pytorch官网上下载libtorch各个版本的url

下载各种版本libtorch（x86架构官网直接下载，arm架构需要源码编译安装）

下载libtorch1.0(1.0.0和1.0.1)版本（cuda8.0、cuda9.0和cuda10.0 ）

Linux version:

cpu版本（Mac仅有cpu版本，和linux的cpu版本链接一样）

https://download.pytorch.org/libtorch/cpu/libtorch-shared-with-deps-1.0.0.zip

cuda版本

https://download.pytorch.org/libtorch/cu80/libtorch-shared-with-deps-1.0.0.zip
https://download.pytorch.org/libtorch/cu90/libtorch-shared-with-deps-1.0.0.zip
`

Windows version:

cpu版本

https://download.pytorch.org/libtorch/cpu/libtorch-win-shared-with-deps-1.0.0.zip

cuda版本

https://download.pytorch.org/libtorch/cu80/libtorch-win-shared-with-deps-1.0.0.zip
https://download.pytorch.org/libtorch/cu90/libtorch-win-shared-with-deps-1.0.0.zip
https://download.pytorch.org/libtorch/cu100/libtorch-win-shared-with-deps-1.0.0.zip

libtorch1.1.0（cuda9.0、cuda10.0）

Linux version:

https://download.pytorch.org/libtorch/cu90/libtorch-shared-with-deps-1.1.0.zip
https://download.pytorch.org/libtorch/cu100/libtorch-shared-with-deps-1.1.0.zip

libtorch1.2.0（cuda9.2、cuda10.0）

Linux version:

https://download.pytorch.org/libtorch/cu92/libtorch-shared-with-deps-1.2.0.zip
https://download.pytorch.org/libtorch/cu100/libtorch-shared-with-deps-1.2.0.zip

libtorch1.3.0（cuda10.0、cuda10.1）

Linux version:

https://download.pytorch.org/libtorch/cu100/libtorch-shared-with-deps-1.3.0.zip

https://download.pytorch.org/libtorch/cu101/libtorch-shared-with-deps-1.3.0.zip

libtorch1.3.1（cuda10.1）

Linux version:

https://download.pytorch.org/libtorch/cu101/libtorch-shared-with-deps-1.3.1.zip

libtorch1.4.0（cuda10.1）

Linux version:

https://download.pytorch.org/libtorch/cu101/libtorch-shared-with-deps-1.4.0.zip

libtorch1.5（cuda10.1、cuda10.2）

Linux version:

https://download.pytorch.org/libtorch/cu101/libtorch-shared-with-deps-1.5.0.zip

https://download.pytorch.org/libtorch/cu102/libtorch-shared-with-deps-1.5.0.zip

https://download.pytorch.org/libtorch/cu102/libtorch-shared-with-deps-1.5.1.zip

libtorch1.6.0（cuda10.2）

Linux version:

https://download.pytorch.org/libtorch/cu102/libtorch-shared-with-deps-1.6.0.zip

libtorch1.7.0（cuda10.2）

Linux version:

https://download.pytorch.org/libtorch/cu110/libtorch-shared-with-deps-1.7.0%2Bcu110.zip

总结

个人实践，发现libtorch对应的cuda版本对于本机的cuda版本是向下兼容的，比如cuda100的libtorch1.1.0可以在cuda9.0的机器上使用，不会报错，但是反过来会报错（cuda90的libtorch在cuda10的机器上去编译运行）。（没有所有版本都去实践，大致情况是如此，不排除个别情况）。另外要注意的是在Linux上，gcc必须要>=4.9.2,否则编译能通过，运行时报free()错。