EXT4 For Linux 6.16 Brings A Change Yielding "Really Stupendous Performance"
Ted Ts'o sent out the EXT4 file-system changes today for the Linux 6.16 kernel. While EXT4 may not see as much code churn these days given its mature state compared to say Btrfs and Bcachefs, with Linux 6.16 are some tantalizing performance improvements.
The EXT4 changes for Linux 6.16 deliver fast-commit performance improvements, multi-fsblock atomic write support for bigalloc file systems, and lastly is large folio support for regular files.
The large folio support for regular files is a particular heavy-hitting performance improvement in Linux 6.16 with the EXT4 file-system. Large folio patches for EXT4 since last year have been showing very nice performance improvements and Ted Ts'o called it out as well in today's merge request. Ted wrote:
Yes, Intel's Kernel Test Robot clocks in the large folio for regular file change to EXT4 as a 37.7% improvement to the FS-Mark benchmark:
More details on the EXT4 changes for Linux 6.16 via this pull request.
With these EXT4 optimizations paired with Bcachefs performance improvements, Btrfs performance work too, and XFS atomic writes, this is one of the more exciting kernel cycles in recent times as it pertains to file-system activity.
The EXT4 changes for Linux 6.16 deliver fast-commit performance improvements, multi-fsblock atomic write support for bigalloc file systems, and lastly is large folio support for regular files.
The large folio support for regular files is a particular heavy-hitting performance improvement in Linux 6.16 with the EXT4 file-system. Large folio patches for EXT4 since last year have been showing very nice performance improvements and Ted Ts'o called it out as well in today's merge request. Ted wrote:
"New ext4 features and performance improvements:
* Fast commit performance improvements
* Multi-fsblock atomic write support for bigalloc file systems
* Large folio support for regular files
This last can result in really stupendous performance for the right workloads. For example, see [1] where the Kernel Test Robot reported over 37% improvement on a large sequential I/O workload.
[1] https://lore.kernel.org/all/[email protected]/"
Yes, Intel's Kernel Test Robot clocks in the large folio for regular file change to EXT4 as a 37.7% improvement to the FS-Mark benchmark:
More details on the EXT4 changes for Linux 6.16 via this pull request.
With these EXT4 optimizations paired with Bcachefs performance improvements, Btrfs performance work too, and XFS atomic writes, this is one of the more exciting kernel cycles in recent times as it pertains to file-system activity.