🥶 TensorFrost (v0.2.0) 🥶

Yet another Python tensor library with autodifferentiation (TODO). Currently very much a work in progress.

The main idea of this library is to compile optimal fused kernels for the GPU given a set of numpy-ish functions, including other more complex things like loops and conditionals (also TODO)

Currently working platforms:

Backend/OS	CPU	CUDA	Vulkan
Windows	🚧	⛔	⛔
Linux	🚧	⛔	⛔

Examples

Installation

From PyPI

Current version is 0.1.3

pip install tensorfrost

From source

You need to have CMake installed to build the library.

First clone the repository:

git clone --recurse-submodules https://github.com/MichaelMoroz/TensorFrost.git
cd TensorFrost

Then run cmake to build the library:

cmake -S . -B build && cmake --build build

The cmake script will automatically install the compiled python module into your python environment.

Building wheel packages (optional)

You can either call clean_rebuild.bat %PYTHON_VERSION% to build the wheel packages for the specified python version (the version needs to be installed beforehand), or you can build them for all versions by calling build_all_python_versions.bat. The scripts will automatically build and install the library for each python version, and then build the wheel packages to the PythonBuild/dist folder.

Usage

Setup

For the library to work you need a C++ compiler that supports C++17 (Currently only Microsoft Visual Studio Compiler).

First you need to import the library:

import TensorFrost as tf

Then you need to initialize the library with the device you want to use and the kernel compiler flags (different for each platform):

tf.initialize(tf.cpu, "/O2 /fp:fast /openmp") # Windows + MSVC (currently the only working compiler out of the box)

TensorFrost will find any available MSVC installation and use it to compile the kernels. If you want to use a different compiler, you can specify the path to the compiler executable (TODO).

Basic usage

Now you can create and compile functions, for example here is a very simple function does a wave simulation:

def WaveEq():
    #shape is not specified -> shape is inferred from the input tensor (can result in slower execution)
    u = tf.input([-1, -1], tf.float32)
    #shape must match 
    v = tf.input(u.shape, tf.float32)

    i,j = u.indices
    laplacian = u[i-1, j] + u[i+1, j] + u[i, j-1] + u[i, j+1] - u * 4.0
    v_new = v + dt*laplacian
    u_new = u + dt*v_new

    return [v_new, u_new]

wave_eq = tf.compile(WaveEq)

The tensor programs take and output tensor memory buffers, which can be created from numpy arrays:

A = tf.tensor(np.zeros([100, 100], dtype=np.float32))
B = tf.tensor(np.zeros([100, 100], dtype=np.float32))

Then you can run the program:

A, B = wave_eq(A, B)

To get the result back into a numpy array, you can use the numpy property:

Anp = A.numpy

TensorFrost does not support JIT compilation (currently no plans either), so you must create the program before running it. Therefore the tensor operations must only be used inside a tensor program. Operations outside the function will throw an error, so if you want to do operations outside you must read the data into a numpy array first.

Operations

TensorFrost supports most of the basic numpy operations, including indexing, arithmetic, and broadcasting (only partially for now). The core operation is the indexing operation, which is used to specify indices for accessing the tensor data. Depending on the dimensinality of the tensor there can be N indices. This operation is similar to numpy's np.ogrid and np.mgrid functions, but it is basically free due to fusion.

#can be created either from a provided shape or from a tensor
i,j = tf.indices([8, 8]) 
i,j = A.indices

For example i contains:

[[0, 0, 0, ..., 0, 0, 0],
 [1, 1, 1, ..., 1, 1, 1],
 [2, 2, 2, ..., 2, 2, 2],
    ...,
 [7, 7, 7, ..., 7, 7, 7]]

And analogously for j.

These indices can then be used to index into the tensor data, to either read or write data:

#set elements [16:32, 16:32] to 1.0
i,j = tf.indices([16, 16]) 
B[i+16, j+16] = 1.0

#read elements [8:24, 8:24]
i,j = tf.indices([16, 16])
C = B[i+8, j+8]

Here we can see that the shape of the "computation" is not the same as the shape of the tensor, and one thread is spawned for each given index. This is the main idea of TensorFrost. Then all sequential computations of the same shape are fused into a single kernel, if they are not dependent on each other in a non-trivial way.

When doing out-of-bounds indexing, the index is currently clamped to the tensor shape. This is not ideal, but it is the simplest way to handle this. In the future there will be a way to specify the boundary conditions.

Scatter operations

These operations allow implementing non-trivial reduction operations, including, for example, matrix multiplication:

def MatrixMultiplication():
    A = tf.input([-1, -1], tf.float32)
    N, M = A.shape
    B = tf.input([M, -1], tf.float32) #M must match
    K = B.shape[1]

    C = tf.zeros([N, K])
    i, j, k = tf.indices([N, K, M])
    tf.scatterAdd(C[i, j], A[i, k] * B[k, j])

    return [C]

matmul = tf.compile(MatrixMultiplication)

Here the 3D nature of the matrix multiplication is apparent. The scatter operation is used to accumulate the results of the row-column dot products into the elements of the resulting matrix.

(This is not the most efficient way to implement matrix multiplication, but it is the simplest way to show how scatter operations work. In the future though, some dimensions will be converted into loop indices, and the scatter operation will be used to accumulate the results of the dot products into the resulting matrix.)

Loops and conditionals

#Mandelbrot set
z_re = tf.zeros([S, S], tf.float32)
z_im = tf.zeros([S, S], tf.float32)

def loop_body(k):
    z_re_new = z_re*z_re - z_im*z_im + c_re
    z_im_new = 2.0*z_re*z_im + c_im
    z_re.set(z_re_new)
    z_im.set(z_im_new)
    tf.if_cond((z_re*z_re + z_im*z_im) > 256.0, lambda: tf.break_loop())
        
tf.loop(loop_body, 0, 128, 1)

In this example, the loop body is a function that takes an index k and updates the z_re and z_im tensors. The loop is then run 128 times, and the loop body is executed for each iteration. The if_cond operation is used to break the loop if the condition is met. Since TensorFrost uses tracing to build the program graph, the loop body must be a function. If you do a normal loop, it will essentially just be unrolled (also wont work if the loop iteration count is not known at compile time).

Also since the setting operation can not be overloaded, the set method must be used to update the tensor data outside of this scope.

z_re.set(z_re_new) #this is fine
z_re = z_re_new #this is not fine

Just setting the tensor to a new value will actually create a new tensor on top of the old one, and the old one will not be updated.

Loops and conditionals can be stacked and nested, but they can be compiled into separate kernels inside the loop body if the data dependencies are not local (look at the QR decomposition example in the examples folder). Not all possible loop and conditional can be valid here, if the loop iteration count has a shape incompatible with the shapes of the tensors in the loop body, the program will not compile correctly.

Autodifferentiation

TODO

Advanced usage

TODO

Roadmap

Core features:

Algorithm library:

Scan, reduction, etc.
Sorting algorithms
Matrix operations (matrix multiplication, etc.)
Advanced matrix operations (QR, SVD, eigenvalues, etc.)
Fast Fourier Transform
High-level neural network layers (convolution, etc.)

Platforms:

Windows
Linux
MacOS

Backends:

(hopefully im not going to abandon this project before finishing lol)

Contributing

Contributions are welcome! If you want to contribute, please open an issue first to discuss the changes you want to make.

PWhiddy / TensorFrost