AI/rocm-systems
2
0
Форкнуть 0
Код Задачи Запросы на слияние Действия Пакеты Проекты Релизы Вики Активность
3 714 коммитов 3 Ветки 0 Теги
16632b627cbaea28d95fd899fcb59e2130ce3939
Граф коммитов

203 Коммитов

Автор SHA1 Сообщение Дата
Evgeny Mankov 24af494e97 [HIP] Fix segfault on uninitialized struct members in hipArrayCreate and hipArray3DCreate 
[ROCm/clr commit: 299fbd4842]
 2019-07-12 16:38:26 +03:00
Evgeny Mankov b5f0cdaa7b [HIP][HIPIFY] Split HIP_ARRAY_DESCRIPTOR struct to HIP_ARRAY_DESCRIPTOR and HIP_ARRAY3D_DESCRIPTOR 
[Reason] To be compatible with CUDA [#1133]

Update HIP code, hipify-clang, tests and docs

[TODO] Add support of the corresponding functions on nvcc fallback path


[ROCm/clr commit: f0832fd968]
 2019-07-11 14:58:16 +03:00
Jatin Chaudhary 6e8edf8890 Adding bounds check before hipMemset (#1190) 
* Adding bounds check in ihipMemset

* Adding ihipMemPtrGetInfo to hipMemPtrGetInfo


[ROCm/clr commit: fcb0a3d4e2]
 2019-07-08 11:00:38 +00:00
Anusha Godavarthy Surya e643bae27d Added missing NULL checks and corrected API return values as per validation 
[ROCm/clr commit: 4989452413]
 2019-06-27 00:19:05 +05:30
Evgeny Mankov 204043c6e0 [HIP][HIPIFY] Make hipMemcpyParam2D coherent with cuMemcpy2D 
+ Makes hip_Memcpy2D struct compatible with CUDA_MEMCPY2D struct
+ Add hipMemcpyParam2D support in nvcc fallback path
+ Update hipify-clang, tests and docs accordingly


[ROCm/clr commit: 9cb3e9aa5e]
 2019-05-22 18:31:39 +03:00
Rahul Garg d44e800a17 Add fine grained host memory lock support (#1095) 
* Add fine grained host memory lock support

* Fix default flag check


[ROCm/clr commit: e1f3dc0c80]
 2019-05-13 11:48:26 +05:30
Rahul Garg d69edbbb7f Add hipMallocManaged default functional support (#1036) 
* Add hipMallocManaged default functional support

* Fix build error

* Add dtest


[ROCm/clr commit: 94769fc8dd]
 2019-04-24 16:50:03 +05:30
Jeff Daily cf4e198a91 In hipFree, synchronize owner of memory (#1018) 
* In hipFree, if memory is associated with a device, synchronize that device's streams.

This changes the behavior from synchronizing the currently set TLS device.

* All devices sync in hipFree for _appId=-1 case.

* Revert "All devices sync in hipFree for _appId=-1 case."

This reverts commit 1efb34d6a8426661e45bc5f763422a1147aeac10.

* add HIP_SYNC_FREE env var


[ROCm/clr commit: cf8fb43e6b]
 2019-04-16 08:35:55 +05:30
Rahul Garg 73bb9a74bb Handle D2D in memcpy2D 
[ROCm/clr commit: 50d623981e]
 2019-03-28 02:21:45 +05:30
Rahul Garg 0d47ae4203 Let hipHostMalloc always share/map pinned host ptr 
[ROCm/clr commit: 9b38380c03]
 2019-03-26 10:19:13 +05:30
Rahul Garg 21d7bbab11 Avoid double mapping of devices to hostMalloc buffer 
[ROCm/clr commit: ad11972f47]
 2019-03-25 23:07:05 +05:30
Maneesh Gupta 817e064745 Merge pull request #970 from mangupta/swdev-172995 
hipExtMallocWithFlags implementation

[ROCm/clr commit: c20d233585]
 2019-03-25 07:46:53 +00:00
Maneesh Gupta 505fc1e98c hipExtMallocWithFlags needs hcc workweek 19115 or higher 
[ROCm/clr commit: 45255ab492]
 2019-03-25 11:41:20 +05:30
Maneesh Gupta f1d064562d hipExtMallocWithFlags implementation 
Change-Id: Iee9e119796472200b2933d5e23be60813f33bc75


[ROCm/clr commit: e44de376f7]
 2019-03-19 11:59:22 +05:30
Rahul Garg da6653482d Add 2D fallback to use copy kernel 
[ROCm/clr commit: af72cde0a1]
 2019-03-14 13:03:06 +05:30
Alex Voicu 45f4ac5023 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.


[ROCm/clr commit: ed48847237]
 2019-03-06 17:31:44 +05:30
Wen-Heng (Jack) Chung e34b0ccd48 Address code review comments to use hipDeviceptr_t 
[ROCm/clr commit: 8b7baa0bd9]
 2019-03-05 05:51:05 +00:00
Wen-Heng (Jack) Chung 2706bf46f2 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


[ROCm/clr commit: 392271f4db]
 2019-03-04 17:00:33 +00:00
Wilkin Chau 19b361281a Fix hipMemset3D test 
Calculate the allocated size based on the width, height and depth.


[ROCm/clr commit: 99540373cf]
 2019-02-28 22:42:46 +00:00
Evgeny a4df59d819 fixing HSA_INIT_API cid args 
[ROCm/clr commit: 47625cb8fd]
 2019-01-16 23:45:44 -06:00
Maneesh Gupta f096437107 Merge pull request #797 from gargrahul/fixhipPointerGetAttributes 
Fixed hipPointerGetAttributes for hostmalloced ptr

[ROCm/clr commit: a778f7cdf7]
 2018-12-12 10:16:07 +05:30
Maneesh Gupta 30ce297635 Merge pull request #608 from gargrahul/add_pinned_2d_sdma_copy 
Added support for pinned 2D SDMA copy

[ROCm/clr commit: 6ce99b066c]
 2018-12-12 07:44:16 +05:30
Rahul Garg 514e5a41e3 Fixed hipPointerGetAttributes for hostmalloced ptr 
[ROCm/clr commit: 77fd517e09]
 2018-12-08 01:42:08 +05:30
Maneesh Gupta 938c731b27 Merge pull request #760 from eshcherb/roctracer-hip-frontend-181113 
Roctracer hip frontend 181113

[ROCm/clr commit: 99bb89b756]
 2018-11-23 11:08:25 +05:30
Maneesh Gupta f0976b3c37 Merge pull request #748 from mkuron/getsymboladdress 
Implement hipGetSymbolAddress and hipGetSymbolSize

[ROCm/clr commit: 40d3184dd1]
 2018-11-21 10:32:01 +05:30
Michael Kuron 4b77b5c1ba Merge branch 'master' into getsymboladdress 
[ROCm/clr commit: e9b88711e2]
 2018-11-20 12:03:22 +01:00
Rahul Garg 796765d53e Fix hipHostRegister 
[ROCm/clr commit: aae87e21d2]
 2018-11-17 05:38:35 +05:30
Evgeny 96b7f0404e renaming HIP_INIT_CB_API to HIP_INIT_API 
[ROCm/clr commit: e362688adf]
 2018-11-13 15:33:26 +00:00
Evgeny 7519b3f75e adding activity prof layer 
[ROCm/clr commit: 084a68be63]
 2018-11-13 15:33:26 +00:00
Rahul Garg 696a6df141 Fixed hipMemcpyToSymbol doesn't work on GPU other than device 0 SWDEV-166881 
[ROCm/clr commit: ac32566d9b]
 2018-11-13 00:49:20 +05:30
Michael Kuron a0c35ab823 Use correct trace macro in hipGetSymbolAddress/hipGetSymbolSize 
[ROCm/clr commit: 357dc8be11]
 2018-11-06 20:46:30 +01:00
Michael Kuron a857354072 Introduce ihipModuleGetGlobal 
[ROCm/clr commit: 4da2d92281]
 2018-11-06 09:54:34 +01:00
Michael Kuron a38366c81e Implement hipGetSymbolAddress and hipGetSymbolSize 
[ROCm/clr commit: 0b6f5791f8]
 2018-11-04 10:39:34 +01:00
Siu Chi Chan e26a728173 Move the global arrays for hip malloc/free 
from a header into a source file such that
there's only an unique copy in an executable
and prevent wasting static memory on the host

Change-Id: Id5b62766f77809c8d7b47892cb7149c490dcbdb9


[ROCm/clr commit: cdd0109e70]
 2018-11-01 16:20:35 -04:00
Anton Gorenko 600e5cdb58 Fix allocation size of arrays with multiple and/or non-32-bit channels 
hipMallocArray and hipMalloc3DArray must use sum of bits
of all components.


[ROCm/clr commit: 6e6297f3cd]
 2018-10-29 18:12:00 +06:00
Rahul Garg e0efc8a09e Return hipSuccess when sizeBytes=0 in hipMemset 
[ROCm/clr commit: bd27310127]
 2018-09-26 12:47:36 +05:30
Rahul Garg 17e35e382c Added support for pinned 2D SDMA copy 
[ROCm/clr commit: 5eb11b58f3]
 2018-07-31 14:05:35 +05:30
Sarunya Pumma 57c15b815d Remove device mapping from shareWithAll memory 
When shareWithAll memory (e.g., host memory) is allocated, set appId
in hc::AmPointerInfo to -1 to indicate that this memory is not mapped
to any device.  Peer checking in ihipStream_t::canSeeMemory is not
necessary if memory is shared with all devices.  Thus, it is skipped.

Note that earlier host memory is always mapped to device 0 and HIP
always performs peer checking for all kinds of hipMemcpy.  Since the
peer checking process requires context locking, hipMemcpy from/to host
memory always grabs device 0's context lock.  Therefore, if there is
another thread holding the context lock of device 0 (e.g.,
hipDeviceSynchronize on device 0), hipMemcpy will have to wait for the
lock until it can actually perform memcpy.  This can significantly
deteriorate execution performance.

Signed-off-by: Sarunya Pumma <sarunya.pumma@amd.com>


[ROCm/clr commit: 84aadb9274]
 2018-07-28 23:15:16 -07:00
Rahul Garg a796eedde9 Revert "Use memcpy kernel for all pinned memory cases in hipMemcpy2DAsync" 
[ROCm/clr commit: f554e48db3]
 2018-07-02 14:32:11 +05:30
Rahul Garg 1c2acf1467 TEMP- fix memcpy2dAsync for trsm issue 
[ROCm/clr commit: 007e2a4b5f]
 2018-06-15 16:08:29 +05:30
Rahul Garg a41cb78337 Fix stream resolution in memcpy2dasync 
[ROCm/clr commit: 2ae3be9773]
 2018-06-14 11:58:56 +05:30
Rahul Garg dfe8629088 Fix retrieved locked ptr offset 
[ROCm/clr commit: 68554e155b]
 2018-06-13 23:10:05 +05:30
Maneesh Gupta e34247ca19 Merge pull request #497 from gargrahul/fix_memcpy3d_fastpath 
Fix hipMemcpy3D for fast path

[ROCm/clr commit: 53037472ff]
 2018-06-06 14:44:02 +05:30
Rahul Garg 914a0aaf9c Fix hipMemcpy3D for fast path 
[ROCm/clr commit: 163d4a5b03]
 2018-06-05 18:54:33 +05:30
Rahul Garg 3a6e8aadbd Fix depth value for 3D allocations 
[ROCm/clr commit: fa6ce7a724]
 2018-06-04 18:00:22 +05:30
Rahul Garg 84cb355df6 Fix memcpy2D for malloc+ hostRegister 
[ROCm/clr commit: a3609eaf61]
 2018-05-31 13:14:27 +05:30
Rahul Garg 0ab5c9d6f1 Use 64x4 grid dims 
[ROCm/clr commit: 27e0566c3a]
 2018-05-24 23:51:52 +05:30
Rahul Garg 439655d213 Clean up and fix remaining bytes copy 
[ROCm/clr commit: aed2653857]
 2018-05-24 23:30:27 +05:30
Rahul Garg 4fc64ff4c5 Fix memcpy2d kernel dims 
[ROCm/clr commit: 53a7c61e9b]
 2018-05-24 17:00:12 +05:30
Rahul Garg 4e52a634a1 Correct remaining bytes in copy 2d kernel 
[ROCm/clr commit: 96b4618d26]
 2018-05-24 08:27:24 +05:30