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

Jump to bottom

Enable trsm test on NVIDIA GPUs when using DEFAULT TUNING_TARGET #518

Merged
merged 2 commits into from
May 27, 2024
Merged

Enable trsm test on NVIDIA GPUs when using DEFAULT TUNING_TARGET #518

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
e0164a8
Add icpx compilation flag to fix trsm
s-Nick May 21, 2024
416aa78
Fix type for gemm using mixed type(half-float)
s-Nick May 23, 2024
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
1 change: 1 addition & 0 deletions CMakeLists.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -169,6 +169,7 @@ if (INSTALL_HEADER_ONLY)
(${TUNING_TARGET} STREQUAL "INTEL_GPU") )
target_compile_options(portblas INTERFACE -fno-fast-math)
target_compile_options(portblas INTERFACE -mllvm=-loopopt=0)
target_link_options(portblas INTERFACE -mllvm=-loopopt=0)
message(STATUS "Adding -fno-fast-math -mllvm=-loopopt=0 to portblas")
endif()
else()
Expand Down
1 change: 1 addition & 0 deletions cmake/Modules/FindDPCPP.cmake
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -102,6 +102,7 @@ function(add_sycl_to_target)
(${TUNING_TARGET} STREQUAL "INTEL_GPU") )
target_compile_options(${SB_ADD_SYCL_TARGET} PRIVATE -fno-fast-math)
target_compile_options(${SB_ADD_SYCL_TARGET} PRIVATE -mllvm=-loopopt=0)
target_link_options(${SB_ADD_SYCL_TARGET} PRIVATE -mllvm=-loopopt=0)
message(STATUS "Adding -fno-fast-math -mllvm=-loopopt=0 to target ${SB_ADD_SYCL_TARGET}")
endif()
if ((${CMAKE_CXX_COMPILER_ID} STREQUAL "IntelLLVM"
Expand Down
2 changes: 1 addition & 1 deletion src/operations/blas3/gemm_no_local_full_vec.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -381,7 +381,7 @@ class Gemm<input_t, output_t, DoubleBuffer, NbcA, NbcB, ClSize, tile_type,
auto C = orig_C;

/* register array used to store the result*/
value_t reg_res[item_rows * item_cols];
element_t reg_res[item_rows * item_cols];
hjabird marked this conversation as resolved.
Show resolved Hide resolved
scaling_c<need_check_boundary, packet_size>(
reg_res, C, ldc, dim_m_a_start, dim_n_b_start, boundary_check_c,
out_of_range);
Expand Down
10 changes: 0 additions & 10 deletions test/unittest/CMakeLists.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -103,16 +103,6 @@ if(is_dpcpp)
)
endif()

if(is_dpcpp AND ${TUNING_TARGET} STREQUAL "DEFAULT")
if (${DPCPP_SYCL_TARGET} MATCHES "nvidia")
set(TESTS_TO_SKIP
${PORTBLAS_UNITTEST}/blas3/blas3_trsm_test.cpp
)
message(WARNING "Targetting NVIDIA hardware with DEFAULT TUNING_TARGET.
Disabling tests for following operator: trsm.")
endif()
endif()

if(GEMM_TALL_SKINNY_SUPPORT)
list(APPEND SYCL_UNITTEST_SRCS ${PORTBLAS_UNITTEST}/blas3/blas3_gemm_tall_skinny_test.cpp)
endif()
Expand Down
Loading