Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Update for ROCm 6.1.0 #3898

Merged
merged 1 commit into from
Apr 25, 2024
Merged

Update for ROCm 6.1.0 #3898

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
24 changes: 2 additions & 22 deletions .github/workflows/hip.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -24,21 +24,11 @@ jobs:
restore-keys: |
ccache-${{ github.workflow }}-${{ github.job }}-git-
- name: Build & Install
# Have to have -Wno-deprecated-declarations due to deprecated atomicAddNoRet
# Have to have -Wno-gnu-zero-variadic-macro-arguments to avoid
# amrex/Src/Base/AMReX_GpuLaunchGlobal.H:15:5: error: must specify at least one argument for '...' parameter of variadic macro [-Werror,-Wgnu-zero-variadic-macro-arguments]
# __launch_bounds__(amrex_launch_bounds_max_threads)
# ^
# /opt/rocm-4.1.1/hip/include/hip/hcc_detail/hip_runtime.h:178:71: note: expanded from macro '__launch_bounds__'
# select_impl_(__VA_ARGS__, launch_bounds_impl1, launch_bounds_impl0)(__VA_ARGS__)
# ^
# /opt/rocm-4.1.1/hip/include/hip/hcc_detail/hip_runtime.h:176:9: note: macro 'select_impl_' defined here
# #define select_impl_(_1, _2, impl_, ...) impl_
# Have to remove "-fno-operator-names to avoid
# /opt/rocm-6.1.0/include/rocprim/device/detail/device_adjacent_difference.hpp:198:26: error: token is not a valid binary operator in a preprocessor subexpression
# 198 | #if defined(__gfx1102__) or defined(__gfx1030__)
# | ~~~~~~~~~~~~~~~~~~~~~^~
env: {CXXFLAGS: "-Werror -Wall -Wextra -Wpedantic -Wnull-dereference -Wfloat-conversion -Wshadow -Woverloaded-virtual -Wextra-semi -Wunreachable-code -Wnon-virtual-dtor -Wno-deprecated-declarations -Wno-gnu-zero-variadic-macro-arguments"}
env: {CXXFLAGS: "-Werror -Wall -Wextra -Wpedantic -Wnull-dereference -Wfloat-conversion -Wshadow -Woverloaded-virtual -Wextra-semi -Wunreachable-code -Wnon-virtual-dtor"}
run: |
export CCACHE_COMPRESS=1
export CCACHE_COMPRESSLEVEL=10
Expand Down Expand Up @@ -92,21 +82,11 @@ jobs:
restore-keys: |
ccache-${{ github.workflow }}-${{ github.job }}-git-
- name: Build & Install
# Have to have -Wno-deprecated-declarations due to deprecated atomicAddNoRet
# Have to have -Wno-gnu-zero-variadic-macro-arguments to avoid
# amrex/Src/Base/AMReX_GpuLaunchGlobal.H:15:5: error: must specify at least one argument for '...' parameter of variadic macro [-Werror,-Wgnu-zero-variadic-macro-arguments]
# __launch_bounds__(amrex_launch_bounds_max_threads)
# ^
# /opt/rocm-4.1.1/hip/include/hip/hcc_detail/hip_runtime.h:178:71: note: expanded from macro '__launch_bounds__'
# select_impl_(__VA_ARGS__, launch_bounds_impl1, launch_bounds_impl0)(__VA_ARGS__)
# ^
# /opt/rocm-4.1.1/hip/include/hip/hcc_detail/hip_runtime.h:176:9: note: macro 'select_impl_' defined here
# #define select_impl_(_1, _2, impl_, ...) impl_
# Have to remove "-fno-operator-names to avoid
# /opt/rocm-6.1.0/include/rocprim/device/detail/device_adjacent_difference.hpp:198:26: error: token is not a valid binary operator in a preprocessor subexpression
# 198 | #if defined(__gfx1102__) or defined(__gfx1030__)
# | ~~~~~~~~~~~~~~~~~~~~~^~
env: {CXXFLAGS: "-Werror -Wall -Wextra -Wpedantic -Wnull-dereference -Wfloat-conversion -Wshadow -Woverloaded-virtual -Wextra-semi -Wunreachable-code -Wnon-virtual-dtor -Wno-deprecated-declarations -Wno-gnu-zero-variadic-macro-arguments"}
env: {CXXFLAGS: "-Werror -Wall -Wextra -Wpedantic -Wnull-dereference -Wfloat-conversion -Wshadow -Woverloaded-virtual -Wextra-semi -Wunreachable-code -Wnon-virtual-dtor"}
run: |
export CCACHE_COMPRESS=1
export CCACHE_COMPRESSLEVEL=10
Expand Down
20 changes: 20 additions & 0 deletions Src/Base/AMReX_Scan.H
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -641,13 +641,33 @@ T PrefixSum (N n, FIN const& fin, FOUT const& fout, TYPE, RetSum a_ret_sum = ret
using ScanTileState = rocprim::detail::lookback_scan_state<T>;
using OrderedBlockId = rocprim::detail::ordered_block_id<unsigned int>;

#if (defined(HIP_VERSION_MAJOR) && (HIP_VERSION_MAJOR < 6)) || \
(defined(HIP_VERSION_MAJOR) && (HIP_VERSION_MAJOR == 6) && \
defined(HIP_VERSION_MINOR) && (HIP_VERSION_MINOR == 0))

std::size_t nbytes_tile_state = rocprim::detail::align_size
(ScanTileState::get_storage_size(nblocks));
std::size_t nbytes_block_id = OrderedBlockId::get_storage_size();

auto dp = (char*)(The_Arena()->alloc(nbytes_tile_state+nbytes_block_id));

ScanTileState tile_state = ScanTileState::create(dp, nblocks);

#else

std::size_t nbytes_tile_state;
AMREX_HIP_SAFE_CALL(ScanTileState::get_storage_size(nblocks, stream, nbytes_tile_state));
nbytes_tile_state = rocprim::detail::align_size(nbytes_tile_state);

std::size_t nbytes_block_id = OrderedBlockId::get_storage_size();

auto dp = (char*)(The_Arena()->alloc(nbytes_tile_state+nbytes_block_id));

ScanTileState tile_state;
AMREX_HIP_SAFE_CALL(ScanTileState::create(tile_state, dp, nblocks, stream));

#endif

auto ordered_block_id = OrderedBlockId::create
(reinterpret_cast<OrderedBlockId::id_type*>(dp + nbytes_tile_state));

Expand Down
Loading