Comments (7)
@boyedarat the next TF release will require DirectML 1.5, which includes some important performance improvements and a few new APIs. We expect to release DML 1.5 and a new version of TF around the end of March / early April. TF 2.x investigation will likely start shortly after the March release.
We do plan to switch over to the open DX headers! This should make it easier to build without requiring an insider SDK. :)
from tensorflow-directml.
It's a good question.
In future releases, one possibility is that we publish a native NuGet package containing the TensorFlow-DirectML C library to nuget.org, which could in turn depend on the matching DirectML redistributable package. This would make integration with VS-based projects quite easy, at least, and it's not too difficult to add NuGet hooks to CMake or Bazel projects either.
For now, I've added the packages under the assets of the last release. I've included the matching DirectML library in the package as well; so long as you keep DirectML..dll next to tensorflow.dll the runtime should find it without any extra work on your end.
from tensorflow-directml.
It's a good question.
In future releases, one possibility is that we publish a native NuGet package containing the TensorFlow-DirectML C library to nuget.org, which could in turn depend on the matching DirectML redistributable package. This would make integration with VS-based projects quite easy, at least, and it's not too difficult to add NuGet hooks to CMake or Bazel projects either.
For now, I've added the packages under the assets of the last release. I've included the matching DirectML library in the package as well; so long as you keep DirectML..dll next to tensorflow.dll the runtime should find it without any extra work on your end.
Just wanted to thank you very much for creating this package. It has been working well so far on with the exception of the following minor errors on startup:
2021-01-05 21:24:31.761912: E tensorflow/core/framework/op_kernel.cc:1575] OpKernel ('op: "ApplyKerasMomentum" device_type: "DML" constraint { name: "T" allowed_values { list { type: DT_FLOAT } } }') for unknown op: ApplyKerasMomentum
2021-01-05 21:24:31.762606: E tensorflow/core/framework/op_kernel.cc:1575] OpKernel ('op: "ApplyKerasMomentum" device_type: "DML" constraint { name: "T" allowed_values { list { type: DT_HALF } } }') for unknown op: ApplyKerasMomentum
Please do continue to release new pre-built C API packages. Thanks!
from tensorflow-directml.
@sritharans thanks for reporting this bug. It looks like one of our (internal) helper kernels is registered but doesn't have an associated opdef. We'll make sure to fix this.
It's good to hear that the C API packages are useful, so we'll make sure to publish them along with pypi. :)
from tensorflow-directml.
@jstoecker When is the next update coming? Would we be able to build the upcoming C API package using the open DirectX headers?
Noticed some info about an upcoming TF 2.4 support in this issue. Is this planned for the upcoming build?
Thanks for the great work.
from tensorflow-directml.
it's good to hear that the C API packages are useful, so we'll make sure to publish them along with pypi. :)
C-API package could be interesting for use by ML.NET dotnet/machinelearning#5240
Might also be interesting for SciSharp: https://scisharp.github.io/SciSharp/ cc @Oceania2018
Consider publishing them in a way that's consumable by NuGet.
from tensorflow-directml.
Latest release is out now: https://github.com/microsoft/tensorflow-directml/releases/tag/v1.15.5.dev210422
We probably will look into making this a native nuget at some point! Perhaps for the v1 release.
from tensorflow-directml.
Related Issues (20)
- Not able to use my own callbacks HOT 3
- Tensorflow-directml is not making any difference in processing times in GPU vs CPU HOT 1
- AMD APU support? HOT 2
- Cannot assign a device for operation embedding/embeddings/Initializer/random_uniform/ HOT 5
- directml on custom tensoflow build ? HOT 1
- Windows Camera post process(DMFT) with DirectML(Tensorflow) HOT 10
- It's not working on Intel Graphics 5500 HOT 1
- Use c api to load pb models HOT 1
- how to set the default device using cAPI?
- session run crashed when runing on nvidia gpu HOT 1
- Could not identify NUMA node of platform GPU ID 0, defaulting to 0. Your kernel may not have been built with NUMA support. HOT 2
- unbox expects Dml at::Tensor as inputs HOT 1
- Is it not supports the amd gpus without rocm?
- Does directml support multi-GPU training
- TensorFlow-DirectML Does Not Exist HOT 7
- AttributeError: module 'tensorflow' has no attribute 'float32' HOT 1
- CPU instructions notification after installing
- python newer versions support
- tensorflow GPU on WSL2 produces a segmentationfault
- RC astro tool operations not supported in DirectML HOT 1
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from tensorflow-directml.