Xen Project Releases Version 4.18 with New Security, Performance, and Architecture Enhancements for AI/ML Applications

Date:

Share post:

The Xen Project, an open source hypervisor hosted at the Linux Foundation, today announced the release of Xen Project Hypervisor 4.18 with architecture enhancements for High Performance Computing (HPC) and Machine Learning (ML) applications, as well as higher security and performance features. As always, a loyal and very active Xen Project community with developers from many organizations and many parts of the world contributed to this release.

“This version provides new enterprise security and high-performance features, but also prepares architectures for HPC and AI/ML applications, which require very large quantities of data processing,” said Kelly Choi, community manager, Xen Project. “We would like to thank the industry leaders and innovators who contributed to the release.”

Notable Features

1. Arm

  • The Scalable Vector Extension (SVE) is now merged in upstream Xen as a tech preview.
  • The Arm® Firmware Framework for Arm A-profile (FF-A) framework support is now merged in upstream Xen as a tech preview.
  • The memory subsystem in Xen on Arm64 is now more compliant with the Arm architecture.

2. x86

  • On all Intel systems, MSR_ARCH_CAPS is now visible in guests, and controllable from the VM’s config file. For CPUs from 2019 onwards, this allows guest kernels to see details about hardware fixes for speculative mitigations.
  • Support for features new in 4th Gen AMD EPYC Processors:
      1. CPUID_USER_DIS (CPUID Faulting) used by Xen to control PV guest’s view of CPUID data
  • Support for features new in Intel Sapphire Rapids CPUs:
      1. PKS (Protection Key Supervisor) available to HVM/PVH guests
      2. VM-Notify used by Xen to mitigate certain micro-architectural pipeline livelocks, instead of crashing the entire server
      3. Bus-lock detection, used by Xen to mitigate (by rate-limiting) the systemwide impact of a guest misusing atomic instructions
  • Support for features new in Intel Granite Rapids CPUs:
      1. AVX512-FP16
  • Add Intel Hardware P-States (HWP) cpufreq driver
  • Support for enforcing system-wide operation in Data Operand Independent Timing Mode

Also Read: The Rise of Quantum Computing: Unlocking the Potential of a Paradigm Shift

3. RISC-V and PowerPC

  • Upstream Xen GitLab CI has been set up with full Xen build and a message printed from Xen early printk

4. Security

  • 20 XSAs has been published, enhancing the security of the project to keep it safe from common vulnerabilities

5. MISRA-C

  • The project has officially adopted more MISRA-C rules, from four directives and 24 rules in 4.17 to 6 directives and 65 rules of MISRA-C

Check Out The New TalkDev Podcast. For more such updates follow us on Google News TalkDev News.

TalkDev Bureau
TalkDev Bureau
The TalkDev Bureau has five well-trained writers and journalists, well versed in B2B enterprise technology industry, and constantly in touch with industry leaders for the latest trends, opinions, and other inputs- to bring you the best and latest in the domain.
spot_img

Related articles

Sony Research and AI Singapore Sign MOU to Conduct Collaborative Research on Large Language Models for Southeast Asian Languages

Sony Research and AI Singapore (AISG) today announced the signing of a memorandum of understanding (MOU) to collaborate on research for the SEA-LION (Southeast Asian...

Drut Technologies Delivers Virtual Disaggregation for GPU Servers

Drut Technologies, today announced the launch of their DX 3.0 advanced software system, which offers virtual disaggregation for...

Micron Unveils 36GB 12-Hi HBM3E Memory for AI and HPC

Micron has unveiled its 12-Hi HBM3E memory stacks, targeting high-performance AI and HPC processors, including Nvidia's next-gen options. These...

Mobilint Unveils Next-Gen AI Chips at 2024 AI Hardware Summit

Mobilint, led by CEO Dongjoo Shin, is making headlines at Silicon Valley's AI Hardware & Edge AI Summit...