Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Multi-stage Kmeans performance comparison with Petuum #66

Open
liuzhanhao opened this issue Aug 19, 2017 · 0 comments
Open

Multi-stage Kmeans performance comparison with Petuum #66

liuzhanhao opened this issue Aug 19, 2017 · 0 comments

Comments

@liuzhanhao
Copy link
Collaborator

liuzhanhao commented Aug 19, 2017

Multi-stage Kmeans performance on PowerPS

Configuration & Result:

Data set: mni8m
# of classes: 10
# of data: 8,100,000
# of features: 784
# of loading workers: 10

Multi-stage with different # of workers

Stage # of worker # of iterations batch size learning rate coefficient time used
Stage 1 5 10000 100 0.03 34.114s
Stage 2 10 10000 300 0.03 47.240s
Stage 3 20 10000 1000 0.03 75.341s

Total time: 218.339s (including loading task, init task and training task)

Multi-stage with constant # of workers

Stage # of worker # of iterations batch size learning rate coefficient time used
Stage 1 20 10000 100 0.03 73.993s
Stage 2 20 10000 300 0.03 74.483s
Stage 3 20 10000 1000 0.03 75.264s

Total time: 292.449s (including loading task, init task and training task)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant