GPU

Graphical processing unit of virtual and physical machines.

URI: gx:class/GPU

Inheritance

• Device
  • GPU

Slots

Name	Cardinality and Range	Description	Inheritance
gpuMemory	0..1 MemorySize	Size of memory of the GPU	direct
gpuInterconnection	0..1 GPUInterconnetionTypes	Interconnection of multiple GPUs within a server system	direct
gpuProcessingUnits	0..1 xsd:integer	Number of processing units, also called compute units, execution units or str...	direct
gpuPassthrough	0..1 xsd:boolean	If true, GPU is provided via passthrough, otherwise GPU is virtualized	direct
vendor	0..1 xsd:string	Vendor of the device	Device
generation	0..1 xsd:string	Vendor specific generation of the device	Device
defaultOversubscriptionRatio	0..1 xsd:integer	Over-subscription means assigning more capacity to virtual machines that the ...	Device
supportedOversubscriptionRatio	0..1 xsd:integer	Over-subscription means assigning more capacity to virtual machines that the ...	Device

Usages

used by	used in	type	used
ServerFlavor	gpu	range	GPU
ContainerImage	gpuReq	range	GPU
ContainerResourceLimits	gpuRequirements	range	GPU
Image	gpuReq	range	GPU
PXEImage	gpuReq	range	GPU
VMImage	gpuReq	range	GPU

Identifier and Mapping Information

Schema Source

• from schema: https://w3id.org/gaia-x/development#gpu

LinkML Source

Direct

name: GPU
description: Graphical processing unit of virtual and physical machines.
from_schema: https://w3id.org/gaia-x/development#gpu
is_a: Device
slots:
- gpuMemory
- gpuInterconnection
- gpuProcessingUnits
- gpuPassthrough

Induced

name: GPU
description: Graphical processing unit of virtual and physical machines.
from_schema: https://w3id.org/gaia-x/development#gpu
is_a: Device
attributes:
  gpuMemory:
    name: gpuMemory
    description: Size of memory of the GPU.
    from_schema: https://w3id.org/gaia-x/development#slots
    alias: gpuMemory
    owner: GPU
    domain_of:
    - GPU
    range: MemorySize
  gpuInterconnection:
    name: gpuInterconnection
    description: Interconnection of multiple GPUs within a server system
    from_schema: https://w3id.org/gaia-x/development#slots
    ifabsent: string(unavailable)
    alias: gpuInterconnection
    owner: GPU
    domain_of:
    - GPU
    range: GPU_InterconnetionTypes
  gpuProcessingUnits:
    name: gpuProcessingUnits
    description: Number of processing units, also called compute units, execution
      units or streaming multiprocessors, of the GPU.
    title: gpu processing units
    from_schema: https://w3id.org/gaia-x/development#slots
    alias: gpuProcessingUnits
    owner: GPU
    domain_of:
    - GPU
    range: integer
    minimum_value: 1
  gpuPassthrough:
    name: gpuPassthrough
    description: If true, GPU is provided via passthrough, otherwise GPU is virtualized.
    title: GPU path through
    from_schema: https://w3id.org/gaia-x/development#slots
    ifabsent: 'False'
    alias: gpuPassthrough
    owner: GPU
    domain_of:
    - GPU
    range: boolean
  vendor:
    name: vendor
    description: Vendor of the device.
    from_schema: https://w3id.org/gaia-x/development#slots
    alias: vendor
    owner: GPU
    domain_of:
    - Device
    range: string
  generation:
    name: generation
    description: Vendor specific generation of the device.
    title: Generation of the device
    from_schema: https://w3id.org/gaia-x/development#slots
    alias: generation
    owner: GPU
    domain_of:
    - Device
    range: string
  defaultOversubscriptionRatio:
    name: defaultOversubscriptionRatio
    description: Over-subscription means assigning more capacity to virtual machines
      that the physical device has. Default over-subscription ratio defines the amount
      of default over-subscription as a value larger or equal to 1.0. 1.0 means no
      over-subscription.
    title: default over-subscription ratio
    from_schema: https://w3id.org/gaia-x/development#slots
    alias: defaultOversubscriptionRatio
    owner: GPU
    domain_of:
    - Device
    range: integer
    minimum_value: 1
  supportedOversubscriptionRatio:
    name: supportedOversubscriptionRatio
    description: Over-subscription means assigning more capacity to virtual machines
      that the physical device has. Supported over-subscription ratio defines the
      amount of supported over-subscription as a value larger or equal to 1.0. 1.0
      means no over-subscription.
    title: supported over-subscription ratio
    from_schema: https://w3id.org/gaia-x/development#slots
    alias: supportedOversubscriptionRatio
    owner: GPU
    domain_of:
    - Device
    range: integer
    minimum_value: 1