search

AutoRally / autorally

Software for the AutoRally platform
http://autorally.github.io
738 stars 229 forks source link

cuda related compile error #67

Closed yfzhang closed 5 years ago

yfzhang commented 6 years ago

Basically followed the installations instructions, but got the following error. Some variables in the .cu files seems not to be defined.

[ 96%] Building NVCC (Device) object autorally/autorally_control/src/path_integral/CMakeFiles/path_integral_bf.dir/path_integral_bf_generated_path_integral_main.cu.o
/home/yf/autorally_ws/src/autorally/autorally_control/src/path_integral/path_integral_main.cu(81): error: identifier "DynamicsModel" is undefined

/home/yf/autorally_ws/src/autorally/autorally_control/src/path_integral/path_integral_main.cu(81): error: identifier "MPPI_NUM_ROLLOUTS__" is undefined

/home/yf/autorally_ws/src/autorally/autorally_control/src/path_integral/path_integral_main.cu(81): error: identifier "BLOCKSIZE_X" is undefined

/home/yf/autorally_ws/src/autorally/autorally_control/src/path_integral/path_integral_main.cu(81): error: identifier "BLOCKSIZE_Y" is undefined

/home/yf/autorally_ws/src/autorally/autorally_control/src/path_integral/path_integral_main.cu(97): error: identifier "DynamicsModel" is undefined

/home/yf/autorally_ws/src/autorally/autorally_control/src/path_integral/path_integral_main.cu(97): error: identifier "model" is undefined

/home/yf/autorally_ws/src/autorally/autorally_control/src/path_integral/path_integral_main.cu(97): error: expected a type specifier

/home/yf/autorally_ws/src/autorally/autorally_control/include/autorally_control/path_integral/mppi_controller.cuh(56): error: name followed by "::" must be a class or namespace name
          detected during instantiation of class "autorally_control::MPPIController<DYNAMICS_T, COSTS_T, ROLLOUTS, BDIM_X, BDIM_Y> [with DYNAMICS_T=<error-type>, COSTS_T=autorally_control::MPPICosts, ROLLOUTS=<error-constant>, BDIM_X=<error-constant>, BDIM_Y=<error-constant>]" 
/home/yf/autorally_ws/src/autorally/autorally_control/src/path_integral/path_integral_main.cu(103): here
nolanwagener commented 6 years ago

Try using an older version of cmake, like 3.5 or 3.6.

zobertke commented 5 years ago

It did not work with 3.6 :cry:

Frak8 commented 5 years ago

Hi guys I have a issue during compile "catkin_make" . /usr/lib/gcc/x86_64-linux-gnu/5/include/mwaitxintrin.h(36): error: identifier "__builtin_ia32_monitorx" is undefined I saw same problem in tensor flow GitHub page.Here I have added this flags to "CMakeLists.txt" exactly like this: set( CMAKE_CXX_FLAGS " -D_MWAITXINTRIN_H_INCLUDED -D_FORCE_INLINES ") in top of "CMakeLists.txt" after set(_cmd "catkin_find_pkg" "catkin" "${CMAKE_SOURCE_DIR}") but doesn't effect.

I verified my cuda before compile :

`+-----------------------------------------------------------------------------+ | NVIDIA-SMI 410.79 Driver Version: 410.79 CUDA Version: 10.0 | |-------------------------------+----------------------+----------------------+ | GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC | | Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. | |===============================+======================+======================| | 0 GeForce MX150 Off | 00000000:01:00.0 Off | N/A | | N/A 53C P3 N/A / N/A | 546MiB / 4042MiB | 4% Default | +-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+ | Processes: GPU Memory | | GPU PID Type Process name Usage | |=============================================================================| | 0 1149 G /usr/lib/xorg/Xorg 28MiB | | 0 1544 G /usr/lib/xorg/Xorg 291MiB | | 0 1981 G compiz 78MiB | | 0 2249 G /proc/self/exe 18MiB | | 0 2411 G ...quest-channel-token=5297298698446075686 103MiB | +-----------------------------------------------------------------------------+

and

nvcc: NVIDIA (R) Cuda compiler driver Copyright (c) 2005-2015 NVIDIA Corporation Built on Tue_Aug_11_14:27:32_CDT_2015 Cuda compilation tools, release 7.5, V7.5.17

Please help me Thanks and best regards Amir

Diandry commented 5 years ago

@farahaniamir08 I had the same error message. Make sure your versions of GCC and NVCC are compatible. In my case, using GCC 5.4 Cuda V7.5 is not supported and updating to NVCC 10.1 solved the issue. You might find this useful

keuntaeklee commented 5 years ago

I had the same issue after upgrading something and I went back to older versions:

cmake version 3.5.1 gcc version 6.5.0 nvcc version 9.2

and it compiles without error.