README.md

Tensors through FFI

In the first tutorial we saw how to edit the values of the array associated with the underlying Storage of a Tensor. Here we'll see how to send the whole Tensor to C and access its internals.

C source code

The main difference is that we need to #include <TH/TH.h> in our src/size.c

#include <TH/TH.h>

void printSize(THFloatTensor* src) {
  ...
}

in order to be able to use THFloatTensor and other Torch releated stuff.

Compilation details

Moreover, we need to tell gcc where to find libraries and shared objects. We can do this by adding some options to our src/Makefile

LIBOPTS = -I$(HOME)/torch/install/include -L$(HOME)/torch/install/lib -lTH

where $(HOME) points to your home ~, -I provide the location of the header file, -L specify where to find the shared object which is located with the -l (lowercase L) argument, and which name here is libTH.so or libTH.dylib.

Scripting in Lua

Finally, from Lua src/main.lua will send the whole Tensor's struct

-- Main program ----------------------------------------------------------------
-- Define a random sized Tensor
math.randomseed(os.time())
x = torch.FloatTensor(unpack(torch.Tensor(math.random(10)):random(10):totable()))
-- Send it to C
C.printSize2D(x:cdata())