-
-
Notifications
You must be signed in to change notification settings - Fork 2.7k
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.
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
web.lua cuda 10 turbo error #307
Comments
I guess that turbo.lua's architecture detection is not working on NVIDIA Jetson Nano(cmds is null). It is a platform support issue of turbo.lua. https://github.com/kernelsauce/turbo |
You can check the value of ffi.arch with the following command. % th -e "ffi = require 'ffi'; print(ffi.arch)"
x64 |
Thanks for the tips. The following command shows:
and arm64 indeed does not show up in the syscall.lua file. Adding an Since this looks more like an issue with Thanks! |
On x64 Linux, syscall ids can be show with the following program. (If the syscall name does not exist, a compile error will occur.) #include <sys/syscall.h>
#include <stdio.h>
int main(void)
{
printf("SYS_stat = %d,\n", SYS_stat);
printf("SYS_fstat = %d,\n", SYS_fstat);
printf("SYS_lstat = %d,\n", SYS_lstat);
printf("SYS_getdents = %d,\n", SYS_getdents);
printf("SYS_io_setup = %d,\n", SYS_io_setup);
printf("SYS_io_destroy = %d,\n", SYS_io_destroy);
printf("SYS_io_getevents = %d,\n", SYS_io_getevents);
printf("SYS_io_submit = %d,\n", SYS_io_submit);
printf("SYS_io_cancel = %d,\n", SYS_io_cancel);
printf("SYS_clock_settime = %d,\n", SYS_clock_settime);
printf("SYS_clock_gettime = %d,\n", SYS_clock_gettime);
printf("SYS_clock_getres = %d,\n", SYS_clock_getres);
printf("SYS_clock_nanosleep = %d,\n", SYS_clock_nanosleep);
return 0;
} % gcc syscall_list.c -o syscall_list
% ./syscall_list
SYS_stat = 4,
SYS_fstat = 5,
SYS_lstat = 6,
SYS_getdents = 78,
SYS_io_setup = 206,
SYS_io_destroy = 207,
SYS_io_getevents = 208,
SYS_io_submit = 209,
SYS_io_cancel = 210,
SYS_clock_settime = 227,
SYS_clock_gettime = 228,
SYS_clock_getres = 229,
SYS_clock_nanosleep = 230, |
Thanks. Not surprisingly, I get an error for SYS_stat, SYS_lstat, and SYS_getdents. |
I'm running Ubuntu 18.04 on an NVIDIA Jetson Nano which has CUDA 10 and cuDNN 7.5 installed by default. I followed issue #253 (comment) and was able to build Torch successfully. Then I installed the additional lua modules for waifu2x and validated successfully when running
th waifu2x.lua -force_cudnn 1
. (waifu2x runs out of memory on this device without using cuDNN. I'm not too concern about this as long as it works using cuDNN.)When I tried running
th web.lua -force_cudnn 1
however, I get the following error:The text was updated successfully, but these errors were encountered: