089ab3b947
Switch the atomic implementation to use Clang builtins.
2018-06-02 12:27:17 +01:00
14e449b5bb
Remove vestigial implementations.
2018-06-02 11:37:08 +01:00
1a02bc364f
Add integrated device property
2018-06-02 13:11:16 +05:30
417869821d
Re-sync with upstream.
2018-06-01 15:49:05 +01:00
df450c6680
Merge pull request #484 from gargrahul/fix_malloc_hiphostreg
...
Fix memcpy2D for malloc+ hostRegister
2018-06-01 16:53:25 +05:30
bdf2645713
Merge pull request #466 from ROCm-Developer-Tools/feature_use_Float16
...
Feature use _Float16 and match CUDA __half behaviour.
2018-06-01 13:50:12 +05:30
ab4b2a650b
Re-sync with upstream. Add integer abs.
2018-05-31 16:38:00 +01:00
8d6357669d
Fix memcpy2D for malloc+ hostRegister
2018-05-31 13:14:27 +05:30
59db16fd36
Switch to using ROCDL directly, as opposed to via HC. Add missing bits.
2018-05-31 03:17:26 +01:00
57fb96013c
Merge pull request #472 from Jorghi12/patch-3
...
Adding double/long int signatures for abs
2018-05-30 08:32:14 +05:30
ec2edb2c92
Update math_functions.cpp
...
CUDA also has a function named labs.
2018-05-26 16:21:14 -04:00
4383d6c6de
Adding double/long int signatures for abs
...
Adding overloads for abs that are found in cuda's math_functions.
2018-05-26 00:40:14 -04:00
d8cb47242b
Use 64x4 grid dims
2018-05-24 23:51:52 +05:30
4ff059d641
Clean up and fix remaining bytes copy
2018-05-24 23:30:27 +05:30
9c7fbdb597
Remove vestigial inline LLVMIR.
2018-05-24 12:46:14 +01:00
981e56a68f
Fix memcpy2d kernel dims
2018-05-24 17:00:12 +05:30
dc179e0c33
Correct remaining bytes in copy 2d kernel
2018-05-24 08:27:24 +05:30
6f819f226b
Missing commit.
2018-05-23 17:57:47 +01:00
9a76d5b94c
Optimize memcpy2D kernel use
2018-05-23 14:43:47 +05:30
323a6226b0
Merge pull request #464 from gargrahul/fix_memcpy2d_pinned_mem_case
...
Fixed memcpy2D for pinned memory case using 2D kernel
2018-05-22 10:42:28 +05:30
df3bb9fc32
Merge pull request #445 from ROCm-Developer-Tools/feature_func_attributes
...
Add support for the hipFuncGetAttributes interface.
2018-05-22 09:37:41 +05:30
f47a8236d7
Fixed memcpy2D for pinned memory case using 2D kernel
2018-05-21 22:14:45 +05:30
0180a82963
hipMemcpy returns success if sizeBytes is 0.
...
Fixes SWDEV-153754 & SWDEV-154178.
2018-05-21 15:38:44 +05:30
cac3f1c7cd
Merge pull request #455 from ROCm-Developer-Tools/magic
...
Change HIP fat binary magic number
2018-05-21 09:52:03 +05:30
cd6c979c27
Update hip_module.cpp
...
Typo.
2018-05-18 17:50:45 +01:00
afe62e7030
Fix for memcpy2DAsync for pinned host memory case
2018-05-18 21:09:50 +05:30
03ac8e6a92
Merge pull request #433 from gargrahul/add_hipmemset3d
...
Added hipMemset3D
2018-05-18 14:54:15 +05:30
ac7713fa34
Merge pull request #448 from 949f45ac/master
...
Provide correct __mul64hi and __umul64hi builtins, using code from ROCm-Device-Libs
2018-05-18 13:18:16 +05:30
d079463887
Change HIP fat binary magic number
2018-05-17 17:04:51 -04:00
8303bfdffd
Reinstate accidentally deleted uchar2Holder
2018-05-17 10:55:45 +02:00
8f010ac68e
Fixed hipMemcpy2D to handle 1D memcpy case
2018-05-16 11:07:10 +05:30
5325b6535e
Update hip_module.cpp
2018-05-14 17:15:36 +01:00
79480d7cbd
Provide correct __mul64hi and __umul64hi builtins, using code from ROCm-Device-Libs
2018-05-14 08:34:56 +02:00
1ba8a35dba
Don't use magic constants, they're evil.
...
Also clarify that the register count cannot be queried at the moment.
2018-05-11 11:31:46 +01:00
13274ce559
Add support for the hipFuncGetAttributes interface.
2018-05-11 03:35:10 +01:00
da302c3e93
Added hipMemset3D
2018-05-07 10:24:30 +05:30
6411ca1f6d
Null checks added for hipmallocpitch and hipmemcpy apis
2018-05-03 09:27:50 +05:30
9de5f23d54
Fix texture 3D for HIP/NVCC
2018-05-02 11:56:37 +05:30
e17ade3afb
Merge pull request #415 from deven-amd/master
...
Checkin to fix bugs in math functions.
2018-05-01 12:29:03 +05:30
af586bbbf2
Checkin to fix bugs in math functions.
...
This change fixes the following bugs that were discovered while debuggnig TF unit test failures (cwise_ops_test)
1. __hisinf and __hisnan routines
Both had incorrect implementations.
2. abs
A "long long" (64bit int) version was missing, resulting in the 32bit version being used for 64bit ints (which resulted in incorrect results, when the value passed in was outside the 32bit int range)
3. lgamma
We seemed to have a custom version for the 'double' datatype (which was giving incorrect results). Replaced it with a call to the 'double' version of the underlying 'hc::precision_math::lgamma'
2018-04-24 18:10:07 +00:00
1c2509dc04
SWDEV-141024
2018-04-20 17:40:00 +05:30
3cfb9c0d40
Added hipMemset2DAsync support
2018-04-17 18:27:27 +05:30
33ada06de9
Merge pull request #400 from gargrahul/hipModule_cleanup
...
hip_module code cleanup
2018-04-17 09:00:15 +05:30
5a80c16cd2
Clean up triple_to_hsa_isa naming ( #403 )
2018-04-17 06:47:55 +05:30
1446f78799
hip_module code cleanup
...
-Fixed missing ihipLogStatus in hipModuleLoad()
-Fixed some ihipXXX functions
2018-04-16 15:35:04 +05:30
1a281240d6
Fix bug when using older rocr naming ( #399 )
2018-04-14 15:47:46 +05:30
ab7e727fa2
Fix new gfx target naming convention ( #397 )
...
This was introduced in recent xnack changes naming changes which affected our offload bundler. This patch will fix issues in HIP samples, ROCR tests, and others.
2018-04-13 16:24:39 +05:30
16c89d101a
Correct missed ihipMemsetCopyDataType change
2018-04-12 10:27:19 +05:30
3d6eb75828
Changed ihipMemsetCopyDataType to ihipMemsetDataType
2018-04-12 09:29:22 +05:30
294bf50f68
Fix hipMemset stream resolution
2018-04-11 19:01:53 +05:30
Alex Voicu
Rahul Garg
Maneesh Gupta
Jorghi12
Yaxun (Sam) Liu
949f45ac
Lakhan Singh
Deven Desai
Aaron Enye Shi