AI/rocm-systems
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
4,389 Commits 3 Branches 0 Tags
cea5489f00be50d2a8bbbad1cb91f80190607236
Commit Graph

4389 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Alex Voicu cea5489f00 Address post-staging issues in #1809 (#1894) 
Fixes SWDEV-223910 and SWDEV-223663
 2020-02-27 16:21:12 +05:30
Maneesh Gupta 3b4b58bda6 bump version to 3.2 (#1898) 
- Bump version to 3.2
- [ci] Enable tests on ROCm 3.1
 2020-02-27 16:18:31 +05:30
Nick Curtis a094c87038 fix long shuffle implementations for windows (#1895) 
Fixes for SWDEV-223694
 2020-02-26 15:53:56 +05:30
Yaxun (Sam) Liu ff4479fcc7 Fix hipcc for extra -mllvm option (#1885) 2020-02-26 15:53:43 +05:30
Sarbojit2019 d81a9a712a [HIPIFY] Add back missing execute permission to hipify-perl (#1881) 
hipify-perl script lost its executable permission hence "samples/0_Intro/square" was failing. Fixes SWDEV 223433.
 2020-02-19 13:48:20 +05:30
eshcherb 82478dca65 adding hipExtModuleLaunchKernel to tracing layer (#1880) 2020-02-19 13:47:49 +05:30
Alex Voicu 730f23829e Tweak synchronous memcpy implementation (#1809) 
The existing one can have issues on certain systems, therefore this limits use of direct memcpy via largeBAR to sizes where it is unequivocally better.

Also addresses SWDEV-220030 and SWDEV-222237.
 2020-02-18 20:50:27 +05:30
Yaxun (Sam) Liu f4120043ed Let HIP-Clang inline all functions by default (#1875) 
This is a quick workaround to match HCC behavior for performance since inlining usually
results in more optimization opportunities therefore better performance.

We will fine tuning inline threashold later.
 2020-02-17 22:49:26 +05:30
Rahul Garg ec84c16d75 Fix hipMemcpy3D (#1798) 
Fixes #1790 and #1791. hipMemcpy3D still requires further refactoring for different input and output combinations.
 2020-02-17 19:35:35 +05:30
Maneesh Gupta 2ec55fde47 [dtests] Fix random timeout failures in hipModuleLoadDataMultThreaded (#1877) 
Limit the max threads that are launched to 16.
 2020-02-17 11:16:20 +05:30
vsytch 5afe43b1aa Add missing __hip_pinned_shadow__ attributes to the texture global vars. (#1866) 2020-02-15 09:52:25 +05:30
Maneesh Gupta e085417370 Use deque instead of vector for code readers so that the iterators and references will be stable (#1851) 
* Use deque instead of vector for code readers so that the iterators and references will be stable

* Fix compile error

* Assign the iterator

* Add multithreaded test

* Make threads a multiple of hardware concurrency

* Output on failure

* Add setDevice to try and initialize the context on cuda

* Create context for cuda

* Set context on each thread

* Reduce threads on cuda

* Skip test on cuda

* Try to initialize the primary context on cuda

* Push ctx to the stack as current

* Revert "Push ctx to the stack as current"

This reverts commit e8a7cc6957.

* Revert "Try to initialize the primary context on cuda"

This reverts commit 86cdb30195.

* updated test for nvidia path

* Add c++11 option for nvcc

Co-authored-by: satyanveshd <53337087+satyanveshd@users.noreply.github.com>
 2020-02-15 09:51:24 +05:30
Nick Curtis 5b0f34c59b Implement long / long long shuffles (#1829) 
Implement additional data-types for shuffles (long and long long).
Based upon the double implementation.
 2020-02-15 09:51:09 +05:30
Siu Chi Chan 13052f27c2 Disabling HCC code object v3 generation by default. 
Some PyTorch unit tests have regression.  Disabling cov3 to allow more
time to debug and unblock PyTorch

Change-Id: Iba7f425ef3499c20c42ec45d9152b5d27ce97d03
 2020-02-14 19:39:27 -05:00
Evgeny Mankov eae0bddbca Merge pull request #1870 from emankov/HIP 
[HIPIFY][doc] Update README.md: LLVM 10.0.0-rc2 - the latest supported LLVM Release
 2020-02-14 13:25:43 +03:00
Evgeny Mankov 0315aefe78 [HIPIFY][doc] Update README.md: LLVM 10.0.0-rc2 - the latest supported LLVM Release 2020-02-14 13:09:31 +03:00
Rahul Garg 893ccc4ddf [sample] Add hipDispatchEnqueueRateMT (#1869) 
* [sample] Add hipDispatchEnqueueRateMT
 2020-02-13 23:21:40 -08:00
Evgeny Mankov 60c76da392 Merge pull request #1867 from emankov/HIP 
[HIPIFY][doc] Update README.md: Windows tested configurations
 2020-02-13 18:48:38 +03:00
Evgeny Mankov fb52a3973b [HIPIFY][doc] Update README.md: Windows tested configurations 2020-02-13 18:34:10 +03:00
Satyanvesh Dittakavi 2034ed3561 Add c++11 option for nvcc 2020-02-13 19:48:26 +05:30
Satyanvesh Dittakavi 627d9a1f46 updated test for nvidia path 2020-02-13 16:34:05 +05:30
Jeff Daily fe47fce496 missing break statement in hipDeviceGetAttribute (#1865) 
The break is missing for hipDeviceAttributeMaxTexture3DDepth.
 2020-02-13 14:22:56 +05:30
Sarbojit2019 a03628335c [hip] Fix for bug introduced in #1770 when blockSize is non-power of 2 (#1864) 
Fixes SWDEV-222161
 2020-02-13 14:22:46 +05:30
Sarbojit2019 e334d3d6ec ihipEnablePeerAccess return error if peer is not accessible (#1858) 
hipDeviceEnablePeerAccess returns success and adds peer into the list even if it is not accessible which creates problem in hipMalloc when it tries to share the ptr to peer device.
Proposed change is to check the access status before updating the peer list and update only when it can access the peer.
 2020-02-13 14:22:11 +05:30
ansurya 888a7f2a90 Reduce GPU copying based on arch it runs on (#1751) 
Implements SWDEV-213230.
 2020-02-13 14:21:51 +05:30
Evgeny Mankov 5930da7f56 Merge pull request #1830 from asalmanp/coop_flag_define 
[HIP][HIPIFY] Add some missing flags for cooperative launch and occup…
 2020-02-12 14:07:39 +03:00
Paul dd6676dadf Revert "Try to initialize the primary context on cuda" 
This reverts commit 86cdb30195.
 2020-02-11 12:34:11 -06:00
Paul cd279cd574 Revert "Push ctx to the stack as current" 
This reverts commit e8a7cc6957.
 2020-02-11 12:34:10 -06:00
Paul e8a7cc6957 Push ctx to the stack as current 2020-02-11 11:46:29 -06:00
Paul 86cdb30195 Try to initialize the primary context on cuda 2020-02-11 11:26:24 -06:00
Aryan Salmanpour 1c013bab3e fix build error in nvcc path 2020-02-11 12:16:51 -05:00
Jatin Chaudhary 18a64fe8d4 Revert "Sync hip-targets*.cmake in package with install changes (#1831)" (#1860) 
Fixes SWDEV-222155 & SWDEV-222158
This reverts commit 2ec7c2ca33.
 2020-02-11 11:56:57 +05:30
Paul accdd882dc Skip test on cuda 2020-02-10 17:23:58 -06:00
Paul 5d24a2beef Reduce threads on cuda 2020-02-10 16:37:34 -06:00
Paul 7a6e88480b Set context on each thread 2020-02-10 16:01:53 -06:00
Paul bb145e77ae Create context for cuda 2020-02-10 15:52:34 -06:00
Paul 9494d0e3c9 Add setDevice to try and initialize the context on cuda 2020-02-10 13:37:45 -06:00
Aryan Salmanpour 504412c1e4 Fix a typo causing a build error 2020-02-10 11:44:40 -05:00
Aryan Salmanpour dfa565d76f resolve merge conflict 2020-02-10 10:30:55 -05:00
Maneesh Gupta 484d67a485 gedit/hip.lang does not need a seperate license 2020-02-10 16:27:20 +05:30
Maneesh Gupta f22d042bc8 Update copyright section in gedit/hip.lang 2020-02-10 16:25:38 +05:30
Maneesh Gupta d032637934 Revert "Match Occupancy APIs syntax with CUDA (#1625)" (#1857) 
Reverting this for now till we figure out how to avoid the build
breakage.

This reverts commit 262ad13dd5.
 2020-02-10 10:45:28 +05:30
Alex Voicu 8f96152d3a (Maybe) Match alignment between Clang and GCC. (#1789) 
Should fix #1740 and the related internal bug.
 2020-02-10 10:44:49 +05:30
mhbliao 8343d7c9aa [hip] Cleanup compiler wrapper for HIP-Clang. (#1847) 2020-02-07 13:28:26 -08:00
Paul 2405ab236f Output on failure 2020-02-07 10:13:28 -06:00
Paul fb8c7cb3ec Make threads a multiple of hardware concurrency 2020-02-06 16:23:29 -06:00
Paul bf69c7ae32 Add multithreaded test 2020-02-06 16:21:40 -06:00
Michael LIAO 59bd608bfe [hipcc] Skip warning on gfx000. 
- The known target checking should skip `gfx000` as well as it won't be
  used in real compilation command formation. The avoid generating
  annoying warning on `gfx000`.
 2020-02-06 17:09:14 -05:00
Paul 84d6eb985d Assign the iterator 2020-02-06 12:15:29 -06:00
Paul bb26e99c73 Fix compile error 2020-02-06 12:04:50 -06:00