AI/rocm-systems
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
12,484 Commits 3 Branches 0 Tags
fe2d747bcc363ed2bbd90a03316db5efdef01857
Commit Graph

12484 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Christophe Paquot b820c66c55 Default HostMalloc to uncached memory 
Change-Id: I72e19c7f7820a77fd5afc09f09cfea9acd0b8e84
 2020-04-03 19:19:33 -04:00
Saleel Kudchadker 74dc32537e Merge "Wake up commandQueue before returning" into amd-master-next 2020-04-03 18:27:03 -04:00
Michael LIAO 6fe3edc5a8 [vdi] Add hipFreeHost 
Change-Id: I8a5b7ff3f0ab4f5674efd6723c18808ad6ef33f5
 2020-04-03 16:34:28 -04:00
German Andryeyev 9e4aeb0f67 Merge "SWDEV-184709 - support hipLaunchCooperativeKernel()" into amd-master-next 2020-04-03 16:23:05 -04:00
Vladislav Sytchenko c9084d0ad2 Take into an account the number of channels... 
when querying the element size of an array.

Change-Id: Id57d3374b14d80a59230ec8286704f2fbabb0fae
 2020-04-03 15:43:18 -04:00
German Andryeyev 5efb3f26c0 SWDEV-184709 - support hipLaunchCooperativeKernel() 
- Add validation checks for cooperative launch to pass Cuda test

Change-Id: Ie296f0c3f113909d9a357879db3b2a833ab314c5
 2020-04-03 15:18:21 -04:00
Michael Hong Bin Liao 16ac35c4d5 Merge "Fix size type in __hipRegisterVar" into amd-master-next 2020-04-03 15:12:21 -04:00
Saleel Kudchadker e2def55164 Wake up commandQueue before returning 
Change-Id: I87eb5a22c81a9cb807474a960b5987d5fb6c2b86
 2020-04-03 10:23:36 -07:00
Aaron En Ye Shi 0bb217245c Merge "Fix path for hip-clang when using hipcc (#1961)" into amd-master-next 2020-04-03 12:28:50 -04:00
German Andryeyev 481cecec78 SWDEV-79445 - HIP improvement 
- Make sure only one GPU barrier is issued per dispatch
when memory tracking is disabled

Change-Id: I974569ab42a8835304a2930eef87b561a3750327
 2020-04-03 12:21:03 -04:00
Vladislav Sytchenko b82ab8b329 Merge branch 'amd-master-next' into amd-npi-next 
Change-Id: I3b91e45da7cb38c2a2c6d8675e7651de754764ec
 2020-04-03 12:11:22 -04:00
Michael LIAO d904f30f9e Fix size type in __hipRegisterVar 
Change-Id: I6b667600ae8f133583b768ab963318882b84179f
 2020-04-03 10:51:58 -04:00
Michael Hong Bin Liao 66aca97d6f Merge "[hip] Clean up unnecessary casting." into amd-master-next 2020-04-03 10:50:40 -04:00
kjayapra-amd cd7de89fc3 SWDEV-229840 - Improve Error Codes. RocKernel and RocProgram 
Change-Id: I8f785308e0562a50924f8bdd02e88c92a759f01a
 2020-04-02 22:30:53 -04:00
Saleel Kudchadker 49f3835666 Promote Khronos OpenCL-ICD-Loader master 
master as of https://github.com/KhronosGroup/OpenCL-ICD-Loader/
commit/c7fda8bb042760b5ead8650c64445f5972a64ad7

Change-Id: Ia852f55ecf0b0aceff915991328812ed9788698f
 2020-04-02 17:02:58 -04:00
German Andryeyev 2e948e4034 SWDEV-184709 - support hipLaunchCooperativeKernel() 
- Enable cooperative tests for single and multiple devices

Change-Id: I54b6713f578b6b5e670f117b17469c0091028c99
 2020-04-02 12:55:05 -04:00
Michael LIAO e3795436b2 [hip] Clean up unnecessary casting. 
Change-Id: I64b08aaef5c67ffb49330c9c605611f1fbd3f5a2
 2020-04-02 12:46:15 -04:00
Paul Fultz II 58f04fb774 Fix path for hip-clang when using hipcc (#1961) 
* Fix path for hip-clang when using hipcc

* Fix typo

* Update regex

Change-Id: I31bbee2e70d58b89191f970f5c6ae7e1c8b40900
 2020-04-02 12:09:31 -04:00
Paul Fultz II 631f82ab82 Add missing flags for hip::device target on hip-clang (#1230) 
This adds the missing compilation flags to hip::device so it can compile with hip-clang compiler.

Change-Id: Ie2b30ea606bfca385a0e84ae03ee0a8d828ad16a
 2020-04-02 12:09:03 -04:00
Chauncey Hui 55fad9b929 SWDEV-2 - Change OpenCL version number from 3115 to 3116 2020-04-02 03:00:03 -04:00
Saleel Kudchadker 5a357a795a Merge "OpenCL2.2 Header changes" into amd-master-next 2020-04-02 02:46:49 -04:00
Saleel Kudchadker 4092e3dee3 Update OpenCL 2.2 headers 
Update headers as of master https://github.com/KhronosGroup/OpenCL-Headers/
commit/9dde5da9c32cbd78656a5e2417c256922e851111

Change-Id: I698a3a1b92730c52b160b51af95469c2b54331c1
 2020-04-02 02:46:20 -04:00
Saleel Kudchadker a97f206ab1 Change for OpenCL2.2 header promotion 
Change-Id: Icd579e5c94af6e8d984a3f6c2606be44dec72e91
 2020-04-01 16:32:03 -04:00
Vladislav Sytchenko a09fadecf2 Add entry points for hipTexObject*() API 
Even though the runtime and driver texture object API is one to one, the structs used by these APIs are not. See hipResourceDesc vs HIP_RESOURCE_DESC differences.

These differences are not trivial and most likely won't be able to handled by hipify, so we need new API entry points.

Change-Id: Id4bcb1ad0ae15378dbdb5a2ed07e5ea30f320082
 2020-04-01 14:51:51 -04:00
Vladislav Sytchenko 2a223ecec6 (SWDEV-210228) 
Don't error when querying the number of devices if there are no devices present in the system.

We should just return 0 for the number of devices in this case and let the application handle this situation.

Change-Id: I20614ade5e649f3ce9ddd970d4b38bfe296f6cdb
 2020-04-01 13:36:44 -04:00
German Andryeyev 481d526859 SWDEV-184709 - support hipLaunchCooperativeKernel() 
- Enable cooperative groups support, based on ROCr capability

Change-Id: I975bcea0af7865009eaed24454ce71d897ea8fc4
 2020-04-01 12:13:33 -04:00
Chauncey Hui 39467ad9e6 SWDEV-2 - Change OpenCL version number from 3114 to 3115 2020-04-01 03:00:03 -04:00
eshcherb 1e5ea70588 adding hipApiString (#1989) 2020-04-01 12:10:33 +05:30
ansurya 68f1b814b1 tex1Dfetch behaviour for different address mode and filter mode (#1772) 
Fixes github issue: #1754

- When ResourceDesc::resType is hipResourceTypeLinear ignore address mode and filter mode.
- When textureDesc::normalizedCoords is set to zero, AddressModeWrap and AddressModeMirror won't be supported and will be switched to AddressModeClamp.
 2020-04-01 12:10:17 +05:30
Saleel Kudchadker 99a024fc14 Merge "Cleanup stream from hip:Event class." into amd-master-next 2020-03-31 20:14:48 -04:00
Vladislav Sytchenko 6e0722a5d0 (SWDEV-229354) 
This patch is a workaround to support user pitch for hipMemcpy{2D/3D}.

Historically OpenCL didn't support pitch with clEnqueueFillBuffer(), so neither did we in VDI. Adding it now will be slightly nontrivial, since the fill kernel and runtime in many places will need to be modified.

As a temporary workaround for cases when pitch > width, we can just enqueue a fill for each row separately. This implementation is slow, but it satisfies the correctness criteria.

Change-Id: Idfeca349288b51d6ff84a7cf001fb63c6a66818a
 2020-03-31 18:12:56 -04:00
Todd tiantuo Li b96e25c5f8 SWDEV-222615 - delete VirtualGPU::Queue and return NULL when Pal::IDevice::CreateQueue() fails, in order to avoid undefined behavior and crash. 
Change-Id: I4b94d3b3b0495530a8bf7840dc77a98aed9aa8ae
 2020-03-31 16:50:53 -04:00
Reshabh Sharma 36f24a40e5 Output file name should not change flags picked for compiler (#1938) 
Fixes SWDEV-207362,

The output file name should not contribute to picking up the right flags for the compiler. This fix solves issues when the output has conflicting extensions which confuses hipcc to treat them as the source files and add the required flags for them.

PS: Output file refers to the file followed by -o

Change-Id: I1095966c11143ad73e81fabc35b4e9de5d3afada
Example: hipcc test.o -o test.hip will add the flags for .hip compilation ignoring the fact that it is an output file
 2020-03-31 16:13:43 -04:00
Saleel Kudchadker 65722f8cca Cleanup stream from hip:Event class. 
Change-Id: I98de07d33bb7fea8f5e2d32b288c15f10ce58902
 2020-03-31 11:22:00 -07:00
Michael Hong Bin Liao 9420f56c66 Merge "[hipcc] Remove the previous workaround." into amd-master-next 2020-03-31 10:05:56 -04:00
Chauncey Hui b0164a9819 SWDEV-2 - Change OpenCL version number from 3113 to 3114 2020-03-31 03:00:03 -04:00
Michael LIAO a14695d4eb [vdi] Fix hipGetSymbol{Address|Size} 
- Use symbol value as the qeury key. Compared to the symbol name, the
  symbol value is more robust as developers may use unqualified or
  qualified identifiers. It also removes the mangling and/or demangling
  requirement for the runtime API.

Change-Id: I9d4259f3842612c7cc98551269fc2092d8b5c19e
 2020-03-31 00:26:53 -04:00
German Andryeyev 9f795d750d SWDEV-224639 - Device memory is still occupied 
- Add cache free on OCL context destroy
- Remove std::mem_fun() usage, since it was removed in c++17

Change-Id: If6acd08f13a2298912ecd78fc025dcf0b32aee54
 2020-03-30 17:39:21 -04:00
German Andryeyev 7ef8dfdfe7 SWDEV-184709 - support hipLaunchCooperativeKernel() 
Add ROCr cooperative queue allocation

Change-Id: I1384482692f4080d31255b09e0f68a21ccad3da8
 2020-03-30 16:09:09 -04:00
Christophe Paquot 47718cbf16 Do not retry to allocate when OOM. Shouldn't be needed since we idle on Free. 
SWDEV-229214

Change-Id: I183006f409388e3c7981f2569649d01d6378be46
 2020-03-30 12:49:48 -07:00
kjayapra-amd 7356a74d35 SWDEV-216213 - Use different static & dynamic module maps for faster lookup. 
Change-Id: Ia605e76a411ad5be04046b9d61f1ac111d49bb4a
 2020-03-30 14:28:07 -04:00
Anusha Godavarthy Surya 9d213070b0 Merge "Update Enable/Disable peers to match cuda behaviour" into amd-master-next 2020-03-30 13:46:42 -04:00
agodavar 96d119ac77 Update Enable/Disable peers to match cuda behaviour 
Change-Id: I67194ccf77a0019368579ff7d95b7790fcf228f3
 2020-03-30 12:49:16 -04:00
Tao Sang 44a145b2a3 Merge "enable HCC printf when using hip-clang" into amd-master-next 2020-03-30 11:37:23 -04:00
Michael LIAO 64bbce8e73 [hipcc] Remove the previous workaround. 
Change-Id: Ie1c2aa21689bc343c10fd4b699546cf500ced5b2
 2020-03-30 10:18:47 -04:00
Michael LIAO 55d869df99 [vdi] Fix calculation of MaxWaves 
- Consider the case where `usedVGPRs` is zero.
- This fixes [SWDEV-228537](http://ontrack-internal.amd.com/browse/SWDEV-228537)

Change-Id: I8675311f5fe24fb59c5d45bada122afefb55b128
 2020-03-30 09:10:16 -04:00
Sarbojit2019 b55775a49e Fix for segfault seen in hipMemcpyDtoD (#1964) 
* Fixes SWDEV-227444.
 2020-03-28 17:29:49 +05:30
Maneesh Gupta 062938a14c Remove address_space(1) typecast and use __ockl_atomic_add_noret_f32 (#1956) 
* Remove address_space(1) typecast for ockl_global_atomic_add_f32
* use __ockl_atomic_add_noret_f32
 2020-03-28 17:28:33 +05:30
Sameer Sahasrabuddhe c014c733c6 enable HCC printf when using hip-clang 
This is cherry-picked from PR#1947 that was committed to the
github repo. It allows printf to work with hip-clang and HCC
runtime.

Change-Id: I754753250ea1e694cf3441722e2d4c9d25fa75bc
 2020-03-28 00:18:21 -04:00
Saleel Kudchadker a5c287554e Merge "Check event status before notify" into amd-master-next 2020-03-27 20:19:40 -04:00