Commit Graph

26 Commits

Author SHA1 Message Date
Alexey Bataev 0ac29350b5 [OPENMP, NVPTX] Fixes for NVPTX RTL
Summary:
Patch fixes several problems in the implementation of NVPTX RTL.
1. Detection of the last iteration for loops with static scheduling, no chunks.
2. Fixes reductions for the serialized parallel constructs.
3. Fixes handling of the barriers.

Reviewers: grokos

Reviewed By: grokos

Subscribers: Hahnfeld, guansong, openmp-commits

Differential Revision: https://reviews.llvm.org/D48480

llvm-svn: 335469
2018-06-25 13:43:35 +00:00
Guansong Zhang f9e56e5982 [OpenMP] [CUDA] Expose teamid to the debug path
Summary: Small bug fix for debug build. A previous fix causing trouble for debug build.

Reviewers: grokos

Reviewed By: grokos

Subscribers: openmp-commits

Tags: #openmp

Differential Revision: https://reviews.llvm.org/D48286

llvm-svn: 335046
2018-06-19 14:05:38 +00:00
Jonas Hahnfeld 17aabf83e9 [libomptarget-nvptx] loop: Determine if runtime uninitialized
The generic entry points for static loop scheduling previously
hardcoded that the runtime was initialized. This can be wrong if
the compiler analyzes that the runtime is not needed and calls
the init functions accordingly.

This didn't affect clang-ykt because they have entry points for
different combinations of SPMD x Runtime not needed. I didn't do
measurements yet but with inlining we might get away with always
calling the generic interface and letting compiler and runtime
figure out the rest.
In any case, a correct runtime is always better than having
functions that may only be called if previous calls passed in
a specific set of arguments!

Differential Revision: https://reviews.llvm.org/D47131

llvm-svn: 333285
2018-05-25 15:56:48 +00:00
Jonas Hahnfeld 65e0b8784c [CMake] Unify install path for libraries
Introduce OPENMP_INSTALL_LIBDIR and use in all install() commands.
This also fixes installation of libomptarget-nvptx that previously
didn't honor {OPENMP,LLVM}_LIBDIR_SUFFIX.

Differential Revision: https://reviews.llvm.org/D47130

llvm-svn: 333284
2018-05-25 15:56:41 +00:00
George Rokos 6da6f433a0 [CUDA]Fix dynamic|guided scheduling.
The existing implementation of the dynamic scheduling
breaks the contract introduced by the original openmp
runtime and, thus, is incorrect. Patch fixes it and
introduces correct dynamic scheduling model.

Thanks to Alexey Bataev for submitting this patch.

Differential Revision: https://reviews.llvm.org/D47333

llvm-svn: 333225
2018-05-24 21:12:41 +00:00
Jonas Hahnfeld 37bbe1a698 [libomptarget-nvptx] Test bitcode compiler flags and enable by default
Move all logic related to selecting the bitcode compiler and linker
into a new file and dynamically test required compiler flags. This
also adds -fcuda-rdc for Clang trunk as previously attempted in D44992
which fixes the build.

As a result this change also enables building the library by default
if all prerequisites are met.

Differential Revision: https://reviews.llvm.org/D46901

llvm-svn: 332494
2018-05-16 17:20:21 +00:00
Gheorghe-Teodor Bercea 787a350021 [OpenMP][libomptarget] Add function for checking SPMD mode
Summary: Add function to the NVPTX libomptarget library that will return true if the current target region is being executed in SPMD mode.

Reviewers: ABataev, grokos, carlo.bertolli, caomhin

Reviewed By: grokos

Subscribers: guansong, openmp-commits

Differential Revision: https://reviews.llvm.org/D46840

llvm-svn: 332360
2018-05-15 15:16:43 +00:00
Guansong Zhang e1c7a46d5b [OpenMP] Use LIBOMPTARGET_DEVICE_RTL_DEBUG env var to control debug messages on the device side
Summary:
Enable the device side debug messages at compile time, use env var to control at runtime.

To achieve this, an environment data block is passed to the device lib when it is loaded.

By default, the message is off, to enable it, a user need to set LIBOMPDEVICE_DEBUG=1.

Reviewers: grokos

Reviewed By: grokos

Subscribers: openmp-commits

Tags: #openmp

Differential Revision: https://reviews.llvm.org/D46210

llvm-svn: 331550
2018-05-04 19:29:28 +00:00
Guansong Zhang ad6c26516b [OpenMP] Remove compilation warning when using clang to compile bc files.
Summary: Minor printf format correction. NVCC ignore those. Clang will give warning on these if debug is enabled.

Reviewers: grokos

Reviewed By: grokos

Subscribers: openmp-commits

Tags: #openmp

Differential Revision: https://reviews.llvm.org/D45528

llvm-svn: 330944
2018-04-26 14:06:53 +00:00
Guansong Zhang 334c379e32 [OpenMP] Make bc file compilation sensitive to LIBOMPTARGET_NVPTX_DEBUG flag
Summary: The LIBOMPTARGET_NVPTX_DEBUG flag is inconsistent between using nvcc to generate .a file and clang to generate .bc file. Sync the two setting so we can get debug messages from the bc file path as well.

Reviewers: grokos

Subscribers: Hahnfeld, openmp-commits, mgorny

Tags: #openmp

Differential Revision: https://reviews.llvm.org/D45530

llvm-svn: 330477
2018-04-20 20:41:00 +00:00
Guansong Zhang f679431f91 [OpenMP] Remove extra warning when we build
Summary:
This one line change is to remove this warning message

"warning: integer conversion resulted in a change of sign"

Reviewers: grokos

Reviewed By: grokos

Subscribers: openmp-commits

Tags: #openmp

Differential Revision: https://reviews.llvm.org/D45415

llvm-svn: 329713
2018-04-10 15:28:31 +00:00
Guansong Zhang f0029a7738 Revert "[OpenMP] enable bc file compilation using the latest clang"
This reverts commit 6849e31c36d712d97433bca9af39b7a09c8c1207.

llvm-svn: 329576
2018-04-09 14:45:41 +00:00
Guansong Zhang e47fbc9da8 [OpenMP] enable bc file compilation using the latest clang
Summary: adding cuda-rdc flag to allow extern global data

Reviewers: grokos

Reviewed By: grokos

Subscribers: gregrodgers, mgorny, openmp-commits

Tags: #openmp

Differential Revision: https://reviews.llvm.org/D44992

llvm-svn: 329072
2018-04-03 15:01:34 +00:00
Gheorghe-Teodor Bercea 4bc36a06e2 [OpenMP][libomptarget] Initialize global memory stack only once.
Summary: The global stack initialization function may be called multiple times. The initialization of the shared memory slots should only happen when the function is called for the first time for a given warp master thread.

Reviewers: grokos, carlo.bertolli, ABataev, caomhin

Reviewed By: grokos

Subscribers: guansong, openmp-commits

Differential Revision: https://reviews.llvm.org/D44754

llvm-svn: 328148
2018-03-21 21:02:55 +00:00
Gheorghe-Teodor Bercea b4332ca3da [OpenMP][libomptarget] Fix master warp check
Summary: The check for the master warp must take into consideration the actual number of warps: the master warp is equal to the last active warp not necessarily WARPSIZE - 1.

Reviewers: grokos, carlo.bertolli, ABataev, caomhin

Reviewed By: grokos

Subscribers: guansong, openmp-commits

Differential Revision: https://reviews.llvm.org/D44537

llvm-svn: 328146
2018-03-21 20:51:16 +00:00
Gheorghe-Teodor Bercea c8d395a168 [OpenMP][libomptarget] Enable globalization for workers
Summary:
This patch allows worker to have a global memory stack managed by the runtime. This patch is needed for completeness and consistency with the globalization policy: if a worker-side variable escapes the current context it then needs to be globalized.
Until now, only the master thread was allowed to have such a stack. These global values can now potentially be shared amongst workers if the semantics of the OpenMP program require it.

Reviewers: ABataev, grokos, carlo.bertolli, caomhin

Reviewed By: grokos

Subscribers: guansong, openmp-commits

Differential Revision: https://reviews.llvm.org/D44487

llvm-svn: 328144
2018-03-21 20:34:19 +00:00
Gheorghe-Teodor Bercea 876c1ed2e5 [OpenMP][libomptarget] Enable usage of shared memory slots
Summary:
Allow the runtime to use the existing shared memory statically allocated slots.

When a variable is globalized, the underlying memory can be either shared or global memory (both have block-wide visibility). In this case, we allow that the storage to use a limited amount of shared memory that has been statically allocated already. Only if shared memory doesn't prove to be enough do we then invoke malloc() to create a new global memory slot.

Reviewers: ABataev, carlo.bertolli, grokos, caomhin

Reviewed By: grokos

Subscribers: guansong, openmp-commits

Differential Revision: https://reviews.llvm.org/D44486

llvm-svn: 327639
2018-03-15 16:05:34 +00:00
Gheorghe-Teodor Bercea f3de222b0d [OpenMP][libomptarget] Enable multiple frames per global memory slot
Summary: To save on calls to malloc, this patch enables the re-use of pre-allocated global memory slots.

Reviewers: ABataev, grokos, carlo.bertolli, caomhin

Reviewed By: grokos

Subscribers: guansong, openmp-commits

Differential Revision: https://reviews.llvm.org/D44470

llvm-svn: 327637
2018-03-15 15:56:04 +00:00
George Rokos 59be4b434f [libomptarget][nvptx] Bug fix: Correctly identify the warp master active thread.
llvm-svn: 327556
2018-03-14 19:11:36 +00:00
Gheorghe-Teodor Bercea 49b62649cf [OpenMP][libomptarget] Add global memory data sharing support for master-worker sharing.
Summary:
This patch adds support for the sharing of variables from the master thread of a team to the worker threads of the team.
The runtime uses a stack structure implemented as a doubly-linked list of slots with each slot having the exact same size as the size requested. This implementation leverages existing data structures. The runtime functions are added as separate functions to avoid interfering with the current interface. 

Limitations to be addressed in future patches:
- This current patch only employs global memory. In a future patch we will enable to usage for shared memory as an optimization.
- Allow the allocation of several requested sizes in the same slot.

Reviewers: ABataev, grokos, caomhin, carlo.bertolli

Reviewed By: grokos

Subscribers: Hahnfeld, guansong, openmp-commits

Differential Revision: https://reviews.llvm.org/D44260

llvm-svn: 327440
2018-03-13 19:44:53 +00:00
Gheorghe-Teodor Bercea d5e5992f9a [OpenMP][libomptarget] Fix union.
Summary: To make the two parts of the union have the same size, the size of vect needs to be increased by 16 bits.

Reviewers: grokos, carlo.bertolli, caomhin, ABataev

Reviewed By: grokos, ABataev

Subscribers: fedor.sergeev, guansong, openmp-commits

Differential Revision: https://reviews.llvm.org/D44254

llvm-svn: 327040
2018-03-08 18:44:02 +00:00
Gheorghe-Teodor Bercea 7a5fa21ae2 [OpenMP] Remove implicit data sharing using device shared memory from libomptarget
Summary:
This patch reverts the changes to libomptarget that were coupled with the changes to Clang code gen for data sharing using shared memory. A similar patch exists for Clang: D43625

Shared memory is meant to be used as an optimization on top of a more general scheme. So far we didn't have a global memory implementation ready so shared memory was a solution which applied to the current level of OpenMP complexity supported by trunk on GPU devices (due to the missing NVPTX backend patch this functionality has never been exercised). Now that we have a global memory solution this patch is "in the way" and needs to be removed (for now). This patch (or an equivalent version of it) will be put out for review once the global memory scheme is in place.


Reviewers: ABataev, grokos, carlo.bertolli, caomhin

Reviewed By: grokos

Subscribers: Hahnfeld, guansong, openmp-commits

Differential Revision: https://reviews.llvm.org/D43626

llvm-svn: 326950
2018-03-07 22:10:10 +00:00
Gheorghe-Teodor Bercea d5ae4e6501 [OpenMP][libomptarget] Enable the compilation of multiple bc libraries for runtime inlining
Summary:
Different NVIDIA GPUs support different compute capabilities. To enable the inlining of runtime functions and the best performance on different generations of NVIDIA GPUs, a bc library for each compute capability needs to be compiled. The same compiler build will then be usable in conjunction with multiple generations of NVIDIA GPUs.
To differentiate between versions of the same bc lib, the output file name will contain the compute capability ID.
Depends on D14254

Reviewers: Hahnfeld, hfinkel, carlo.bertolli, caomhin, ABataev, grokos

Reviewed By: Hahnfeld, grokos

Subscribers: guansong, mgorny, openmp-commits

Differential Revision: https://reviews.llvm.org/D41724

llvm-svn: 324904
2018-02-12 16:45:20 +00:00
Gheorghe-Teodor Bercea aaeab8d4ef [OpenMP][libomptarget] Add data sharing support in libomptarget
Summary: This patch extends the libomptarget functionality in patch D14254 with support for the data sharing scheme for supporting implicitly shared variables. The runtime therefore maintains a list of references to shared variables.

Reviewers: carlo.bertolli, ABataev, Hahnfeld, grokos, caomhin, hfinkel

Reviewed By: Hahnfeld, grokos

Subscribers: guansong, llvm-commits, openmp-commits

Differential Revision: https://reviews.llvm.org/D41485

llvm-svn: 324495
2018-02-07 18:21:55 +00:00
Carlo Bertolli 57e9f44a8c [OpenMP-RT] Fix debug string for NVPTX runtime library
https://reviews.llvm.org/D42757

The method ThreadsInTeam is used to determine the number of threads to be used in a parallel region under SPMD mode (see line 127 of supporti.h in libomptarget/deviceRTLs/nvptx/src/). This patch fixes the corresponding debug print upon initialization of the kernel in SPMD mode.

llvm-svn: 323978
2018-02-01 16:12:16 +00:00
George Rokos 0dd6ed74fd [OpenMP] Initial implementation of OpenMP offloading library - libomptarget device RTLs.
This patch implements the device runtime library whose interface is used in the code generation for OpenMP offloading devices.
Currently there is a single device RTL written in CUDA meant to CUDA enabled GPUs.
The interface is a variation of the kmpc interface that includes some extra calls to do thread and storage management that only make sense for a GPU target.

Differential revision: https://reviews.llvm.org/D14254

llvm-svn: 323649
2018-01-29 13:59:35 +00:00