News › performance
By Zayden R., September 5, 2026
AWS has unveiled the public preview of Amazon Linux 2027, boasting significant performance improvements on AMD EPYC hardware. Engineers should note upgrades like kernel 7.1, GCC 16.1, and LTO optimizations.
Amazon Web Services has rolled out the public preview of Amazon Linux 2027, a noteworthy successor to Amazon Linux 2023. This release is particularly significant for engineers working with AMD EPYC processors in the AWS cloud, as it promises substantial performance improvements. The new version introduces several key software upgrades, including the leap to Linux kernel 7.1 from 6.18 and a major compiler upgrade from GCC 11.5 to GCC 16.1. These changes alone set the stage for enhanced efficiency and capability.
Among the most impactful enhancements is the use of link-time optimizations (LTO) in package building, which can lead to more efficient code execution. Additionally, Amazon Linux 2027 enforces x86-64-v3 builds, assuming AVX/AVX2 support, further optimizing performance on compatible hardware. Python developers will also appreciate the upgrade from Python 3.9 to Python 3.14, aligning with more modern development practices.
While some updates, such as those to OpenJDK from version 25 to 26, are less dramatic, the cumulative effect of these changes can be seen in performance benchmarking. The benchmarks conducted by Phoronix, albeit limited due to budget constraints, reveal meaningful out-of-the-box performance improvements. Tests were carried out on an m8a.4xlarge instance powered by the AMD EPYC 5th Gen "Turin" processor, with 16 vCPUs, 64GB of DDR5-6400 memory, and 400GB of local storage.
Amazon Linux 2027's packages are largely based on Fedora 44~45 sources, augmented by AWS-specific changes. This preview offers a glimpse into the future of cloud-based Linux distributions, with AWS clearly prioritizing performance and modern software support. Engineers running workloads in the AWS cloud should evaluate this new release, especially if leveraging AMD EPYC hardware, to determine the potential benefits for their specific use cases.