github30 / tvm

TVM integration into PyTorch

Pytorch TVM Extension

Build

You'll need to build PyTorch on top of this PR: pytorch/pytorch#18588

cd pytorch
git fetch origin pull/18588/head:tvm_dev
git checkout tvm_dev
python setup.py install

Then, you'll need to build this repo seperately

# Make sure the right llvm-config is in your PATH
python setup.py install

Test

python test/operators.py

Usage

This package transparently hooks into PyTorch's JIT, so the same tooling is applicable (see @torch.jit.script, torch.jit.trace and graph_for). See below for an example.

from tvm import relay # This imports all the topi operators
import torch_tvm

torch_tvm.enable()

# The following function will be compiled with TVM
@torch.jit.script
def my_func(a, b, c):
    return a * b + c

To disable the JIT hooks, use torch_tvm.disable().

Code Layout

• register.cpp: Sets up pybind bindings and invokes the registration of a TVM backend.
• compiler.{h,cpp}: Main logic to compile a PyTorch JIT graph with TVM.
• operators.{h,cpp}: Location of mapping from JIT IR to TVM operators.

TODO

• Add ability to register translations from opaque op names to TVM (as is done in operator.cpp) from Python.
• Zero copy set_input.
• Threadpool
• Allocator
• Operator translation
  • Add
  • Multiply
  • Convolution
  • BatchNorm
  • Relu
  • AveragePool
  • MaxPool
  • Linear
• Tensor manipulation
  • Reshape
  • Views