AMD cpu 下 Pytorch 多卡并行卡死问题解决

Posted 2021-02-12 jiangoil

dataparallel not working on nvidia gpus and amd cpus

 

https://github.com/pytorch/pytorch/issues/13045

 

问题:

 

多卡运行时, 网络会卡在那里不能运行.

系统是 AMD Ryzen5 1600x 和 两张taitanXP

之前两张卡是2070+taitanXP是可以多卡运行的, 只不过是显存不一样大...

 

看了下日志, 都是下面的错误

 

these error messages were found in the dmesg log:

[1118468.873266] nvidia 0000:0a:00.0: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x000f address=0x00000000ea13a000 flags=0x0020]
[1118468.942145] nvidia 0000:0a:00.0: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x000f address=0x00000000ea139068 flags=0x0020]
[1118468.942189] nvidia 0000:0a:00.0: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x000f address=0x00000000d0000040 flags=0x0020]
[1118468.942227] nvidia 0000:0a:00.0: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x000f address=0x00000000d00007c0 flags=0x0020]
[1118468.942265] nvidia 0000:0a:00.0: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x000f address=0x00000000d0001040 flags=0x0020]
[1118468.942303] nvidia 0000:0a:00.0: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x000f address=0x00000000d0000f40 flags=0x0020]
[1118468.942340] nvidia 0000:0a:00.0: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x000f address=0x00000000d00016c0 flags=0x0020]
[1118468.942377] nvidia 0000:0a:00.0: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x000f address=0x00000000d0002040 flags=0x0020]

 

搜了一下, 似乎是一个bug . . .

 

临时解决办法:

 

修改 /etc/default/grub

 

GRUB_DEFAULT=0
GRUB_TIMEOUT_STYLE=hidden
GRUB_TIMEOUT=10
GRUB_DISTRIBUTOR=`lsb_release -i -s 2> /dev/null || echo Debian`
GRUB_CMDLINE_LINUX_DEFAULT="quiet splash"
GRUB_CMDLINE_LINUX="iommu=soft" # 注意修改这一行 ...

 

然后

sudo update grub

最后重启

 

这样就可以正常运行了