nicovasilache

4 ms·

1.
by nicovasilache 9y ago
Python bindings are in there :) This needs a tensor library callable from python that works on GPUs. One direction we are going towards is PyTorch via ATen / Torch tensors; we already use the C++ parts of ATen. Of course any other CUDA
2.
by nicovasilache 9y ago
This is one reason I also prefer to call TC a notation personally. We can't allocate and declare inside TC, that may change in the future but for now we went for the easiest entry point to Halide and polyhedral IR we could think of. Yo
3.
by nicovasilache 9y ago
@phaedrus take a look at Weld from Stanford then, may also be useful to you. Edit: in addition to what @ozinenko pointed to of course..
4.
by nicovasilache 9y ago
The thing is that compilers usually quickly go to SSA form and it is not the best IR to optimize loops in. Then you fight it to extract a high-level IR and this little process makes it very easy to lose high-level information. We don't
5.
by nicovasilache 9y ago
punchcards .. days .. ouch that's painful :)
6.
by nicovasilache 9y ago
Hello, evolutionary algorithms by themselves I am not sure and don't have enough experience with atm. In our particular context the key is the intermediate representation on which the evolution happens and the compiler behaviors it tri
7.
by nicovasilache 9y ago
Hello, Tensor Comprehensions absolutely use techniques that have been existing for a few years (Halide) or many decades (polyhedral model, Einstein notation(century old?), ...). The TCE is definitely also a motivational prior work which als