Source: Reprinted with permission from OSC Open Source Community (ID: oschina2013)
Recently, Tencent engineer Kairui Song submitted a new patch to the Linux kernel mailing list, proposing the introduction of a new architecture called Swap Table, which integrates swap cache, swap mapping, and swap allocator into a new backend infrastructure.
According to reports, the latest Phase I patch includes a total of 9 changes and has already shown significant results. In multiple tests, both small ARM devices and large x86_64 servers exhibited performance improvements:
Overall acceleration of approximately 5%–20%;
In virtual machine scalability scenarios, some tests showed improvements exceeding 20%;
Kernel build time reduced by several percentage points;
Throughput of in-memory databases such as Redis and Valkey increased by 6%–7%.
This optimization not only brings higher throughput and lower latency but also reduces memory usage, laying the foundation for future expansion and functional optimization.
In fact, the first round of Swap Table patches released in May this year already demonstrated a 20–30% performance gain, and this further strengthens the potential in this direction.
Although some developers believe that modern systems often reduce or disable swap usage, the community generally recognizes the value of this improvement in high-load scenarios and virtualization environments.
As the patch progresses, this optimization is expected to become the default mechanism in future Linux kernels, providing the system with more stable and efficient memory management capabilities.
1. Why is code that is clearly typed by hand called a script?
2. DeepSeek's "Extreme You Are Beautiful" bug has received an official response.
3. What is UE8M0 that caused domestic chips to surge overnight?
4. Interviewer: You will be notified of the interview results within six months...
5. A single statement from DeepSeek caused domestic chips to collectively surge! What exactly is the UE8M0 FP8 behind it?