-
Notifications
You must be signed in to change notification settings - Fork 21
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
[Feature] send and recv use multiple cpus #131
Comments
@NKcqx to triage. |
This feature is somewhat important in secretflow's HEU based algorithms. please help. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
The current send and recv is an grpc service. The grpc service limits the use of CPU to 1 core per connection. grpc/grpc-swift#992
This limit is by design, because grpc service considers the senario when a servie should handle multiple connections.
However, between two clusters, we may have only a handful connections that each needs to send a large object to the other end.
Sometimes we can superposition computation with such communications but sometimes it is hard to do so, particularly when communication time is much longer than computation time.
For example, we have two secretflow clusters, and one needs to send a large ciphertext to another. Using multiple CPUs to do send and receive can really speed up communications in this case.
Can you provide the feature to use multiple cores when doing send and receive?
The text was updated successfully, but these errors were encountered: