[mlpack] Intel nGraph Integration

[Cale McCollough]

My name is Cale McCollough: Software and Computer Engineer and founder of
Kabuki Starship, Kabuki Toolkit, Script2, and the Serial Chinese Room,
Interprocess, and Telemetry (SCRIPT) Specification, which is like MavLink
mixed with mbed, gRPC, Qt, and coAPP. I'm attempting to put together a
unified C++ virtual machine and SDK, but it's a bit like sweeping dirt
around on the forest floor and I'm not super far. I am now using MLPack for
Kabuki Toolkit and I'm trying to make it not a nightmare to work with both
libraries. I recently came across Intel nGraph, which uses LLVM to compile
neural net models. The cuDNN support is in progress and it's Intel so it
shouldn't take too long. I'm wondering if the MLPack community has explored
if an MLPack-to-nGraph bridge is worth investing in and if there is any
support for the project. It would provide industry-standard
parallelizability. Any advice on integrating large C++ libraries is
appreciated.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://knife.lugatgt.org/pipermail/mlpack/attachments/20190211/42a2bb24/attachment.html>