cuProj: GPU-Accelerated Coordinate Projection

Note

cuProj depends on RMM from RAPIDS.

Resources

• cuProj User's Guide: Python API reference and guides
• cuProj Developer Documentation: Understand cuProj's architecture
• Getting Started: Instructions for installing cuSpatial/cuProj
• cuProj/cuSpatial Community: Get help, collaborate, and ask the team questions
• cuProj Issues: Request a feature/documentation or report a bug (file issues in cuSpatial repo).

Overview

cuProj is a library and Python package for accelerated geographic and geodetic coordinate transformations. cuProj can transform billions of geospatial coordinates per second from one coordinate reference system (CRS) to another on GPUs. This includes cartographic projections as well as geodetic transformations. cuProj is implemented in CUDA C++ to run on GPUs to provide the highest performance.

cuProj provides a Python API that closely matches the PyProj API, as well as a header-only C++ API. While the C++ API does not match the API of Proj, it is designed to eventually expand to support many of the same features and transformations that Proj supports.

Currently cuProj only supports a subset of the Proj transformations. The following transformations are supported:

• WGS84 (EPSG: 4326) to/from any of the 60 UTM zone transformations (EPSG: 32601-32660, 32701-32760).

Example

The Python API is closely matched to PyProj and data can seamlessly transition between the two:

import cuproj
import pyproj

# Create a PyProj transformer
pyproj_transformer = pyproj.Transformer.from_crs("EPSG:4326", "EPSG:32613")

# Create a cuProj transformer
cuproj_transformer = cuproj.Transformer.from_crs("EPSG:4326", "EPSG:32613")

# Transform a grid of points around the San Francisco Bay using PyProj
num_points = 10000
grid_side = int(np.sqrt(num_points))

x, y = np.meshgrid(np.linspace(min_corner[0], max_corner[0], grid_side),
                   np.linspace(min_corner[1], max_corner[1], grid_side))
grid = [x.reshape(-1), y.reshape(-1)]

pyproj_result = pyproj_transformer.transform(*grid)

# Transform a grid of points around the San Francisco Bay using cuProj
cuproj_result = cuproj_transformer.transform(*grid)

Note that the cuProj transformer is created from the same CRSs as the PyProj transformer. The transformer can then be used to transform a grid of points. The result of the transformation is returned as a tuple of x and y coordinates. The result of the PyProj transformation is a tuple of Numpy arrays, while the result of the cuProj transformation is a tuple of CuPy arrays.

Also note that in the above example, the input data are in host memory, so cuProj will create a copy in device memory first. Data already on the device will not be copied, resulting in higher performance. See the simple cuProj Benchmark notebook for an example.