Commit Graph

193 Commits

Author SHA1 Message Date
Shilei Tian ca999f7191 [OpenMP][Offloading] Use bitset to indicate execution mode instead of value
The execution mode of a kernel is stored in a global variable, whose value means:
- 0 - SPMD mode
- 1 - indicates generic mode
- 2 - SPMD mode execution with generic mode semantics

We are going to add support for SIMD execution mode. It will be come with another
execution mode, such as SIMD-generic mode. As a result, this value-based indicator
is not flexible.

This patch changes to bitset based solution to encode execution mode. Each
position is:
[0] - generic mode
[1] - SPMD mode
[2] - SIMD mode (will be added later)

In this way, `0x1` is generic mode, `0x2` is SPMD mode, and `0x3` is SPMD mode
execution with generic mode semantics. In the future after we add the support for
SIMD mode, `0b1xx` will be in SIMD mode.

Reviewed By: jdoerfert

Differential Revision: https://reviews.llvm.org/D110029
2021-09-22 11:40:52 -04:00
Shilei Tian 49e976c934 [OpenMP][NVPTX] Fix a warning that data argument not used by format string
Reviewed By: jhuber6, grokos

Differential Revision: https://reviews.llvm.org/D110104
2021-09-20 17:22:14 -04:00
Joseph Huber f1c821fa85 [OpenMP] Add support for dynamic shared memory in new RTL
This patch adds support for using dynamic shared memory in the new
device runtime. The new function `__kmpc_get_dynamic_shared` will return a
pointer to the buffer of dynamic shared memory. Currently the amount of memory
allocated is set by an environment variable.

In the future this amount will be added to the amount used for the smart stack
which will be configured in a similar way.

Reviewed By: tianshilei1992

Differential Revision: https://reviews.llvm.org/D110006
2021-09-17 21:25:36 -04:00
Joseph Huber ec02c34b6d [OpenMP] Add additional fields to device environment
This patch adds fields for the device number and number of devices into
the device environment struct and debugging values.

Reviewed By: jdoerfert

Differential Revision: https://reviews.llvm.org/D110004
2021-09-17 21:25:32 -04:00
Hansang Bae ae2a5facce [OpenMP][libomptarget] Minor fix in x86_64 plugin
Call to remove() was passing invalid address for the file name.

Differential Revision: https://reviews.llvm.org/D109846
2021-09-15 15:57:06 -05:00
Jon Chesterfield 6760234e8d [libomptarget][amdgpu] Precisely manage hsa lifetime
The hsa library must be initialized before any calls into it and
destructed after the last call into it. There have been a number of bugs in
this area related to member variables which would like to use raii to manage
resources acquired from hsa.

This patch moves the init/shutdown of hsa into a class, such that when used as
the first member variable (could be a base), the lifetime of other member
variables are reliably scoped within it. This will allow other classes to use
raii reliably when used as member variables within the global.

Reviewed By: pdhaliwal

Differential Revision: https://reviews.llvm.org/D109512
2021-09-09 17:28:11 +01:00
Jon Chesterfield d642156f8f [libomptarget][nfc] Hoist hsa_init into rtl.cpp 2021-09-09 16:09:34 +01:00
Jon Chesterfield 3153bdd547 [libomptarget][amdgpu] Drop env variables
Use the same debug print as the rest of libomptarget plugins with
the same environment control. Also drop the max queue size debugging hook as
I don't believe it is still in use, can bring it back near the rest of the env
handling in rtl.cpp if someone objects.

That makes most of rt.h and all of utils.cpp unused. Clean that up and simplify
control flow in a couple of places.

Behaviour change is that debug prints that used to use the old environment
variable now use the new one and print in slightly different format, and the
removal of the max queue size variable.

Reviewed By: pdhaliwal

Differential Revision: https://reviews.llvm.org/D108784
2021-09-02 11:02:39 +01:00
Jon Chesterfield f8bcbb82a7 [libomptarget] Normalise a cmake debug string, checking it triggers CI 2021-09-01 14:24:28 +01:00
Shilei Tian e8fdacfd81 [OpenMP][NVPTX] Fixed missing variables for CUDA free compilation in NVPTX plugin
`CU_EVENT_DEFAULT` is defined in CUDA header. It should be added to
`openmp/libomptarget/plugins/cuda/dynamic_cuda/cuda.h` for CUDA free build.

Reviewed By: ronlieb

Differential Revision: https://reviews.llvm.org/D108878
2021-08-28 18:08:10 -04:00
Shilei Tian 29df4ab3f3 [OpenMP][Offloading] Add support for event related interfaces
This patch adds the support form event related interfaces, which will be used
later to fix data race. See D104418 for more details.

Reviewed By: jdoerfert, ye-luo

Differential Revision: https://reviews.llvm.org/D108528
2021-08-28 16:24:14 -04:00
Jon Chesterfield 78f92c3810 [openmp][amdgpu] Initial gfx10 offloading implementation
Lets wavefront size be 32 for amdgpu openmp, as well as 64.

Fixes up as little as possible to pass that through the libraries. This change
is end to end, as opposed to updating clang/devicertl/plugin separately. It can
be broken up for review/commit if preferred. Posting as-is so that others with
a gfx10 can try it out. It works roughly as well as gfx9 for me, but there are
probably bugs remaining as well as the todo: for letting grid values vary more.

Reviewed By: ronlieb

Differential Revision: https://reviews.llvm.org/D108708
2021-08-27 12:34:03 +01:00
Jon Chesterfield 3d85342982 [libomptarget][amdgpu][nfc] Rename variables, delete dead code 2021-08-26 19:58:38 +01:00
Jon Chesterfield 68ab93f4d7 [libomptarget][amdgpu][nfc] Rename source files 2021-08-26 18:29:44 +01:00
Jon Chesterfield ba0af885e7 [libomptarget][amdgpu][nfc] Make grid value access match devicertl 2021-08-25 15:11:19 +01:00
Jon Chesterfield 9b2c6c07b5 [libomptarget][amdgpu] Refactor debug printing
Move most debug printing in rtl.cpp behind DP() macro
Adjust the print output for gpu arch mismatch when the architectures match
Convert an assert into graceful failure

Reviewed By: jdoerfert

Differential Revision: https://reviews.llvm.org/D108562
2021-08-25 14:57:51 +01:00
Jon Chesterfield ba8547775b [libomptarget][amdgpu] Fix debug build from D104696 2021-08-25 01:27:51 +01:00
Pushpinder Singh 9b8b7c1180 [AMDGPU][Libomptarget] Delete g_atl_machine global
With uses of g_atl_machine gone, a significant portion of dead
code has been removed.

This patch depends on D104691 and D104695.

Reviewed By: JonChesterfield

Differential Revision: https://reviews.llvm.org/D104696
2021-08-24 07:59:40 +00:00
Jon Chesterfield 77579b99e9 [openmp][nfc] Replace OMPGridValues array with struct
[nfc] Replaces enum indices into an array with a struct. Named the
fields to match the enum, leaves memory layout and initialization unchanged.

Motivation is to later safely remove dead fields and replace redundant ones
with (compile time) computation. It should also be possible to factor some
common fields into a base and introduce a gfx10 amdgpu instance with less
duplication than the arrays of integers require.

Reviewed By: ronlieb

Differential Revision: https://reviews.llvm.org/D108339
2021-08-19 13:25:42 +01:00
Joseph Huber edb8acdc6e [Libomptarget] Correctly default to Generic if exec_mode is not present
Currently, the runtime returns an error when the `exec_mode` global is
not present. The expected behvaiour is that the region will default to
Generic. This prevents global constructors from being called because
they do not contain execution mode globals.

Reviewed By: jdoerfert

Differential Revision: https://reviews.llvm.org/D108255
2021-08-18 11:24:28 -04:00
Dimitry Andric 400cd6d2f0 [libomptarget][amdgpu] use --allow-shlib-undefined to link on FreeBSD
On FreeBSD, the `environ` symbol is undefined at link time for shared
libraries, but resolved by the dynamic linker at runtime. Therefore,
allow the symbol to be undefined when creating a shared library, by
using the `--allow-shlib-undefined` linker flag, instead of `-z defs`
(a.k.a `--no-undefined`).

Reviewed By: JonChesterfield

Differential Revision: https://reviews.llvm.org/D107698
2021-08-08 13:52:44 +02:00
Dimitry Andric 71ae2e0221 [libomptarget][amdgpu] don't declare Elf_Note on FreeBSD
On FreeBSD, the system `<libelf.h>` already declares `struct Elf_Note`
indirectly (via `<sys/elf_common.h>`). This results in compile errors
when building the libomptarget amdgpu plugin. Avoid redeclaring `struct
Elf_Note` on FreeBSD to fix the errors.

Reviewed By: JonChesterfield

Differential Revision: https://reviews.llvm.org/D107661
2021-08-06 21:45:26 +02:00
Shilei Tian 28939b6ae5 [NFC] Clean up and clang-format openmp/libomptarget/plugins/cuda/src/rtl.cpp 2021-08-05 22:32:28 -04:00
Jon Chesterfield a90da62adb [libomptarget][amdgpu] Update printed plugin name 2021-07-29 14:46:42 +01:00
Jose M Monsalve Diaz 88e66fa60a [OpenMP] Fixing missing variables when CUDA SDK not in system
This patch fixes the error reported in D106751. When there is no CUDA SDK
installed in the system, the build fails due to missing `CU_DEVICE_ATTRIBUTE`
variables.

Using @zsrkmyn sugested fix

Reviewed By: jdoerfert

Differential Revision: https://reviews.llvm.org/D106933
2021-07-27 23:46:15 -05:00
Jose M Monsalve Diaz 313c523995 [OpenMP][Tool] Introducing the `llvm-omp-device-info` tool
This patch introduces the `llvm-omp-device-info` tool, which uses the
omptarget library and interface to query the device info from all the
available devices as seen by OpenMP. This is inspired by PGI's `pgaccelinfo`

Since omptarget usually requires a description structure with executable
kernels, I split the initialization of the RTLs and Devices to be able to
initialize all possible devices and query each of them.

This revision relies on the patch that introduces the print device info.

A limitation is that the order in which the devices are initialized, and the
corresponding device ID is not necesarily the one seen by OpenMP.

The changes are as follows:
1. Separate the RTL initialization that was performed in `RegisterLib` to its own `initRTLonce` function
2. Create an `initAllRTLs` method that initializes all available RTLs at runtime
3. Created the `llvm-deviceinfo.cpp` tool that uses `omptarget` to query each device and prints its information.

Example Output:
```
Device (0):
    print_device_info not implemented

Device (1):
    print_device_info not implemented

Device (2):
    print_device_info not implemented

Device (3):
    print_device_info not implemented

Device (4):
    CUDA Driver Version:                11000
    CUDA Device Number:                 0
    Device Name:                        Quadro P1000
    Global Memory Size:                 4236312576 bytes
    Number of Multiprocessors:          5
    Concurrent Copy and Execution:      Yes
    Total Constant Memory:              65536 bytes
    Max Shared Memory per Block:        49152 bytes
    Registers per Block:                65536
    Warp Size:                          32 Threads
    Maximum Threads per Block:          1024
    Maximum Block Dimensions:           1024, 1024, 64
    Maximum Grid Dimensions:            2147483647 x 65535 x 65535
    Maximum Memory Pitch:               2147483647 bytes
    Texture Alignment:                  512 bytes
    Clock Rate:                         1480500 kHz
    Execution Timeout:                  Yes
    Integrated Device:                  No
    Can Map Host Memory:                Yes
    Compute Mode:                       DEFAULT
    Concurrent Kernels:                 Yes
    ECC Enabled:                        No
    Memory Clock Rate:                  2505000 kHz
    Memory Bus Width:                   128 bits
    L2 Cache Size:                      1048576 bytes
    Max Threads Per SMP:                2048
    Async Engines:                      Yes (2)
    Unified Addressing:                 Yes
    Managed Memory:                     Yes
    Concurrent Managed Memory:          Yes
    Preemption Supported:               Yes
    Cooperative Launch:                 Yes
    Multi-Device Boars:                 No
    Compute Capabilities:               61
```

Reviewed By: tianshilei1992

Differential Revision: https://reviews.llvm.org/D106752
2021-07-27 22:38:35 -04:00
Jose M Monsalve Diaz d2f85d0910 [OpenMP][Libomptarget] Adding `print_device_info` to RTL and `omptarget`
This patch introduces a function in the device's plugin to print the
device information. This patch relates to another patch that introduces
a CLI tool to obtain the device information from the omplibrary directly.
It is inspired by PGI's pgaccelinfo.

The modifications are as follows:
1. Introduce the optional `void __tgt_rtl_print_device_info(RTLdevID)` function into the RTL.
2. Introduce the `bool __tgt_print_device_info(devID)` function into `omptarget` interface. Returns false if the RTL is not implemented
3. Added `bool printDeviceInfo(RTLDevID)` to the `DeviceTy`
4. Implement the `__tgt_rtl_print_device_info` for CUDA. Added additional CUDA Runtime calls.

Reviewed By: jdoerfert

Differential Revision: https://reviews.llvm.org/D106751
2021-07-27 21:47:57 -04:00
Jon Chesterfield 2a613a7790 [libomptarget] Build amdgpu plugin without hsa
Default to building the amdgpu plugin to use dlopen when hsa is
not found instead of disabling it.

Reviewed By: jdoerfert

Differential Revision: https://reviews.llvm.org/D106600
2021-07-26 09:54:51 +01:00
Jon Chesterfield dd0b463dd9 [libomptarget][amdgpu] More robust handling of failure to init HSA
If hsa_init fails, subsequent calls into hsa are not safe. Except for
hsa_init, but we don't retry on failure.

This patch:
- deletes a print that called into hsa to ask why it can't call into hsa
- drops a merge conflict block next to that print
- reliably initializes number of devices to zero
- skips the plugin destructor contents if the constructor failed to init hsa

Tested by making hsa_init return error, and by forcing the dynamic library
use which was then deleted from disk. Before this patch, both segv. After it,
friendly message about offloading being unavailable.

Reviewed By: jdoerfert

Differential Revision: https://reviews.llvm.org/D106774
2021-07-25 23:15:58 +01:00
Jon Chesterfield e3251f2ec4 Revert "[libomptarget] Build amdgpu plugin without hsa"
Inaccurate error handling around hsa_init

This reverts commit e30b3b23a4.
2021-07-25 21:03:51 +01:00
Jon Chesterfield e30b3b23a4 [libomptarget] Build amdgpu plugin without hsa
Default to building the amdgpu plugin to use dlopen when hsa is
not found instead of disabling it.

Reviewed By: jdoerfert

Differential Revision: https://reviews.llvm.org/D106600
2021-07-25 19:33:36 +01:00
Abhinav Gaba f7c92995c0 [OpenMP] Fix CUDA plugin build after 3817ba13ae.
The build was broken on machines that don't have Cuda SDK installed.

See https://reviews.llvm.org/D106627 for the original discussion.
2021-07-23 16:50:00 +08:00
Joseph Huber 76c0c0ca86 [OpenMP][NFC] Fix formatting in CUDA plugin 2021-07-22 21:50:40 -04:00
Joseph Huber 3817ba13ae [OpenMP] Add environment variables to change stack / heap size in the CUDA plugin
This patch adds support for two environment variables to configure the device.
``LIBOMPTARGET_STACK_SIZE`` sets the amount of memory in bytes that each thread
has for its stack. ``LIBOMPTARGET_HEAP_SIZE`` sets the amount of heap memory
that can be allocated using malloc / free on the device.

Reviewed By: jdoerfert

Differential Revision: https://reviews.llvm.org/D106627
2021-07-22 21:40:02 -04:00
Jon Chesterfield 9e05c084e5 [libomptarget][amdgpu][nfc] Normalise license headers
Reviewed By: gregrodgers, jdoerfert

Differential Revision: https://reviews.llvm.org/D106581
2021-07-22 20:23:41 +01:00
Jon Chesterfield 14e34a83b0 [libomptarget][amdgpu][nfc] Replace use of gelf.h with libelf.h
AMDGPU can assume Elf64 so doesn't need to abstract over Elf32

Drop a few other unused headers at the same time. Now only llvm elf
and libelf are used by the plugin.

Reviewed By: jdoerfert

Differential Revision: https://reviews.llvm.org/D106579
2021-07-22 20:04:13 +01:00
Jon Chesterfield 1a96570621 [libomptarget][amdgpu] Implement dlopen of libhsa
AMDGPU plugin equivalent of D95155, build without HSA installed locally

Compiles a new file, plugins/amdgpu/dynamic_hsa/hsa.cpp, to an object file that
exposes the same symbols that the plugin presently uses from hsa. The object
file contains dlopen of hsa and cached dlsym calls. Also provides header files
corresponding to the subset that is used.

This is behind a feature flag, LIBOMPTARGET_FORCE_DLOPEN_LIBHSA, default off.
That allows developers to build against the dlopen/dlsym implementation, e.g.
while testing this mode.

Enabling by default will cause this plugin to build on a wider variety of
machines than it does at present so may break some CI builds. That risk can
be minimised by reviewing the header dependencies of the library and ensuring
it doesn't use any libraries that are not already used by libomptarget.

Separating the implementation from enabling by default in case the latter needs
to be rolled back after wider CI results.

Reviewed By: jdoerfert

Differential Revision: https://reviews.llvm.org/D106559
2021-07-22 16:54:10 +01:00
Joseph Huber a158d3663f [OpenMP] Fix warnings for uninitialized block counts
Summary:
Fixes some warning given for uninitialized block counts if the exection mode is
not recognized. This shouldn't happen in practice because the execution mode is
checked when it's read from the device.
2021-07-22 09:24:07 -04:00
Jon Chesterfield dc1f6f8b92 [libomptarget][amdgpu][nfc] Drop dead signal pool setup
This class is instantiated once in rtl.cpp before hsa_init is
called. The hsa_signal_create call therefore fails leaving the pool empty.

This signal pool is a legacy from ATMI where it was constructed after hsa_init.
Moving the state into the rtl.cpp global class disabled the initial populating
of the pool without noticeably changing performance. Just rechecked with a fix
that allocates the signals after hsa_init and that also doesn't noticeably
change performance.

This patch therefore drops the initialisation. Only change from main is to
drop a DEBUG_PRINT statement that would say the pool initial size is zero.

Reviewed By: jdoerfert

Differential Revision: https://reviews.llvm.org/D106515
2021-07-22 10:29:32 +01:00
Joseph Huber 7d57639264 [OpenMP] Add new execution mode for SPMD execution with Generic semantics
Qualified kernels can be transformed from generic-mode to SPMD mode using an
optimization in OpenMPOpt. This patch introduces a new execution mode to
indicate kernels that have been transformed from generic-mode to SPMD-mode.
These kernels have SPMD-mode execution, but need generic-mode semantics for
scheduling the blocks and threads. Without this far too few blocks will be
scheduled for a generic region as SPMD mode expects the trip count to be
divided by the number of threads.

Reviewed By: ggeorgakoudis

Differential Revision: https://reviews.llvm.org/D106460
2021-07-21 20:57:28 -04:00
Jon Chesterfield a733bbbd17 [libomptarget][amdgpu][nfc] Refactor #includes
Create a hsa_api.h header that includes the ROCr headers in use
Drop some unused headers and _cplusplus macros

Reviewed By: jdoerfert

Differential Revision: https://reviews.llvm.org/D106455
2021-07-21 17:28:07 +01:00
Jon Chesterfield ddfb074a80 [libomptarget][nfc] Group environment variables, drop accesses to DeviceInfo global
[libomptarget][nfc] Group environment variables, drop accesses to DeviceInfo global

Folds some duplicates logic into a helper function, passes the new environment
struct into getLaunchVals which no longer reads the DeviceInfo global.

Implemented on top of D105237

Reviewed By: dhruvachak

Differential Revision: https://reviews.llvm.org/D105239
2021-07-06 17:06:38 +01:00
Atmn Patel 21e92612c0 [Libomptarget] Experimental Remote Plugin Fixes
D97883 introduced a compile-time error in the experimental remote offloading
libomptarget plugin, this patch fixes it and resolves a number of
inconsistencies in the plugin as well:

1. Non-functional Asynchronous API
2. Unnecessarily verbose debug printing
3. Misc. code clean ups

This is not intended to make any functional changes to the plugin.

Differential Revision: https://reviews.llvm.org/D105325
2021-07-02 12:38:34 -04:00
Jon Chesterfield db89414da4 [libomptarget][nfc] Move grid size computation
Change getLaunchVals to return the integers used for launch

Reviewed By: pdhaliwal

Differential Revision: https://reviews.llvm.org/D105237
2021-07-01 12:53:04 +01:00
Dhruva Chakrabarti 98c36f0079 Revert "[libomptarget] [amdgpu] Fix default setting of max flat workgroup size"
This reverts commit 2240b41ee4.
A value of 0 for KernDescVal WG_Size implies it is unknown, so it should be
set to the default. The above change was made without this assumption.

Reviewed By: JonChesterfield

Differential Revision: https://reviews.llvm.org/D105250
2021-06-30 17:15:00 -07:00
Jon Chesterfield 4b0926b044 [libomptarget][nfc] Replace out arguments with struct return
A step towards making this function adequately self contained that it
can be tested easily. No functional change intended here, left variable
names unchanged.

Reviewed By: ronlieb

Differential Revision: https://reviews.llvm.org/D105229
2021-06-30 22:40:07 +01:00
Jon Chesterfield d86b0073cf [libomptarget][amdgpu][nfc] Fix build warnings, drop some headers
Removes stdarg header, drops uses of iostream, fix some format string errors.
Also changes a C style struct to C++ style to avoid a warning from clang/

Reviewed By: pdhaliwal

Differential Revision: https://reviews.llvm.org/D104923
2021-06-30 22:23:36 +01:00
Dhruva Chakrabarti e0b713a035 [libomptarget] [amdgpu] Change default number of teams per computation unit
This patch is related to https://reviews.llvm.org/D98832. Based on discussions there, I decided to separate out the teams default as this patch. This change is to increase the number of teams per computation unit so as to provide more wavefronts for hiding latency. This change improves performance for some programs, including 20-50% for some Stream benchmarks.

Reviewed By: JonChesterfield

Differential Revision: https://reviews.llvm.org/D99003
2021-06-29 15:34:35 -07:00
Dhruva Chakrabarti 2240b41ee4 [libomptarget] [amdgpu] Fix default setting of max flat workgroup size
When max flat workgroup size is not specified, it is set to the default
workgroup size. This prevents kernel launch with a workgroup size larger
than the default. The fix is to ignore a size of 0 and treat it as
unspecified.

Reviewed By: JonChesterfield

Differential Revision: https://reviews.llvm.org/D105073
2021-06-29 13:47:24 -07:00
Pushpinder Singh 20df2c7052 [AMDGPU][Libomptarget] Collect allocatable memory pools using HSA
The logic is almost similar to that of system.cpp with one change that
instead of adding all the memory pools to a device struct it only
keeps a single pool. The existing approach also always allocated memory on
the first HSA pool found for a GPU.

This depends on D104691. The goal of this series of patches is to remove
_atl_machine global. The next patch will drop g_atl_machine entirely.

Reviewed By: JonChesterfield

Differential Revision: https://reviews.llvm.org/D104695
2021-06-28 11:28:04 +00:00