transport(ptp): shortcut need to order messages on reception #389
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
csegarragonz
changed the title
csegarragonz
added a commit
that referenced
this pull request
Mar 25, 2024
In this PR we bypass the usage of the PTP layer for data-plane (i.e. message-passing) operations in our MPI implementation. The goal is to increase throughput and latency, particularly of small MPI messages, by giving each MPI rank a (thread-local) receiver thread and a separate port. Note that this approach, unlike our original PTP-less MPI transport layer, only requires one additional port per physical core, irrespective of the size of applications, or number of concurrent MPI applications. For the moment, this is very MPI specific, but in the future we could use a similar per-Faaslet recv-socket approach whenever we need low-latency and high-throughput. Colaterally this fixes all the (re-)ordering issues in the PTP layer that were forcing us to run faabric with a very high number of PTP server threads. Closes #335 Closes #389
csegarragonz
added a commit
that referenced
this pull request
Mar 25, 2024
In this PR we bypass the usage of the PTP layer for data-plane (i.e. message-passing) operations in our MPI implementation. The goal is to increase throughput and latency, particularly of small MPI messages, by giving each MPI rank a (thread-local) receiver thread and a separate port. Note that this approach, unlike our original PTP-less MPI transport layer, only requires one additional port per physical core, irrespective of the size of applications, or number of concurrent MPI applications. For the moment, this is very MPI specific, but in the future we could use a similar per-Faaslet recv-socket approach whenever we need low-latency and high-throughput. Colaterally this fixes all the (re-)ordering issues in the PTP layer that were forcing us to run faabric with a very high number of PTP server threads. Closes #335 Closes #389
csegarragonz
added a commit
that referenced
this pull request
Mar 25, 2024
In this PR we bypass the usage of the PTP layer for data-plane (i.e. message-passing) operations in our MPI implementation. The goal is to increase throughput and latency, particularly of small MPI messages, by giving each MPI rank a (thread-local) receiver thread and a separate port. Note that this approach, unlike our original PTP-less MPI transport layer, only requires one additional port per physical core, irrespective of the size of applications, or number of concurrent MPI applications. For the moment, this is very MPI specific, but in the future we could use a similar per-Faaslet recv-socket approach whenever we need low-latency and high-throughput. Colaterally this fixes all the (re-)ordering issues in the PTP layer that were forcing us to run faabric with a very high number of PTP server threads. Closes #335 Closes #389
csegarragonz
added a commit
that referenced
this pull request
Mar 25, 2024
In this PR we bypass the usage of the PTP layer for data-plane (i.e. message-passing) operations in our MPI implementation. The goal is to increase throughput and latency, particularly of small MPI messages, by giving each MPI rank a (thread-local) receiver thread and a separate port. Note that this approach, unlike our original PTP-less MPI transport layer, only requires one additional port per physical core, irrespective of the size of applications, or number of concurrent MPI applications. For the moment, this is very MPI specific, but in the future we could use a similar per-Faaslet recv-socket approach whenever we need low-latency and high-throughput. Colaterally this fixes all the (re-)ordering issues in the PTP layer that were forcing us to run faabric with a very high number of PTP server threads. Closes #335 Closes #389
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.