RAS Features

CPU

Corrected Machine Check Interrupt (CMCI)

Corrects error-triggered interrupts.

Memory

Failed DIMM Isolation

Identifies faulty DIMMs to facilitate isolation and replacement of the faulty DIMMs.

Memory Thermal Throttling

Automatically adjusts the memory temperature to prevent the memory from being damaged due to overheat.

Rank Sparing

Uses some memory ranks for backup to prevent the system from breaking down due to uncorrectable errors.

Memory Address Parity Protection

Detects memory command and address errors.

Memory Demand and Patrol Scrubbing

Corrects correctable errors upon detection. If these errors are not corrected in a timely manner, uncorrectable errors may occur.

Memory Mirroring

Provides high reliability for the system via mirroring.

Single Device Data Correction (SDDC)

Corrects single-chip multi-bit errors to improve memory reliability.

Device Tagging

Degrades and rectifies memory faults to improve memory availability.

Data Scrambling

Optimizes data flow distribution to reduce the error probability and improve memory data flow reliability and address error detection.

PCIe

PCIe Advanced Error Reporting

Provides a PCIe advanced error reporting mechanism to improve server serviceability.

UPI

Intel UPI Link Level Retry

Provides a retry mechanism to improve the reliability of UPI links.

Intel UPI Protocol Protection via CRC

Provides cyclic redundancy check (CRC) protection for UPI data packets to improve system reliability.

System

Core Disable For FRB (Fault Resilient Boot)

Isolates a faulty CPU core during startup to improve system reliability and availability.

Corrupt Data Containment Mode

Marks the memory storage unit when a data error occurs to limit the impact on the running program and improve system reliability.

Socket disable for FRB (Fault Resilient Boot)

Isolates a faulty socket during the BIOS startup process to improve system reliability.

Architected Error Records

With the features such as eMCA, the BIOS collects error information recorded in hardware registers in compliance with UEFI specifications, notifies the OS through the APEI interface of the ACPI, and locates the error unit, improving system availability.

Error Injection Support

Implements fault injection to verify RAS features.

Machine Check Architecture (MCA)

Provides a software repair function to rectify uncorrectable errors to improve system availability.

Enhanced Machine Check Architecture (eMCA): Gen2

Improves system availability.

OOB access to MCA registers

The out-of-band system can access MCA registers through the PECI. When a fatal error occurs in the system, the out-of-band system can collect onsite data to facilitate subsequent fault analysis and locating and improve system serviceability.

BIOS Abstraction Layer for Error Handling

The BIOS processes errors and reports error information to the OS based on specifications, improving system serviceability.

BIOS-based Predictive Failure Analysis (PFA)

The OS takes the lead. The BIOS provides information about physical memory error units. The OS tracks, predicts, and handles the errors.