Feature #3028
support shared vgpu, eg: nvidia GRID
Status: | Closed | Start date: | 07/10/2014 | |
---|---|---|---|---|
Priority: | Normal | Due date: | ||
Assignee: | - | % Done: | 0% | |
Category: | Core & System | |||
Target version: | Release 4.14 | |||
Resolution: | fixed | Pull request: |
Description
Nvidia has released GRID, it can be shared by more than one vm.
XenDesktop and WindowsServer has supported it.
Can opennebula integrate this feature?
Associated revisions
feature #3028: Adds PCI_DEVICES to HostShare
feature #3028: kvm probe to get PCI devices
feature #3028: Fix PCI set bugs
feature #3028: Add PCI capacity when deploying a VM
feature #3028: Fix PCI device assignment
feature #3028: Delete pci devices from host when removing the VM. Always
include VMID in PCI elements
feature #3028: Print HostShare and HostXML into streams. Scheduler now parses PCI device lists
feature #3028: show pci devices in onehost show
feature #3028: add pci passthrough to deployment file
feature #3028: Scheduler check and set PCI device requests
feature #3028: Add comments to PCI format. Init pci list in get_*
methods
Feature #3028: Add PCI_DEVICES to onedb migrator
Feature #3028: Add PCI dev checks to onedb fsck
feature #3028: check all needed pci params for deployment file
Feature #3028: Add PCI info to im_dummy driver
Feature #3028: Fix sched log formatting
Feature #3028: Check pci devices in Host::test_capacity
Feature #3028: Add pci devices to sunstone
Feature #3028: Set maxlength for pci inputs
feature #3028: Remove devices not shown in monitor from host. Recover
constness of test method. Get rid of unneeded methods
feature #3028: Keep used PCI devices when updating monitoring
feature #3028: tidy up pci devices output
Feature #3028: Fix pci table for 1 element, remove table columns
feature #3028: Make PCI restricted attribute for VMs
feature #3028: bug assigning pci devices to VMs
Feature #3028: Update host xsd with pci_devices
Feature #3028: Ignore cpu,mem when capacity is not enforced
feature #3028: Update HostShare::get_pci_value check for empty and wrong
values
feature #3028: Fix get_pci_value logic
Feature #3028: Another fix for get_pci_value
History
#1 Updated by Ruben S. Montero almost 7 years ago
- Tracker changed from Feature to Request
This depends on the underlying hypervisor, moving it to request to find out how this can be configured.
#2 Updated by 海涛 肖 almost 7 years ago
Now a days, physical servers are coming up with graphic cards that have multiple GPUs, VMs running on cloud can leverage the high computation power of GPU to meet high graphics processing requirements like Auto-CAD, Photoshop etc. Also, there are cards in market which supports sharing of a GPU cards among multiple VMs by creating vGPUs for each VM, e.g. NVIDIA has introduced vGPU capable cards GRID K1 and K2 which allow multiple vGPUs on a single physical GPU.
With vGPU technology, the graphics commands of each virtual machine are passed directly to the underlying dedicated GPU, without translation by the hypervisor. This allows the GPU hardware to be time-sliced and shared across multiple VMs.
Xenserver has added the support for NVIDIA GRID K1 and GRID K2 cards. It allows the VMs on XenServer hosts to use the GPU cards in following ways:
GPU-passthrough: It allows the hypervisor to assign the entire PGPU to a VM, this is useful for a power users.
VGPU: It allows the VM to share a PGPU device with other VMs, this is useful for tier 2 users.
#4 Updated by Carlos Martín almost 6 years ago
- Tracker changed from Request to Feature
- Category set to Core & System
- Status changed from Pending to Closed
- Target version set to Release 4.14
- Resolution set to fixed