HPC Glossary - Revision history

M Janczyk at 09:35, 15 July 2025

2025-07-15T09:35:48Z

← Older revision		Revision as of 11:35, 15 July 2025
Line 1:		Line 1:
	A short definition of the typical elements of an HPC cluster.		A short definition of the typical elements of an HPC cluster.

	;<span id="Batch_Scheduler">‎</span>Batch Scheduler / [[~~Batch_system~~]]		;<span id="Batch_Scheduler">‎</span>Batch Scheduler / [[Batch system]]
	: The software that distributes the [[#Compute_Job\|compute jobs]] of the users on the available resources (compute nodes).		: The software that distributes the [[#Compute_Job\|compute jobs]] of the users on the available resources (compute nodes).

M Janczyk at 09:32, 15 July 2025

2025-07-15T09:32:15Z

← Older revision		Revision as of 11:32, 15 July 2025
Line 1:		Line 1:
	A short definition of the typical elements of an HPC cluster.		A short definition of the typical elements of an HPC cluster.

	;<span id="Batch_Scheduler">‎</span>[[Batch Scheduler]] / ~~Batch System~~		;<span id="Batch_Scheduler">‎</span>Batch Scheduler / [[Batch_system]]
	: The software that distributes the [[#Compute_Job\|compute jobs]] of the users on the available resources (compute nodes).		: The software that distributes the [[#Compute_Job\|compute jobs]] of the users on the available resources (compute nodes).

K Siegmund at 10:07, 19 February 2025

2025-02-19T10:07:01Z

← Older revision		Revision as of 12:07, 19 February 2025
Line 14:		Line 14:

	;<span id="CPU">‎</span>CPU		;<span id="CPU">‎</span>CPU
	: Central Processing Unit. It performs the actual computation in a compute node. A modern CPU is composed of numerous [[#Core\|cores]] and layers of cache.		: Central Processing Unit. It performs the actual computation in a compute node. A modern CPU is composed of numerous [[#Core\|cores]] and layers of cache. The term is sometimes also used to describe a single core. To distinguish, one now often uses the term [[#socket\|socket]] for the whole cpu.

	;<span id="CPU_Time">‎</span>CPU Time		;<span id="CPU_Time">‎</span>CPU Time

K Siegmund at 10:01, 19 February 2025

2025-02-19T10:01:14Z

← Older revision		Revision as of 12:01, 19 February 2025
Line 83:		Line 83:

	;<span id="Socket">‎</span>Socket		;<span id="Socket">‎</span>Socket
	: Physical socket where the CPU ~~capsules~~ are placed. Often used as a synonym to CPU if a computer has more than one socket and one wants to make clear that only one of the CPU chips sitting in one socket is meant.		: Physical socket where the CPU chips are placed. Often used as a synonym to CPU if a computer has more than one socket and one wants to make clear that only one of the CPU chips sitting in one socket is meant.

	;<span id="Submit">‎</span>Submit		;<span id="Submit">‎</span>Submit

K Siegmund at 09:45, 19 February 2025

2025-02-19T09:45:02Z

← Older revision		Revision as of 11:45, 19 February 2025
Line 30:		Line 30:
	;<span id="Hyperthreading">‎</span>Hyperthreading		;<span id="Hyperthreading">‎</span>Hyperthreading
	: Modern computers can be configured so that one real compute-[[#Core\|core]] appears like two "logical" cores on the system. These two "hyperthreads" can sometimes do computations in parallel, if the calculations use two different sub-units of the compute-core - but most of the time, two calculations on two hyperthreads run on the same physical hardware and both run half as fast as if one thread had a full core. Some programs (e.g. gromacs) can profit from running with twice as many threads on hyperthreads and finish 10-20% faster if run in that way.		: Modern computers can be configured so that one real compute-[[#Core\|core]] appears like two "logical" cores on the system. These two "hyperthreads" can sometimes do computations in parallel, if the calculations use two different sub-units of the compute-core - but most of the time, two calculations on two hyperthreads run on the same physical hardware and both run half as fast as if one thread had a full core. Some programs (e.g. gromacs) can profit from running with twice as many threads on hyperthreads and finish 10-20% faster if run in that way.

			;Infiniband
			: [[Infiniband]] is a high-speed network often used to connect nodes on a HPC cluster. Omni-Path is the name of the same technology by different vendor.

	;Job		;Job
Line 48:		Line 51:
	;<span id="Node">‎</span>Node		;<span id="Node">‎</span>Node
	: An individual computer with one or more sockets, part of an HPC cluster.		: An individual computer with one or more sockets, part of an HPC cluster.

			;Omni-Path
			: [[Omni-Path]] is a high-speed network often used to connect nodes on a HPC cluster. It is basically Infiniband by a different vendor.

	;<span id="OpenMP">‎</span>OpenMP		;<span id="OpenMP">‎</span>OpenMP

K Siegmund at 09:26, 19 February 2025

2025-02-19T09:26:25Z

← Older revision		Revision as of 11:26, 19 February 2025
Line 1:		Line 1:
	A short definition of the typical elements of an HPC cluster.		A short definition of the typical elements of an HPC cluster.

	;<span id="Batch_Scheduler">‎</span>Batch Scheduler		;<span id="Batch_Scheduler">‎</span>[[Batch Scheduler]] / Batch System
	: The software that distributes the [[#Compute_Job\|compute jobs]] of the users on the available resources (compute nodes).		: The software that distributes the [[#Compute_Job\|compute jobs]] of the users on the available resources (compute nodes).

	;<span id="Batch_Script">‎</span>Batch Script		;<span id="Batch_Script">‎</span>Batch Script
	: A [[#Script\|script]] that contains information in the form of special comments at the beginning of the script which contain information about how many compute resources of what kind are needed.		: A [[#Script\|script]] that contains information in the form of special comments at the beginning of the script which contain information about how many compute resources of what kind are needed.

	;Batch System
	: See [[#Batch_Scheduler\|Batch Scheduler]]

	;<span id="Core">‎</span>Core		;<span id="Core">‎</span>Core

R Keller: MPI explanation

2024-02-06T12:29:17Z

MPI explanation

← Older revision		Revision as of 14:29, 6 February 2024
Line 44:		Line 44:

	;<span id="MPI">‎</span>MPI		;<span id="MPI">‎</span>MPI
			: Standard for the Message Passing Interface, mainly for distributed memory machines (like HPC Clusters with many compute nodes, each a shared memory system with many cores) scaling from one node to thousands of compute nodes.

	;<span id="Multithreading">‎</span>Multithreading		;<span id="Multithreading">‎</span>Multithreading

R Keller: Add OpenMP and clarify some explanations.

2024-02-06T12:27:21Z

Add OpenMP and clarify some explanations.

← Older revision		Revision as of 14:27, 6 February 2024
Line 52:		Line 52:

	;<span id="OpenMP">‎</span>OpenMP		;<span id="OpenMP">‎</span>OpenMP
			: Specification for Shared Memory parallelization based on Threads for many-core CPUs and GPUs.

	;<span id="Parallelization">‎</span>Parallelization		;<span id="Parallelization">‎</span>Parallelization
	: ~~Making it possible for~~ programs to calculate parts of the problem ~~they~~ ~~want~~ to ~~solve~~ in ~~parallel~~.		: Enabling programs to calculate parts of the problem in parallel, reducing overall wallclock time (albeit with some parallelization overhead like runtime environment, communication and synchronization).

	;<span id="RAM">‎</span>RAM		;<span id="RAM">‎</span>RAM
Line 84:		Line 85:

	;<span id="Thread">‎</span>Thread		;<span id="Thread">‎</span>Thread
	: Logical unit that can be executed independently.		: Logical unit that can be executed independently within a process, sharing resources such as allocated memory, open files and signal handlers.

	;<span id="Wall_Clock_Time">‎</span>Wall Clock Time		;<span id="Wall_Clock_Time">‎</span>Wall Clock Time

M Janczyk at 16:30, 4 December 2023

2023-12-04T16:30:41Z

← Older revision		Revision as of 18:30, 4 December 2023
Line 17:		Line 17:

	;<span id="CPU">‎</span>CPU		;<span id="CPU">‎</span>CPU
	: Central Processing Unit. It performs the actual computation in a compute node. A modern CPU is composed of numerous cores and layers of cache.		: Central Processing Unit. It performs the actual computation in a compute node. A modern CPU is composed of numerous [[#Core\|cores]] and layers of cache.

	;<span id="CPU_Time">‎</span>CPU Time		;<span id="CPU_Time">‎</span>CPU Time
	: The time that CPUs have spent to calculate something. If 10 CPU cores calculate something for 1 hour each (even if it happens within the same hour), then 10 CPU-hours have been used for this calculation.		: The time that CPUs have spent to calculate something. If 10 CPU [[#Core\|cores]] calculate something for 1 hour each (even if it happens within the same hour), then 10 CPU-hours have been used for this calculation.

	;<span id="GPU">‎</span>GPU		;<span id="GPU">‎</span>GPU
Line 32:		Line 32:

	;<span id="Hyperthreading">‎</span>Hyperthreading		;<span id="Hyperthreading">‎</span>Hyperthreading
	: Modern computers can be configured so that one real compute-[[core]] appears like two "logical" cores on the system. These two "hyperthreads" can sometimes do computations in parallel, if the calculations use two different sub-units of the compute-core - but most of the time, two calculations on two hyperthreads run on the same physical hardware and both run half as fast as if one thread had a full core. Some programs (e.g. gromacs) can profit from running with twice as many threads on hyperthreads and finish 10-20% faster if run in that way.		: Modern computers can be configured so that one real compute-[[#Core\|core]] appears like two "logical" cores on the system. These two "hyperthreads" can sometimes do computations in parallel, if the calculations use two different sub-units of the compute-core - but most of the time, two calculations on two hyperthreads run on the same physical hardware and both run half as fast as if one thread had a full core. Some programs (e.g. gromacs) can profit from running with twice as many threads on hyperthreads and finish 10-20% faster if run in that way.

	;Job		;Job
Line 46:		Line 46:

	;<span id="Multithreading">‎</span>Multithreading		;<span id="Multithreading">‎</span>Multithreading
	: Multithreading means that one computer program runs calculations on more than one compute-core using several logical "threads" of serial compute instructions to do so (eg. to work through different and independent data arrays in parallel). Specific types of multithreaded parallelization are [[#OpenMP\|OpenMP]] or [[#MPI\|MPI]].		: Multithreading means that one computer program runs calculations on more than one compute-[[#Core\|core]] using several logical "threads" of serial compute instructions to do so (eg. to work through different and independent data arrays in parallel). Specific types of multithreaded parallelization are [[#OpenMP\|OpenMP]] or [[#MPI\|MPI]].

	;<span id="Node">‎</span>Node		;<span id="Node">‎</span>Node
Line 57:		Line 57:

	;<span id="RAM">‎</span>RAM		;<span id="RAM">‎</span>RAM
	: Random Access Memory. It is used as the working memory for the cores.		: Random Access Memory. It is used as the working memory for the [[#Core\|cores]].

	;<span id="Runtime">‎</span>Runtime		;<span id="Runtime">‎</span>Runtime

M Janczyk at 16:28, 4 December 2023

2023-12-04T16:28:53Z

Show changes