Граф коммитов

4341 Коммитов

Автор SHA1 Сообщение Дата
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
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
Paul 1354a447e1 Use deque instead of vector for code readers so that the iterators and references will be stable 2020-02-06 11:56:17 -06:00
Siu Chi Chan 713702eebc Enable code object v3 codgen when compiling with HCC (#1836)
Change-Id: I590c05a418cc37ec247b89124cef17a7fe8b27e2
2020-02-05 20:56:33 +05:30
vsytch fa363ae721 Device texture functions should not normalize the sampled pixel (#1826)
* Device texture functions should not normalize the sampled pixel. This is already done by HW.
* Add support to use h/w capability for normalized float data convertion for driver API's

Co-authored-by: ansurya <50609411+ansurya@users.noreply.github.com>
2020-02-05 20:56:17 +05:30
saleelk 854d3103dd Implement __hipPushCallConfiguration/__hipPopCallConfiguration for hip_clang (#1845)
This is needed so that the right symbols are present if we want to use hip-clang with hip/hcc runtime
2020-02-04 19:37:57 +05:30
lthakur 2ef4dead24 [dtest] Update hipLaunchCoopMultiKernel.cpp (#1844)
used reinterpret_cast<uint32_t*> for numBlocks, as expected by hipOccupancyMaxActiveBlocksPerMultiprocessor() api.
2020-02-04 19:37:38 +05:30
Siu Chi Chan 14e235378f Fix C-style hipLaunchKernel (#1835)
* Fix bug in LaunchKernel test
Instead of passing the address of the gpu buffer, pass the address
of the pointer that holds the address of the gpu buffer

* Fix hipLaunchKernel's kernarg buffer construction.
The hipLaunchKernel implementation should rely on ihipModuleLaunchKernel
to construct the kernarg buffer correctly based on kernel metadata.

* Fix a bug in get_functions where the Kernel_descriptor wasn't constructed with the correct kernarg layout information.

* Fix a bug in kernarg layout parsing dealing with kernel without any arg

* teach ihipModuleLaunchKernel to handle kernel without any arg

* Add a more interesting test
2020-02-04 19:37:16 +05:30
Sarbojit2019 6e62ea5ee3 Added overflow check in kernel launch (#1770) 2020-02-04 09:02:16 +05:30
Daniil Fukalov 8d59305879 Update INSTALL.md (#1834) 2020-02-04 09:01:38 +05:30
Jatin Chaudhary 2ec7c2ca33 Sync hip-targets*.cmake in package with install changes (#1831)
Fixes SWDEV-204247
2020-02-04 08:59:28 +05:30
Rahul Garg 27fbfd9d66 [docs] Fix links in cookbook samples (#1824) 2020-02-04 08:56:31 +05:30
kpyzhov 951d7d4ba6 hipcc script: Refined device selection; added gfx1011 target. (#1821) 2020-02-04 08:56:16 +05:30
ansurya 696c85aec9 [hit] Syntax support for EXCLUDE_HIP_RUNTIME and EXCLUDE_HIP_COMPILER (#1800)
* Added support to disable test based on HIP Runtime(HCC/VDI) and Compiler(hcc/clang)
* Updated README.md, HIT.cmake for new options EXCLUDE_HIP_RUNTIME and EXCLUDE_HIP_COMPILER
2020-02-04 08:56:02 +05:30
gandryey 3c718ebe15 [dtest] RCCL multi device launch test (#1731)
Simple test for hipLaunchCooperativeKernelMultiDevice API.
2020-02-04 08:53:43 +05:30
Maneesh Gupta 00494e8c08 Update Jenkinsfile
Switch using to newer --gpus all instead of older(deprecated) --runtime nvidia
2020-02-03 15:29:35 +05:30
Evgeny Mankov 1aaeb5d485 Merge pull request #1841 from emankov/hipify-clang
[HIPIFY][doc] Update README.md
2020-02-02 01:58:34 +03:00
Evgeny Mankov 5062082157 [HIPIFY][doc] Update README.md
+ Links, versioning, formatting, cleanup
2020-02-02 01:46:58 +03:00
Evgeny Mankov 4f35b8905d Merge pull request #1840 from emankov/hipify-clang
[HIPIFY][doc] CUDA 10.2 is now supported by LLVM 10.0.0-rc1
2020-02-01 18:49:33 +03:00
Evgeny Mankov 244689a2f3 [HIPIFY][doc] CUDA 10.2 is now supported by LLVM 10.0.0-rc1 2020-02-01 18:47:46 +03:00
Evgeny Mankov 2b996fb9fb Merge pull request #1839 from emankov/hipify-clang
[HIPIFY][cmake] CUDA 10.2 support starting from LLVM 10.0
2020-02-01 16:32:24 +03:00
Evgeny Mankov 6cc3554501 [HIPIFY][cmake] CUDA 10.2 support starting from LLVM 10.0
+ hipify-clang unit tests are passed against CUDA 10.2
2020-02-01 16:30:05 +03:00
Evgeny Mankov c6f8166e1e Merge pull request #1838 from emankov/hipify-clang
[HIPIFY] Fix build failure against ToT trunk LLVM 11.0.0
2020-02-01 14:43:19 +03:00
Evgeny Mankov c396b5b6e5 [HIPIFY] Fix build failure against ToT trunk LLVM 11.0.0
+ Add an explicit cast when assigning StringRef to std::string
2020-02-01 14:41:16 +03:00
jujiang-del cb632005bc [dtest] Avoid resViewDesc for non-array resource in texture creation (#1823) 2020-01-30 14:04:41 -08:00
Evgeny Mankov f1c70ea6ba Merge pull request #1700 from scchan/object_load_error_check
add error checking for code object loading
2020-01-30 21:31:03 +03:00
satyanveshd 262ad13dd5 Match Occupancy APIs syntax with CUDA (#1625)
* Match Occupancy APIs syntax with CUDA and fix tests using these APIs
2020-01-29 13:05:53 -08:00
vsytch 6da0c82b78 Add missing texturePitchAlignment member to the hipDeviceProp_t struct. (#1802)
* Add missing texturePitchAlignment member to the hipDeviceProp_t struct.

* Add missing hipDeviceAttributeTexturePitchAlignment enumerator to the hipDeviceAttribute_t enum.

* Initialize texturePitchAlignment to 256. This works for gfx9+, but is technically overaligned in most cases for pre-gfx9.

* Add the texturePitchAlignment property to the NVCC path.
2020-01-27 16:37:00 -08:00
Evgeny Mankov e419d738e7 Merge pull request #1820 from emankov/HIP
[HIPIFY] Update Readme.md
2020-01-27 17:23:06 +03:00
Evgeny Mankov 5a788eb048 [HIPIFY] Update Readme.md
Restructure contents in order to have separated same-level information about hipify-clang and hipify-perl.
2020-01-27 17:21:32 +03:00
Evgeny Mankov 37da8171bc Merge pull request #1813 from emankov/HIP
[HIP][HIPIFY] Readmes actualization
2020-01-27 12:50:48 +03:00
Evgeny Mankov 5fae88f84b [HIPIFY][#1819] Fix for not found clang's system header files
Use absolute paths instead of relative ones.
2020-01-27 12:37:26 +03:00
Evgeny Mankov e57c2b3147 Merge branch 'master' of https://github.com/ROCm-Developer-Tools/HIP into HIP 2020-01-27 12:08:22 +03:00
vsytch e419a1609f Update the HIP_TRSF_* flags to match their Cuda equivalents. (#1801) 2020-01-24 11:41:15 -08:00
Evgeny Mankov 2b77ca7d2e Merge branch 'master' of https://github.com/ROCm-Developer-Tools/HIP into HIP 2020-01-24 16:14:16 +03:00
aakanksha555 ef3dbc1f91 Fix for a syntax error in deb packages (#1814) 2020-01-24 16:23:47 +05:30
mshivama 0605426049 SWDEV-220503: this_grid().thread_rank() gives incorrect result (#1808)
* fix a minor bug while computing this.grid()::thread_rank()
2020-01-24 16:23:28 +05:30
paulfreddy 5cd5c62f29 Fix install script syntax error (#1805)
Fix hip-nvcc install warning on postinstall and prerm script
2020-01-24 16:22:57 +05:30
Siu Chi Chan 8fc7cad90f Fix associate code object symbols with host allocation bug (#1799)
The current implementation skips this procedure for a given device
object when a global symbol is found in the cache.  This is incorrect:

 - There could be other undefined globals that have not been previously
encountered further down the list
 - If a symbol is found in the cache, it doesn't need to be pinned again
but it still need to be defined for the current executable

Added special case for the printf buffer symbol (already pinned by HCC)

The bug was exposed by running printf on different GPUs.
2020-01-24 16:22:49 +05:30
Alexey Chernov 9f5a4148ce Clear HIP_PATH before version detection (#1786)
Don't allow `HIP_PATH` to be propagated to `hipconfig`,
when run by CMake to detect the package version, as it
leads to the wrong version is detected: when there's
already HIP of some different version installed in the
system and `HIP_PATH` points to its location, `hipconfig`
tends to return the version of the installed HIP, rather
than the value defined for the distribution. The compiled
results report wrong version and spoils the rest of the
stack in this case.
2020-01-24 16:21:49 +05:30
Evgeny Mankov 0a380694c5 [HIP][HIPIFY] Readmes actualization 2020-01-23 20:55:10 +03:00
Evgeny Mankov 9f40056297 Merge pull request #1812 from emankov/HIP
[HIPIFY][doc] Readme - update versions
2020-01-23 18:04:39 +03:00
Evgeny Mankov 85b76bac70 [HIPIFY][doc] Readme - update versions 2020-01-23 18:01:48 +03:00
Evgeny Mankov ea9307b60b Merge pull request #1811 from emankov/HIP
[HIPIFY][cmake] Formatting
2020-01-23 17:48:52 +03:00
Evgeny Mankov 584ce03173 [HIPIFY][cmake] Formatting 2020-01-23 17:47:11 +03:00
Evgeny Mankov e563b67207 Merge pull request #1810 from emankov/HIP
[HIPIFY][cmake] Send error on CUDA 10.2 and higher as yet unsupported
2020-01-23 17:29:38 +03:00
Evgeny Mankov 9b739bfa15 [HIPIFY][cmake] Send error on CUDA 10.2 and higher as yet unsupported 2020-01-23 16:37:25 +03:00