ac Merge pull request #927 from emankov/master 
[HIPIFY][Caffe2] Initial Caffe2 support
 2019-02-25 16:41:17 +03:00
Evgeny Mankov 2619f22e5c [HIPIFY][Caffe2] Initial Caffe2 support 2019-02-23 20:46:22 +03:00
Maneesh Gupta 31c4cc78b1 Merge pull request #925 from yxsamliu/h2f 
Add __gnu_h2f_ieee and __gnu_f2h_ieee
 2019-02-22 13:38:15 +05:30
Maneesh Gupta 53f98edc3d Merge pull request #923 from aaronenyeshi/fix-co-v3-arg 
Consume the code obj args to prevent duplicates
 2019-02-22 13:38:08 +05:30
Yaxun Sam Liu c5e813f64c Add __gnu_h2f_ieee and __gnu_f2h_ieee 
The implementation is copied from HCC runtime.

For hcc it has no effect since apps can find them in either hcc runtime or HIP
runtime.

hip-clang needs it in HIP/HCC runtime so that HIP/HCC and HIP/VDI runtime are
swappable.
 2019-02-21 12:48:28 -05:00
Aaron Enye Shi 2cd34f3c01 Consume the code obj args to prevent duplicates 2019-02-19 17:02:27 +00:00
Jenkins b6c261f14b Merge 'master' into 'amd-master' 
Change-Id: I13b597d44e0f6261d6a35b9541da0bc3e81f581c
 2019-02-18 10:34:42 -06:00
Evgeny Mankov 6eb8256883 Merge pull request #919 from emankov/master 
[HIPIFY][DNN] Revise and sync with CUDNN 10 and hipDNN
 2019-02-15 15:06:28 +03:00