- A machine-readable CPUID data repository and code generator
- Purpose
- x86 CPUID data is now huge: up to 830+ bitfields, scattered over 15+ CPU manuals and 52 CPUID leaves.
- Such bitfields can be vendor specific and include multiple flags related to CPU vulnerabilities.
- Hypervisors and Operating Systems in the future will need to support heterogeneous x86 cores and live microcode updates that might change the CPUID bits on-the-fly. This means that a proper, extensive, per-CPU, CPUID data model is needed for such projects.
- We have meticulously encoded all of the 830+ publicly-known CPUID bitfields, along with their metadata, and aim for the generated CPUID files to be used by the Linux Kernel, the Xen hypervisor, UEFI TianoCore EDK II, and other foundational projects.
- Development
- NEWS
- Funding