Skip to main content

Cornelis Technical Documentation

Glossary

ACPI

Advanced Configuration and Power Interface - This industry standard is an interface that allows the BIOS, OS, and peripherals to communicate with each other about power usage.

AOC

Active Optical Cable - This electrical-to-optical cable is used to connect modules in the CN5000 Omni-Path Fabric.

AR

Adaptive Routing (also called Dynamic Routing) - This technique uses algorithms to send data packets along the best routes, reading and responding to changes in network topology.

ASIC

Application-Specific Integrated Circuit - This is an integrated circuit chip used in Omni-Path Switches and SuperNICs.

BaseSL

Base Service Level - This is the fundamental service level provided when data is transmitted.

BECN

Backward Explicit Congestion Notification - This is an alert sent to the sender that congestion has been detected. The sender should implement congestion avoidance procedures.

BMC

Baseboard Management Controller - This is a specialized processor used for remote monitoring and management of a host system.

CM

Connected Mode - This mode allows for an IP layer MTU to be larger than what can fit in a single packet. Connected mode is less efficient for bulk data transfers using smaller segment sizes.

CRC

Cyclic Redundancy Check - This is a common, error-detecting code used to detect accidental changes to digital data.

CRL

Certificate Revocation List - This is a list of digital certificates revoked by an issuing certificate authority (CA) before their scheduled expiration. Entries on the list should no longer be trusted.

CSM

Cluster System Management - This is a system administrator tool for monitoring and managing the cluster.

CSM

Control and Switching Module - This is a hardware component, often used in networking or automation systems, that allows for the control and redirection of signals or data between different points or devices

CSP

Cloud Service Provider - A third-party company offering on-demand, internet-based IT infrastructure, platforms, or software.

CUDA

Compute Unified Device Architecture - This is a platform and API created by NVIDIA that allows software to use certain types of GPUs for general-purpose processing.

DCS

Director Class Switch - This is a high-performance, scalable switch designed for interconnecting mid-sized and large High-Performance Computing (HPC) clusters, offering high bandwidth and low latency.

DDR

Double Data Rate - This transfers data on both the rising and falling edges of the clock signal.

DHCP

Dynamic Host Configuration Protocol - This network management protocol is used to assign an Internet Protocol (IP) address to a device so they can communicate using IP.

DIMM

Dual In-Line Memory Module - This type of computer memory uses random access memory mounted on a printed circuit board to connect to a motherboard, often called a stick

DLID

Destination Local Identifier - This is the address within the header of a data packet that identifies the destination computing node within the fabric.

DSAP

Distributed Subnet Administration Provider - This is a component designed to manage records for Virtual Fabrics, specifically those that match SIDs (Subnet IDs) in the DSAP's configuration file, ensuring efficient and accurate subnet management.

ECC

Error-Correcting Code - This technique detects and fixes errors in data transmission and storage.

ESM

Embedded Subnet Manager - This is a centralized entity running in a managed switch that uses a switch port to access and manage the fabric for fabrics consisting of 100 SuperNIC ports or less.