Expanding CXL Software Ecosystem through HMSDK on Linux
Honggyu Kim, SK hynix - With the advent of CXL memory, some systems are becoming more heterogeneous having memory systems with different latency and bandwidth characteristics. They are usually handled as different NUMA nodes in separate memory tiers and CXL memory is used as slow tiers because of its protocol overhead compared to DIMM attached local DRAM. In this kind of systems, we need to be careful placing memory pages on proper NUMA nodes. If the placement is suboptimal, some frequently accessed pages might reside on slow tiers and it makes performance degradation unexpectedly. This session will introduce a heterogeneous memory software development kit, called HMSDK[1]. This project includes weighted interleaving[2] for bandwidth expansion and DAMON based memory tiering[3] for capacity expansion. In addition, it also provides heterogeneous memory allocation APIs and a user control tool[4] so that developers can fine tune their program by explicitly managing specific memory area using hmalloc APIs and hmctl tool. HMSDK has contributed its kernel features to the mainline Linux kernel successfully. The weighted interleaving is supported from linux-v6.9 and linux-v6.11 was recently released with its DAMON based tiered memory management. In addition, the userspace tools such as numactl and damo has accepted HMSDKs features for the ease of use for general users. In this talk, Honggyu will explain how HMSDK tries to expand CXL software ecosystem on Linux environment.

Exploring CXL Memory Disaggregation Use Cases and System Benefits

Microsoft Just Released Their Own Linux Distro: Should You Be Worried?

The Moment That Changed Software Development!

Turing Award Winner: Disagreeing with Google, Postgres, Future Problems | Mike Stonebraker

Linux File System Structure Explained: From / to /usr | Linux Basics

Creator of C++: Bell Labs, Negative Overhead Abstraction, Mistakes | Bjarne Stroustrup

MIT Just Revealed the AI Bubble's Fatal Flaw

OSDI '24 - Managing Memory Tiers with CXL in Virtualized Environments

The Open Source community is collapsing

Abstract Black and White wave pattern| Height Map Footage| 3 hours Topographic 4k Background

Building the PERFECT Linux PC with Linus Torvalds

10 Images | Coastal Citrus Floral Summer Paintings Screensaver l Frame TV ART |

Linux 7.1, KDE Plasma 6.7, Antergos Returns?, Commodore Callback, AUR Update & more Linux news

This is why more and more projects are leaving GitHub!

Andrej Karpathy: From Vibe Coding to Agentic Engineering w/ Stephanie Zhan

Boost your AI workload performance using CXL memory

UnifabriX Memory over Fabrics: An Open Journey from CXL to UALink in AI Infrastructures

Anthropic is Completely F*cked.

