AI/rocm-systems
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
2,397 Commits 3 Branches 0 Tags
295cd826d9dd59cb398d0680834f99a678a2aecc
Commit Graph

80 Commits

Author SHA1 Message Date
Rahul Garg 181e0ee8ee hip_module code cleanup 
-Fixed missing ihipLogStatus in hipModuleLoad()
-Fixed some ihipXXX functions


[ROCm/clr commit: c23898f49b]
 2018-04-16 15:35:04 +05:30
Maneesh Gupta 46ddefedee Apply .clangformat to all repo source files 
Change-Id: I7e79c6058f0303f9a98911e3b7dd2e8596079344


[ROCm/clr commit: 9e47fccc89]
 2018-03-12 11:29:03 +05:30
Alex Voicu eab8624938 Re-sync with upstream. 
[ROCm/clr commit: a704bd8b44]
 2018-02-12 20:20:24 +00:00
Rahul Garg f15543d5cf Fixed host allocated globals address lookup for host usage 
Fixed texture driver APIs failure


[ROCm/clr commit: b8c23f979b]
 2018-01-30 18:06:31 +05:30
Rahul Garg 1ee7e7c2f9 Fixed build error 
[ROCm/clr commit: d1dcc5025d]
 2017-12-28 16:15:45 +05:30
Alex Voicu aa48cc7b55 This introduces LipoProteinLipase (lpl), a simple tool for creating fat binaries. It represents a direct replacement of the creaky hccgenco.sh script, which had various issues. The format it uses is that of a code object bundle, generated by the Clang Offload Bundler. The output is always suffixed with the ".adipose" extension. It is shared with HCC. The hipcc script and associated tests are modified to use lpl. Help can be obtained by invoking lpl --help. A more computer-sciency / corporate friendly name is likely to be beneficial, which is a reason for choosing easily searchable/replaceable names such as lpl or adipose. 
[ROCm/clr commit: 4e0739c68a]
 2017-12-08 04:22:57 +00:00
Alex Voicu 37e40fcd40 This is primarily intended as an additional cleanup of the module functionality, in the aftermath of adopting module based dispatch. The main effort was associated with refactoring the questionable ihipModuleGetSymbol. It was quaintly written and misleading, in that it had little to do with getting symbols, and was exactly retrieving a kernel object. Error handling is modified so as to reduce branching depth. Functions which serve as interfaces to the HSA RT are moved in a separate helper header. Code object readers are properly deleted. Some leftover dead functionality pertaining to associating namespace scope variables with their allocated memory is removed. Executable loading is changed to use a string which holds the ELF image of the code object being loaded, thus avoiding some corner cases where using a istream would fail. 
[ROCm/clr commit: e186bd9533]
 2017-12-03 23:09:06 +00:00
Alex Voicu cdac21c81a Choose whether or not to use functional grid_launch based on the version of HCC used to compile. 
[ROCm/clr commit: c6ab6f292b]
 2017-11-29 00:17:44 +00:00
Alex Voicu 914acbe88f Re-sync with upstream and re-factor platform global management for texture references. 
[ROCm/clr commit: 6a0efb7ed2]
 2017-11-28 19:15:29 +00:00
Alex Voicu 579a3187da Merge remote-tracking branch 'origin/master' into feature_use_module_based_dispatch_instead_of_pfe 
# Conflicts:
#	src/hip_module.cpp


[ROCm/clr commit: d37a5a6008]
 2017-11-28 17:29:11 +00:00
Rahul Garg 212afdf56b Fixed review comments 
[ROCm/clr commit: 38029f2849]
 2017-11-21 21:19:06 +05:30
Alex Voicu 1366b2d5ae Refactor the __device__ versions of memset and memcpy to be less awkward i.e. not return nullptr as opposed to the destination pointer (it can only be assumed it was done for maximum confusion) and actually unroll as they claim to. Change all of the {to, from}Symbol functions to use hipModuleGetGlobal, as opposed to hc::accelerator::get_symbol_address which is no longer valid with module based dispatch. 
[ROCm/clr commit: f8c1c1b38e]
 2017-11-21 02:40:34 +00:00
Rahul Garg 9cee9705dc -Moved coGlobals in hipModule class (takes care of multi module case) 
-Used mutex scope for updating coGlobals


[ROCm/clr commit: eff09fcd86]
 2017-11-20 16:23:18 +05:30
Rahul Garg 7453c8b25c Update hipModuleGetTexRef API 
[ROCm/clr commit: 76f174b536]
 2017-11-19 22:10:46 +05:30
Alex Voicu 93df8a2a57 Merge remote-tracking branch 'origin/master' into feature_use_module_based_dispatch_instead_of_pfe 
[ROCm/clr commit: 1518611976]
 2017-11-09 23:43:07 +00:00
Rahul Garg bb8c31c198 Texture driver APIs support 
[ROCm/clr commit: 1851c153f6]
 2017-11-09 22:10:55 +05:30
Alex Voicu e35ce82eff This introduces correct support for agent global variables, and implements hipModuleGetGlobal as an actual equivalent for cuModuleGetGlobal. 
[ROCm/clr commit: d12cf0da7d]
 2017-11-03 01:44:48 +00:00
Ben Sander 19935275ea Merge pull request #245 from scchan/centos_fixes 
various fixes for centos/rhel

[ROCm/clr commit: 5a80c4108e]
 2017-11-01 18:10:29 +01:00
Alex Voicu b738633bcb This switches HIP from its currently convoluted macro + pfe based dispatch mechanism to a more natural one partially based on the existing module API. The basic idea is that HCC will always correctly emit __global__ functions: as empty-bodied stubs, on host, and as kernels, on device. It then becomes trivial to obtain the mangled name on host, at dispatch, from the function's address, and then to use the mangled name to retrieve the kernel. This should address all problems stemming from serialisation, dubious mismatches due to the manufactured functor, macro-isms et al. It also immediately enables support for generalised globals as a consequence of that being available in the module API. Finally, it will make debug much easier, since the actual names of the __global__ functions will automatically be used in traces etc. One detail is that due to how dispatch works now (hipLaunchKernel and hipLaunchKernelGGL are themselves variadic function templates which deduce the function type of the callee), in certain cases it may be necesssary to insert explicit casts to ensure that the variadic argument list selects a viable overload - this can be observed in some unit tests. Eventually we may be able to remove this limitation, but for now it does not appear terribly onerous. The code is not extremely HIPpie, nor is it fully optimised, but rather is intended as a starting point for the HIP team to make its own. 
[ROCm/clr commit: 28f87f7d2e]
 2017-11-01 15:09:59 +00:00
Siu Chi Chan 2094b92a3e Centos/RHEL - remove usage of constexpr since libc++ doesn't enable ctor for constexpr pair in C++11 
[ROCm/clr commit: c1a4ff7b20]
 2017-10-31 18:16:12 +00:00
Ben Sander 4df6b371ab Add ns-level timer for HIP API routines 
Refactor some miuses of ihipLogStatus, these should only be in top-level
HIP APIs and should be paired with HIP_API_INIT calls.


[ROCm/clr commit: dc7d993a02]
 2017-10-30 20:20:51 +00:00
Maneesh Gupta 941682c701 Make elfio headers private 
Change-Id: I3ba174bb46e84a75380207d93a0da6fe3703689e


[ROCm/clr commit: 281c543bc3]
 2017-10-23 10:24:36 +05:30
Alex Voicu 1c8db33de9 This fixes incorrect usage of the reader object, which created arcane 
mismatches when one reader accessed another's section.


[ROCm/clr commit: b222448efd]
 2017-10-09 15:46:38 +01:00
Alex Voicu d9075301c3 This adds cursory support for globals to the HIP module loading API. The 
style is purposefully alien so as to signal that HIP experts should turn
it into HIP worthy code as soon as possible.


[ROCm/clr commit: 582779c6c6]
 2017-10-09 13:27:11 +01:00
Maneesh Gupta 51c48e4cef Bump min hcc_workweek required for named kernel dispatch to 17312 
Change-Id: I8c7b58306b279ed113d03260e4bc6086bb8b4e68


[ROCm/clr commit: dabd6d2bc6]
 2017-08-08 11:08:55 +05:30
Ben Sander 004080bff8 Add workweek check to make sure we have a new enough compiler 
[ROCm/clr commit: 2781293e89]
 2017-07-27 23:00:58 -05:00
Ben Sander c8b72cb926 Pass kernel name to HCC dispatch_hsa_kernel, for debug/profile 
[ROCm/clr commit: a282351763]
 2017-07-27 22:00:15 -05:00
Ben Sander ccc085e4db Add option to pass names to HCC dispatch API (for debug) 
[ROCm/clr commit: c7382f7da6]
 2017-06-23 18:05:30 -05:00
Ben Sander 81354999e8 Set event->_stream on hipHccModuleLaunchKernel path if start/stop used 
Ensure _stream is always non-null in recorded events.
Fixes isDefaultStream fault.


[ROCm/clr commit: 6cc5dc0326]
 2017-05-30 21:55:46 -05:00
Evgeny Mankov 052f630bd3 [HIP] [HIPIFY] [FIX] cuModuleLoadDataEx -> hipModuleLoadDataEx 
https://github.com/GPUOpen-ProfessionalCompute-Tools/HIP/issues/81

1.	Do not use JIT options on HCC path, call hipModuleLoadData instead.
2.	NVCC path is unchanged, to call cuModuleLoadDataEx with all options.
3.	Get rid of manual hipification, based on #ifdef #else for NVCC/HIP.
4.	Update documentation accordingly.


[ROCm/clr commit: ae9f14ef9c]
 2017-05-19 17:22:14 +03:00
Ben Sander 4ac6ac9d1d Add initial HIP_SYNC_NULL_STREAM=0 mode. 
This eliminates host-synchronization for null stream.  Instead, the
null-stream uses GPU-side events to wait for other streams.
Default is OFF pending additional testing.

Add enhanced null-stream test.

Also refine HIP_TRACE_API.


[ROCm/clr commit: 8bc6ee5932]
 2017-05-16 19:04:25 -05:00
Ben Sander dc9e957a36 Finish adding start/stop event support to hipHccModuleLaunchKernel. 
Change interface to use hipEvent_t rather than hipEvent_t*

Change-Id: I259062dc087a13d51dc27f84e1e8861f332a104d


[ROCm/clr commit: e9eaadd135]
 2017-04-06 21:02:50 -05:00
sunway513 1897ca4a20 Refactor events and add initial event option for hipHccModuleLaunchKernel 
- Change hipEvent_t to a class.
- Move event logic inside the class.
- Add _type to support Independent, StartCommand, StopCommand events.
  StartCommand returns start timestamp from events.

Change-Id: I4ddd694f2645a3ff7170c9111dc1d3e39931ca21


[ROCm/clr commit: f442e975c6]
 2017-04-06 21:02:50 -05:00
Aditya Atluri 28c8d15412 added debug support for HIP sample 
Change-Id: Ia7265234082039b68114f7421f4dbcb7149d4d2b


[ROCm/clr commit: 81d3725a16]
 2017-03-31 14:13:46 -05:00
Aditya Atluri 3651cb672a added new api hipHccModuleLaunchKernel 
1. hipHccModuleLaunchKernel is same as hipModuleLaunchKernel with OpenCL workitem model
2. Added copy right
3. Fixed header naming

Change-Id: I6a7c35a3566e2f8d3f5056613e34193775d4b236


[ROCm/clr commit: 1cead6a4cd]
 2017-03-31 12:11:34 -05:00
Aditya Atluri bdaf6a3aaa disabled metadata apis 
Change-Id: Ifb8839c581644cccc2afcd18c38a866f649a4144


[ROCm/clr commit: 4e5861df7a]
 2017-03-28 10:46:31 -05:00
Aditya Atluri c3f1a37c28 Re-enabled metadata parsing in HIP 
Change-Id: If8caa844571cb8581450df9ffdb76e2445c75f13


[ROCm/clr commit: abf369049b]
 2017-03-27 11:00:39 -05:00
Aditya Atluri 478eb4e350 removed llvm dependency and metadata functionality 
Change-Id: Ib9783b75d326559ed29c5aa2218aed40d20ad0fb


[ROCm/clr commit: 5a0685c137]
 2017-03-23 10:16:37 -05:00
Aditya Atluri b4b87b8786 Added default module launch api functionality 
1. As in hipModuleLaunchKernel(..., kernelParams, nullptr); works with this commit
    2. Added headers AMDGPUPTNote.h, AMDGPURuntimeMetadata.h to do code object meta data parsing
    3. Changed CMake to look at llvm link libraries
    4. HIP developer should set env variable LLVM_HOME to remove link errors
    5. HIP depends on installed LLVM (not source, not build)
    6. Added sample to test out the feature
    7. Right now HCC does not support embedding metadata in code object. Use clang opencl
    8. Changed HIPCC to read LLVM_HOME env var
    9. New argument to CMake should be given -DLLVM_HOME=<where llvm 5.0 is installed>

Change-Id: Iba38194aa872d97cc2c90a8e5ff746c48055c868


[ROCm/clr commit: 9645b0e0dc]
 2017-03-17 13:11:34 -05:00
Ben Sander 0710dbde55 Fix bug in hipModuleGetFunction. 
Modules with > 1 function didn't return the function correctly.
Also fix coding convention issues


[ROCm/clr commit: 09df0977c0]
 2017-03-08 13:52:38 -06:00
Aditya Atluri 25210492b0 v3: added free for ihipModuleSymbol_t structures inside tracker 
Change-Id: Ib8041a05312c08cbdf2d4fee5e7cbae17df6efff


[ROCm/clr commit: 1cffe49775]
 2017-02-10 13:42:10 -06:00
Aditya Atluri 3577c19d66 v2: Fixed hipModule memory management 
1. Changed test to assert for same hipFunction values
2. Added better memory management for hipModule

Change-Id: I10d7aef13c215a2211e262f3c79017f26a17d9a7


[ROCm/clr commit: 554fe191e1]
 2017-02-10 13:32:13 -06:00
Aditya Atluri 6c76483ac8 fixed hipFunction memory management 
Change-Id: I7ebb323419bcd220ebd6466a8eb38e7bfdb1520a


[ROCm/clr commit: 6110fcdfb6]
 2017-02-09 17:22:55 -06:00
Ben Sander ad2ba66ce6 Read HCC_OPT_FLUSH and optimize dispatch accordingly. 
If HCC is in this mode, we can use less aggressive flushes in some
cases.


[ROCm/clr commit: 2f7a8ec39c]
 2017-01-25 21:50:52 -06:00
Ben Sander 2ea139e30f Show dynamic shared mem usage not static. 
[ROCm/clr commit: 27447f10c2]
 2017-01-23 22:34:41 -06:00
Ben Sander 3bc2e3ba02 Fix debug display for Module launch kernels 
[ROCm/clr commit: 2ffc9f4e22]
 2017-01-19 12:40:45 -06:00
Aditya Atluri a86633f210 changed copyright year from 2016 to 2017 in src directory 
Change-Id: Idb97db509b2b4b1656b2df7a14a62ade38c9d574


[ROCm/clr commit: e9ff23e5f9]
 2017-01-11 18:05:41 -06:00
Ben Sander 8ab567e331 Add name for function 
[ROCm/clr commit: 5d815937de]
 2016-12-17 08:54:09 -06:00
Ben Sander c59c82916a Remove HSA dependency from hipFunction_t 
Place _groupSegmentSize and _privateSegmentSize inside Function,
remove hsa_executable_symbol_t.


[ROCm/clr commit: 2bd70ff345]
 2016-12-17 07:22:56 -06:00
Ben Sander a3064807b5 Remove USE_DISPATCH_HSA_KERNEL=0 path. 
[ROCm/clr commit: 06d382bc6d]
 2016-12-17 07:22:56 -06:00