Skip to content
No description, website, or topics provided.
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
GPUAcceleration
M 1.0.6 Nov 27, 2018
TFUtility 1.0.6 Nov 27, 2018
Warp 1.0.6 Nov 27, 2018
WarpControls 1.0.6 Nov 27, 2018
WarpLib 1.0.6 Nov 27, 2018
icon
.gitattributes 1.0.0 Jun 10, 2018
.gitignore 1.0.0 Jun 10, 2018
LICENSE
README.md
Warp.sln 1.0.6 Nov 27, 2018

README.md

Using Warp

If you just want to use Warp, all tutorials and binaries can be found at http://www.warpem.com. Continue reading this README only if you're interested in compiling Warp from source.

Compiling Warp

Prerequisites

CUDA SDK 10.0 with $PATH and $CUDA_PATH environment variables set correctly
cuDNN 7.4
Visual Studio 2017 with the Windows SDK 10.0 included during installation

Modified TensorFlow 1.10 source code

GTOM
liblion

Preparations

Make sure the Warp, GTOM and liblion folders are located in the same parent folder, or modify the include paths accordingly.

Compile GTOM and liblion using the same configuration (Debug/Release) you intend to use for Warp's compilation.

Compile TensorFlow 1.10. This is a tricky thing to do on Windows, but our modified version should help. If it still doesn't compile, some googling will be required. Once compiled, create an environment variable $TENSORFLOW_LIBS that points to TF's 'build' folder. This will be required to configure all paths correctly in Warp's TFUtility project.

Compilation

Open Warp.sln in Visual Studio. If everything is configured correctly, you shouldn't see any error messages. Now build the 'Warp' project, which will also build all of its dependencies. That's it!

Other operating systems

Warp's GUI definitely won't build on Linux or OS X as there are no ongoing efforts to port WPF to these platforms. There is a good chance you will manage to build WarpLib, or any other WPF-free parts of the solution. However, we haven't tried it. We definitely intend to make a head-less, cross-platform version for processing in HPC clusters, but that is still a long way down the road.

Authorship

Warp is being developed by Dimitry Tegunov (tegunov@gmail.com) in Patrick Cramer's lab at the Max Planck Institute for Biophysical Chemistry in Göttingen, Germany.

You can’t perform that action at this time.