Comments (7)
agibsonccc
commented on May 11, 2024
Sorry about that! Was playing with the GPU support. Fixed/reverted. Download one of the releases here in just a few minutes. Thanks!
from deeplearning4j.
EmergentOrder
commented on May 11, 2024
Mind if I ask, what GPU support ?
from deeplearning4j.
agibsonccc
commented on May 11, 2024
Not quite there for a release. Depending on how much longer he waits, I'm going to fork it and publish a temp release.
This will be a good drop in till the next release when I can do proper AMD/Nvidia support for matrices with jcublas and aparapi. That's too big of a departure for me to do till I have a stable release out of this base part though.
from deeplearning4j.
EmergentOrder
commented on May 11, 2024
(y) cool, I was just wondering whether you had taken a different approach. I tried using MAGMA as well via jblas but found that its performance was poorer in my tests. In a heterogenous / multi-GPU environment it might be worth it though. I also used jcublas batch gemm and got good results, but ran into JVM memory overhead limitations when it comes to batches of small matrix / matrix-vector ops.
Looking forward to your work with jcublas / aparapi.
from deeplearning4j.
EmergentOrder
commented on May 11, 2024
oh actually, I should mention too that clMAGMA, the openCL AMD-supporting variant should work in place of MAGMA, meaning similar drop-in functionality for AMD cards could be straightforward to achieve, though I can't attest to the relative performance.
from deeplearning4j.
agibsonccc
commented on May 11, 2024
Great! I appreciate the references. It's going to be a bit annoying to develop the wrappers around each of the matrix libs, but having a common API for matrices that call out to native/GPU/(heaven forbid)/ plain java, will be a requirement if this lib is going to be used in any production envs. I'll be experimenting and will probably be asking around again when I go to develop the API. Thanks for the interest!
from deeplearning4j.
lock
commented on May 11, 2024
This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.
from deeplearning4j.
Related Issues (20)
- Can't use the ND4J backend on mac os M1 HOT 2
- TimeSeriesUtils.getFormatFromRnnLayer should support SameDiffLayer as it can be Recurrent as well HOT 2
- DL4JInvalidInputException when Using Convolutional Layers and Dense Layers HOT 1
- SameDiff UI Visualization
- Error on saving large SameDiff model with FlatBuffers
- Sd.evaluate failure
- Nd4j.ScatterUpdates has a large overhead
- INDArray#fmod doesn't support long data type
- Nullpointerexception in ComputationGraph.outputOfLayersDetached
- Name scope is being ignored while creating a variable
- Unsupported keras layer type GRU HOT 3
- ND4J is conflicting with Android dependencies
- Apply to remove dependency deeplearning4j-datasets from project deeplearning4j-core HOT 2
- Keras 2.14 optimizer format changed causing simple models to not import HOT 19
- Init pretrained models HTTP status code 403 HOT 3
- A script that parses XML takes only 0.00718 seconds to execute through Python, but it takes 17 seconds to execute through Python4J. Why is there such a difference in performance? HOT 6
- Please ensure that you have an nd4j backend on your classpath. Please see: https://deeplearning4j.konduit.ai/nd4j/backend HOT 2
- Please update the version of opencsv you are using
- module-info.java for 'modelimport' contains invalid (old) exports HOT 4
- python4j spring boot jar PythonExecutioner init error HOT 3
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 deeplearning4j.