You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Edited CMAKE_CUDA_FLAGS in CMakeLists.txt, removing --use-local-env as that wasn't working on SCALE - set(CMAKE_CUDA_FLAGS "${CMAKE_CUDA_FLAGS} -g -lineinfo -use_fast_math")
Started building cmake .. -DCMAKE_BUILD_TYPE=Release -DCMAKE_CUDA_COMPILER="${SCALE_PATH}/bin/nvcc" -DCMAKE_CUDA_ARCHITECTURES=86 -DCMAKE_INSTALL_RPATH_USE_LINK_PATH=ON -DCMAKE_C_COMPILER=gcc-11 -DCMAKE_CXX_COMPILER=g++-11
Stuck here now cmake --build . --config Release -j8:
[ 27%] Building CUDA object source/EngineGpuKernels/CMakeFiles/EngineGpuKernels.dir/EditKernelsLauncher.cu.o
In file included from /home/redacted/projects/alien/source/EngineGpuKernels/DataAccessKernels.cu:1:
/home/redacted/projects/alien/source/EngineGpuKernels/DataAccessKernels.cuh:4:10: fatal error: 'sm_60_atomic_functions.h' file not found
#include "sm_60_atomic_functions.h"
^~~~~~~~~~~~~~~~~~~~~~~~~~
In file included from /home/redacted/projects/alien/source/EngineGpuKernels/DebugKernels.cu:1:
/home/redacted/projects/alien/source/EngineGpuKernels/DebugKernels.cuh:4:10: fatal error: 'sm_60_atomic_functions.h' file not found
#include "sm_60_atomic_functions.h"
^~~~~~~~~~~~~~~~~~~~~~~~~~
In file included from /home/redacted/projects/alien/source/EngineGpuKernels/EditKernels.cu:1:
/home/redacted/projects/alien/source/EngineGpuKernels/EditKernels.cuh:8:10: fatal error: 'sm_60_atomic_functions.h' file not found
#include "sm_60_atomic_functions.h"
^~~~~~~~~~~~~~~~~~~~~~~~~~
In file included from /home/redacted/projects/alien/source/EngineGpuKernels/DataAccessKernelsLauncher.cu:1:
In file included from /home/redacted/projects/alien/source/EngineGpuKernels/DataAccessKernelsLauncher.cuh:7:
/home/redacted/projects/alien/source/EngineGpuKernels/Base.cuh:6:10: fatal error: 'device_launch_parameters.h' file not found
#include <device_launch_parameters.h>
^~~~~~~~~~~~~~~~~~~~~~~~~~~~
In file included from /home/redacted/projects/alien/source/EngineGpuKernels/FlowFieldKernels.cu:1:
In file included from /home/redacted/projects/alien/source/EngineGpuKernels/FlowFieldKernels.cuh:3:
In file included from /home/redacted/projects/alien/source/EngineGpuKernels/Math.cuh:5:
/home/redacted/projects/alien/source/EngineGpuKernels/Base.cuh:6:10: fatal error: 'device_launch_parameters.h' file not found
#include <device_launch_parameters.h>
^~~~~~~~~~~~~~~~~~~~~~~~~~~~
In file included from /home/redacted/projects/alien/source/EngineGpuKernels/EditKernelsLauncher.cu:1:
In file included from /home/redacted/projects/alien/source/EngineGpuKernels/EditKernelsLauncher.cuh:6:
/home/redacted/projects/alien/source/EngineGpuKernels/Base.cuh:6:10: fatal error: 'device_launch_parameters.h' file not found
#include <device_launch_parameters.h>
^~~~~~~~~~~~~~~~~~~~~~~~~~~~
[ 28%] Building CUDA object source/EngineGpuKernels/CMakeFiles/EngineGpuKernels.dir/GarbageCollectorKernels.cu.o
1 error generated when compiling for gfx1100.
make[2]: *** [source/EngineGpuKernels/CMakeFiles/EngineGpuKernels.dir/build.make:122: source/EngineGpuKernels/CMakeFiles/EngineGpuKernels.dir/DebugKernels.cu.o] Error 1
make[2]: *** Waiting for unfinished jobs....
1 error generated when compiling for gfx1100.
make[2]: *** [source/EngineGpuKernels/CMakeFiles/EngineGpuKernels.dir/build.make:92: source/EngineGpuKernels/CMakeFiles/EngineGpuKernels.dir/DataAccessKernels.cu.o] Error 1
1 error generated when compiling for gfx1100.
1 error generated when compiling for gfx1100.
make[2]: *** [source/EngineGpuKernels/CMakeFiles/EngineGpuKernels.dir/build.make:167: source/EngineGpuKernels/CMakeFiles/EngineGpuKernels.dir/FlowFieldKernels.cu.o] Error 1
make[2]: *** [source/EngineGpuKernels/CMakeFiles/EngineGpuKernels.dir/build.make:137: source/EngineGpuKernels/CMakeFiles/EngineGpuKernels.dir/EditKernels.cu.o] Error 1
1 error generated when compiling for gfx1100.
make[2]: *** [source/EngineGpuKernels/CMakeFiles/EngineGpuKernels.dir/build.make:107: source/EngineGpuKernels/CMakeFiles/EngineGpuKernels.dir/DataAccessKernelsLauncher.cu.o] Error 1
1 error generated when compiling for gfx1100.
make[2]: *** [source/EngineGpuKernels/CMakeFiles/EngineGpuKernels.dir/build.make:152: source/EngineGpuKernels/CMakeFiles/EngineGpuKernels.dir/EditKernelsLauncher.cu.o] Error 1
In file included from /home/redacted/projects/alien/source/EngineGpuKernels/GarbageCollectorKernels.cu:1:
/home/redacted/projects/alien/source/EngineGpuKernels/GarbageCollectorKernels.cuh:4:10: fatal error: 'sm_60_atomic_functions.h' file not found
#include "sm_60_atomic_functions.h"
^~~~~~~~~~~~~~~~~~~~~~~~~~
1 error generated when compiling for gfx1100.
make[2]: *** [source/EngineGpuKernels/CMakeFiles/EngineGpuKernels.dir/build.make:182: source/EngineGpuKernels/CMakeFiles/EngineGpuKernels.dir/GarbageCollectorKernels.cu.o] Error 1
make[1]: *** [CMakeFiles/Makefile2:438: source/EngineGpuKernels/CMakeFiles/EngineGpuKernels.dir/all] Error 2
make[1]: *** Waiting for unfinished jobs....
[ 28%] Linking CXX static library libNetwork.a
[ 28%] Built target Network
make: *** [Makefile:91: all] Error 2
cmake --build . --config Release -j8 67.06s user 2.87s system 428% cpu 16.301 total
I know that something about atomic functions is crucial for the performance of the project, but maybe I can work around this somehow? Or maybe I'm misconfiguring something and it is an easy fix?
The text was updated successfully, but these errors were encountered:
Hey @ashirviskas -- pretty sure this is not a problem for the authors of alien (Cool project btw!) to consider, given that it's a compatibility issue in the SCALE compiler toolkit which we'll fix over there.
Hey @Sheeo, thanks for the response here! To the creator of ALIEN, @chrxh - we're solving these issues one by one in SCALE discord server, might be running on AMD soon 🚀
Hi! I'm very passionate about this project and I tried compiling it on my AMD 7900XTX using SCALE. Here is my current progress on Arch Linux:
CMAKE_CUDA_FLAGS
in CMakeLists.txt, removing--use-local-env
as that wasn't working on SCALE -set(CMAKE_CUDA_FLAGS "${CMAKE_CUDA_FLAGS} -g -lineinfo -use_fast_math")
cmake .. -DCMAKE_BUILD_TYPE=Release -DCMAKE_CUDA_COMPILER="${SCALE_PATH}/bin/nvcc" -DCMAKE_CUDA_ARCHITECTURES=86 -DCMAKE_INSTALL_RPATH_USE_LINK_PATH=ON -DCMAKE_C_COMPILER=gcc-11 -DCMAKE_CXX_COMPILER=g++-11
cmake --build . --config Release -j8
:I know that something about atomic functions is crucial for the performance of the project, but maybe I can work around this somehow? Or maybe I'm misconfiguring something and it is an easy fix?
The text was updated successfully, but these errors were encountered: