The news
Linux 7.3-rc1 has been released. The kernel now stands at 40.98 million lines in total. New AMD code and related changes form one third of the entire update.
Context
Release candidates mark the start of the stabilization phase for each kernel cycle. This rc1 is described as a big one. Earlier kernels followed the same process of collecting features and fixes before the first test release. The cycle begins with a merge window that pulls in new material from maintainers and vendors. Once that window closes, the focus shifts to testing and refinement rather than new additions. The appearance of a sizable rc1 signals that a substantial volume of work arrived during the merge period.
Details
The 40.98 million line count covers the full codebase at this stage. AMD contributions make up roughly one third of the changes that entered during this cycle. No other specific component sizes or patch counts appear in the available report. The total reflects the cumulative growth of the tree after the merge window closed. One vendor’s share reaching one third points to concentrated activity around hardware enablement and related driver updates. The rest of the changes come from the usual mix of architecture work, networking, filesystems, and core fixes, though their individual weights are not broken out here.
The release candidate process itself remains unchanged from prior cycles. Maintainers submit pull requests, Linus Torvalds integrates them, and the first tagged rc appears once the merge window ends. In this case the resulting rc1 is large enough to stand out in the regular announcements that accompany each new kernel series. The line count serves as a simple metric that developers track across releases to gauge overall growth.
Reactions / counterpoints
No on-the-record comments from other maintainers or vendors appear in the source material. The report limits itself to the headline numbers and the observation that AMD changes dominate the diff.
Why it matters
Kernel size growth reflects ongoing additions for new hardware and drivers. When one vendor supplies a third of the changes in a single cycle, the result is faster support for that vendor’s platforms. Engineers who maintain distributions or run AMD systems will see the impact first in upcoming test kernels. The pattern also shows how vendor code continues to drive kernel development volume. Over time this can shift maintenance effort toward reviewing and integrating external contributions rather than core architecture work.
For users the practical outcome is earlier access to fixes and features tied to recent AMD silicon. The current rc1 size suggests the final 7.3 release will carry a noticeable share of that vendor work into stable use. Distribution teams that backport patches will have a larger set of AMD-specific commits to evaluate. Hardware vendors gain quicker validation of their platforms inside the mainline tree, which can shorten the time between silicon launch and broad software support.
At the same time, a single vendor’s outsized share raises questions about long-term balance inside the project. Core kernel developers must still review and merge the incoming code, even when the bulk originates elsewhere. This dynamic has existed for years across multiple vendors, yet the one-third figure in a single rc1 makes the concentration visible. Readers who track kernel statistics will compare this cycle against previous ones to see whether the proportion holds or returns to a more even split.
The line-count metric itself is crude but consistent. It captures everything from new drivers to documentation updates and does not distinguish between high-value and low-value additions. Still, when the number moves by several hundred thousand lines in one merge window and one company accounts for a third, the signal is clear: hardware enablement remains a primary engine of kernel growth. Teams that ship products on AMD platforms can plan integration work around the upcoming 7.3 schedule rather than waiting for later point releases.
---
Sources:
{"word_count": 612, "sources_used": 1}
No comments yet