Comments (5)
@dejunzhang What's is the network bandwidth b/w your nodes? It's very likely your network is congested. It will be interesting to get some network usage info.
May I assume that both runs use identical batch size? Batch size is for per worker instance. You should adjust batch size by dividing it by $(SPARK_WORKER_INSTANCES).
from caffeonspark.
Several factors could contribute to the slowness:
-
weakest node in the cluster. Training is done synchronously, the slowest node brings down the over speed.
-
network bandwidth. Two drawbacks with ethernet connections: (a) GPU memory needs to be copied to host memory before transmission. The copying can be slow. (b) Ethernet is usually slow. We use Infiniband and RDMA for fast memory and network access.
-
as @anfeng mentioned, your effective batch size doubles when you use two nodes since the batch size in the prototxt file is per device.
from caffeonspark.
@anfeng the slave nodes are connected with a 1Gb/s switch. Yes, I didn't modify the batch size just now, and they are the same as in single node. I will have a try. thanks.
@junshi15 the slave nodes have the same hardware and environment. Now we only have ethernet connections between nodes.
from caffeonspark.
@dejunzhang Let us know if you see differences with above config changes etc, whenever
from caffeonspark.
@mriduljain the network usage during the trainning is about 90 ~ 300Mb/s and the time cost is slight higher than the single node but with lower loss. Thank you.
from caffeonspark.
Related Issues (20)
- MNIST example at one out of four worker nodes only HOT 2
- CaffeOnSpark CPU model failed but GPU model success(same scripts and same data) HOT 1
- Trouble in performing test with existing model, dataframe empty. HOT 2
- Could anyone help about build CaffeOnSpak while caffe-distri failed ? HOT 5
- Core dump failures HOT 3
- CaffeOnSpark use infiniband but Cannot find the address of another infiniband host. HOT 4
- Use infiniband HOT 2
- Infiniband not work, Help me
- Feature extraction mode running slow HOT 7
- DataLayer use data_param instead of memory_data_param HOT 2
- Parameter synchronization mode HOT 2
- Error: Exception in thread "AWT-EventQueue-0" java.lang.UnsatisfiedLinkError: /Applications/Alice 2.4.app/Contents/Required/lib/osx/libjogl_awt.jnilib: Library not loaded: /System/Library/Frameworks/JavaVM.framework/Libraries/libjawt.dylib
- err “java.lang.UnsupportedOperationException: empty.reduceLeft”
- Error running javah command: Error executing command line HOT 2
- hive java.io.filenotfoundexception system cannot find specified path
- Is CaffeOnSpark still being maintained and developed? HOT 2
- Attribute protoFile not valid
- Exception in thread "main" org.apache.spark.SparkException: Cannot load main class from JAR file:/data/lenet_memory_solver.prototxt,/data/lenet_memory_train_test.prototxt
- have anyone faced that cannot fine class google V3 when doing test HOT 2
- java.lang.RuntimeException: Error while encoding: java.lang.ArrayIndexOutOfBoundsException: 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 caffeonspark.