Skip to content

fix: Respect native Orin CUDA architecture and CuTe link requirements #118

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

Open
suharvest wants to merge 2 commits into
base: main
Choose a base branch
from
+19 −3
Open

fix: Respect native Orin CUDA architecture and CuTe link requirements #118

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
68f2409
fix: Respect explicit CUDA architectures
suharvest Jul 25, 2026
b406116
fix: Propagate CuTe DSL shim through static libraries
suharvest Jul 25, 2026
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
12 changes: 11 additions & 1 deletion CMakeLists.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15,6 +15,16 @@

cmake_minimum_required(VERSION 3.20 FATAL_ERROR)

# Record whether the user explicitly requested CUDA architectures before
# project() runs. enable_language(CUDA) initializes CMAKE_CUDA_ARCHITECTURES to
# a compiler default, so testing it afterwards cannot distinguish a
# user-provided value from that default.
set(_EDGELLM_USER_CUDA_ARCHS FALSE)
if(DEFINED CMAKE_CUDA_ARCHITECTURES
AND NOT CMAKE_CUDA_ARCHITECTURES STREQUAL "")
set(_EDGELLM_USER_CUDA_ARCHS TRUE)
endif()

project(
tensorrt_edgellm_sdk
VERSION 0.9.1
Expand Down Expand Up @@ -61,7 +71,7 @@ endif()
set_ifndef(CUDA_CTK_VERSION 12.8)
set_ifndef(CUDA_DIR /usr/local/cuda-${CUDA_CTK_VERSION})

if(NOT DEFINED AARCH64_BUILD)
if(NOT _EDGELLM_USER_CUDA_ARCHS AND NOT DEFINED AARCH64_BUILD)
set(CMAKE_CUDA_ARCHITECTURES 80;86;89)
if(CUDA_CTK_VERSION VERSION_GREATER_EQUAL 12.8)
list(APPEND CMAKE_CUDA_ARCHITECTURES 100a 120)
Expand Down
10 changes: 8 additions & 2 deletions cmake/CuteDsl.cmake
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1043,7 +1043,8 @@ function(cute_dsl_setup)
foreach(_tgt ${ARG_LINK_TARGETS})
get_target_property(_tgt_type ${_tgt} TYPE)
if(_tgt_type STREQUAL "STATIC_LIBRARY")
target_link_libraries(${_tgt} PUBLIC ${_link_libs})
target_link_libraries(${_tgt} PUBLIC ${_link_libs}
trt_edgellm_cutedsl_cudart_shim)
else()
target_link_libraries(${_tgt} PRIVATE ${_link_libs}
trt_edgellm_cutedsl_cudart_shim)
Expand All @@ -1056,7 +1057,12 @@ function(cute_dsl_setup)
if(NOT _cute_dsl_cuda_ver STREQUAL ""
AND _cute_dsl_cuda_ver VERSION_GREATER_EQUAL 12.0
AND _cute_dsl_cuda_ver VERSION_LESS 12.8)
target_link_options(${_tgt} PRIVATE "-Wl,--wrap=_cudaLaunchKernelEx")
if(_tgt_type STREQUAL "STATIC_LIBRARY")
target_link_options(${_tgt} INTERFACE
"-Wl,--wrap=_cudaLaunchKernelEx")
else()
target_link_options(${_tgt} PRIVATE "-Wl,--wrap=_cudaLaunchKernelEx")
endif()
endif()
endforeach()

Expand Down