This document is relevant for: Inf1, Inf2, Trn1, Trn2, Trn3
AWS Neuron SDK 2.27.0: Neuron Runtime release notes#
Date of release: December 19, 2025
Go back to the AWS Neuron 2.27.0 release notes home
Released versions#
Neuron Driver:
2.25.4.0Neuron Runtime Library:
2.29.40.0Neuron Collectives:
2.29.41.0
Compatibility Support Tables#
Runtime Version: 2.29.40.0
The Neuron runtime was tested for the following EC2 instances and configurations:
Instance Family |
OS Type |
OS Version |
Kernel Version |
GLIBC Version |
|---|---|---|---|---|
|
Ubuntu |
U24 |
6.14 |
2.39 |
|
Ubuntu |
U22 |
6.8 |
2.35 |
|
Rocky Linux |
RL9 |
5.14 |
2.34 |
|
Amazon Linux |
AL2023 |
6.12 |
2.34 |
|
Ubuntu |
U24 |
6.14 |
2.39 |
|
Ubuntu |
U22 |
6.8 |
2.35 |
|
Rocky Linux |
RL9 |
5.14 |
2.34 |
|
Amazon Linux |
AL2023 |
6.12 |
2.34 |
|
Ubuntu |
U24 |
6.14 |
2.39 |
|
Ubuntu |
U22 |
6.8 |
2.35 |
|
Amazon Linux |
AL2023 |
6.12 |
2.34 |
Driver Version: 2.25.4.0
The Neuron driver was tested for the following EC2 instances and configurations:
Instance Family |
OS Type |
OS Version |
Kernel Version |
GLIBC Version |
|---|---|---|---|---|
|
Ubuntu |
U24 |
6.14 |
2.39 |
|
Ubuntu |
U22 |
6.8 |
2.35 |
|
Rocky Linux |
RL9 |
5.14 |
2.34 |
|
Amazon Linux |
AL2023 |
6.12 |
2.34 |
|
Amazon Linux |
AL2 |
5.10 |
2.26 |
|
Ubuntu |
U24 |
6.14 |
2.39 |
|
Ubuntu |
U22 |
6.8 |
2.35 |
|
Rocky Linux |
RL9 |
5.14 |
2.34 |
|
Amazon Linux |
AL2023 |
6.12 |
2.34 |
|
Amazon Linux |
AL2 |
5.10 |
2.26 |
|
Ubuntu |
U24 |
6.14 |
2.39 |
|
Ubuntu |
U22 |
6.8 |
2.35 |
|
Amazon Linux |
AL2023 |
6.12 |
2.34 |
|
Amazon Linux |
AL2 |
5.10 |
2.26 |
What’s New#
Neuron Runtime Library 2.29.40.0#
Added support for Trainium3 (single node mode)
Neuron Driver 2.25.4.0#
Added support for Trainium3
Improvements#
Neuron Runtime Library 2.29.40.0#
Reduced the overhead of reprogramming the Collectives Engine by up to 100x for NEFFs compiled with the
-O1flag. This improves end-to-end performance of these NEFFs by up to 15%.Reduced NeuronCore branch overhead by up to 3x, decreasing the overhead of starting a NEFF program by up to 5%.
Reduced the overhead of starting a NEFF program by up to 50% with an on-device hardware barrier between ranks.
Improved all-gather latency by up to 35% for messages greater than 1MB in TP8 (LNC2) and TP16 (LNC1) collectives.
Added support for NRT Debug Stream APIs.
Bug fixes#
Neuron Runtime Library 2.29.40.0#
Fixed scratchpad page allocation bug that caused excessive page allocations due to page rounding error.
Fixed segfault that occurred when freeing an empty tensor.
Previous release notes#
This document is relevant for: Inf1, Inf2, Trn1, Trn2, Trn3