Difference between revisions of "GPGPU-FedCloud"

From EGIWiki
Jump to: navigation, search
Line 1: Line 1:
 
= Status of accelerated computing in Clouds  =
 
= Status of accelerated computing in Clouds  =
 
Need modification/support at all levels
 
Need modification/support at all levels
 
  
 
*Chipset : HW virtualization support (otherwise some limitation)  
 
*Chipset : HW virtualization support (otherwise some limitation)  
Line 14: Line 13:
  
 
== GPGPU (General-Purpose computing on Graphical Processing Units)  ==
 
== GPGPU (General-Purpose computing on Graphical Processing Units)  ==
 +
 
NVIDIA GPU/Tesla/GRID, AMD Radeon/FirePro, Intel HD Graphics,...
 
NVIDIA GPU/Tesla/GRID, AMD Radeon/FirePro, Intel HD Graphics,...
  
Line 19: Line 19:
  
 
== Intel Many Integrated Core Architecture  ==
 
== Intel Many Integrated Core Architecture  ==
<div>Xeon Phi Coprocessor</div><div>Virtualization using PCI pass-through</div><div><br></div>
+
 
 +
Xeon Phi Coprocessor
 +
 
 +
Virtualization using PCI pass-through
 +
 
 
== Specialized PCIe cards with accelerators  ==
 
== Specialized PCIe cards with accelerators  ==
<div>DSP (Digital Signal Processors)</div><div>FPGA (Field Programmable Gate Array)</div><div>Not commonly used in cloud environment</div><div><br></div><div></div>
+
 
 +
DSP (Digital Signal Processors)
 +
 
 +
FPGA (Field Programmable Gate Array)
 +
 
 +
Not commonly used in cloud environment
 +
 
 
= Hypervisors  =
 
= Hypervisors  =
  
 
== QEMU/KVM  ==
 
== QEMU/KVM  ==
<div>Supports only pass-through virtualization model</div><div>vGPU support is under development</div><div><br></div>
+
 
 +
Supports only pass-through virtualization model
 +
 
 +
vGPU support is under development
 +
 
 
== Citrix XenServer 6, VMware ESXi 5.1  ==
 
== Citrix XenServer 6, VMware ESXi 5.1  ==
  

Revision as of 13:43, 30 April 2015

Status of accelerated computing in Clouds

Need modification/support at all levels

  • Chipset : HW virtualization support (otherwise some limitation)
  • OS level: correct kernel configuration for the accelerators
  • Hypervisor: configuration pass-through, vGPU
  • CMFs: VM start, scheduler
  • FedCloud facilities: accounting, information discovery
  • Application: VM images with correct drivers for specific chipsets


Accelerators

GPGPU (General-Purpose computing on Graphical Processing Units)

NVIDIA GPU/Tesla/GRID, AMD Radeon/FirePro, Intel HD Graphics,...

Virtualization using VGA pass-through, vGPU (GPU partitioning) - NVIDIA GRID accelerators

Intel Many Integrated Core Architecture

Xeon Phi Coprocessor

Virtualization using PCI pass-through

Specialized PCIe cards with accelerators

DSP (Digital Signal Processors)

FPGA (Field Programmable Gate Array)

Not commonly used in cloud environment

Hypervisors

QEMU/KVM

Supports only pass-through virtualization model

vGPU support is under development

Citrix XenServer 6, VMware ESXi 5.1

Support both pass-through and vGPU virtualization models

Limitations:

  • vGPU support require certified server HW
  • Live VM migration is not supported
  • VM snapshot with memory is not supported