You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@cloudstack.apache.org by "ASF GitHub Bot (JIRA)" <ji...@apache.org> on 2017/11/26 13:59:00 UTC
[jira] [Commented] (CLOUDSTACK-10159) Addition of NVIDIA Tesla M60
card in CloudStack (XenServer 6.5 SP1)
[ https://issues.apache.org/jira/browse/CLOUDSTACK-10159?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16266011#comment-16266011 ]
ASF GitHub Bot commented on CLOUDSTACK-10159:
---------------------------------------------
nitin-maharana opened a new pull request #2339: [WIP] CLOUDSTACK-10159 Addition of NVIDIA Tesla M60 card in CloudStack (XenServer 6.5 SP1)
URL: https://github.com/apache/cloudstack/pull/2339
Introduction:
==========
Currently, CloudStack gets GPU capability through NVIDIA GRID K1 and GRID K2 cards only on the XenServer Host. Now, XenServer 6.5 SP1 has added support for NVIDIA Maxwell card(Tesla M60), which supports both Passthrough and vGPU features. This improvement would allow the user to get benefits of Telsa GPU cards while deploying a VM with GPU enabled.
Benefits of Tesla M60:
=================
1. The Tesla M60 card supports latest NVIDIA GRID 2.0 technology.
2. The Tesla M60 can run more VMs per server. It supports a maximum of 128 VMs on a XenServer 6.5 SP1 host.
3. Now It can run on a blade server and supports vGPU on Linux VMs.(Only on M60-8Q, M60-4Q, M60-2Q, M60-1Q, M60-0Q)
4. The Tesla M60 card supports a maximum resolution of 2560 X 1600.
5. Tesla M60 has a maximum CUDA core of 4096 on both the GPUs (2048 each).
Tesla M60 Virtual GPU Types:
======================
Physical GPUs | GRID Virtual GPU | Frame Buffer (MB) | Virtual Display Heads | Maximum Resolution per Display Head | Maximum vGPUs per GPU | Maximum vGPU per Board
-- | -- | -- | -- | -- | -- | --
2 | M60-8Q | 8192 | 4 | 4096×2160 | 1 | 2
2 | M60-4Q | 4096 | 4 | 4096×2160 | 2 | 4
2 | M60-2Q | 2048 | 4 | 4096×2160 | 4 | 8
2 | M60-1Q | 1024 | 2 | 4096×2160 | 8 | 16
2 | M60-0Q | 512 | 2 | 2560×1600 | 16 | 32
2 | M60-1B | 1024 | 4 | 2560×1600 | 8 | 16
2 | M60-0B | 512 | 2 | 2560×1600 | 16 | 32
2 | M60-8A | 8192 | 1 | 1280×1024 | 1 | 2
2 | M60-4A | 4096 | 1 | 1280×1024 | 2 | 4
2 | M60-2A | 2048 | 1 | 1280×1024 | 4 | 8
2 | M60-1A | 1024 | 1 | 1280×1024 | 8 | 16
Reference:
========
[1] https://www.citrix.com/blogs/2015/11/06/nvidia-m60-support-on-xenserver/
[2] http://hcl.xenserver.org/gpus/28/Nvidia_Tesla_M60
[3] http://images.nvidia.com/content/grid/pdf/GRID-vGPU-User-Guide.pdf
Note:
====
1. The change is not tested against the Tesla M60 card.
2. If any community user has the physical card available in their lab, I request you to test this change and report the result. I would appreciate that.
3. The Marvin tests were run against NVIDIA GRID K1 and K2 cards, the result is successful.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
> Addition of NVIDIA Tesla M60 card in CloudStack (XenServer 6.5 SP1)
> -------------------------------------------------------------------
>
> Key: CLOUDSTACK-10159
> URL: https://issues.apache.org/jira/browse/CLOUDSTACK-10159
> Project: CloudStack
> Issue Type: Improvement
> Security Level: Public(Anyone can view this level - this is the default.)
> Components: API, UI
> Affects Versions: 4.11.0.0
> Reporter: Nitin Kumar Maharana
> Fix For: 4.11.0.0
>
>
> Introduction:
> ==========
> Currently, CloudStack gets GPU capability through NVIDIA GRID K1 and GRID K2 cards only on the XenServer Host. Now, XenServer 6.5 SP1 has added support for NVIDIA Maxwell card(Tesla M60), which supports both Passthrough and vGPU features. This improvement would allow the user to get benefits of Telsa GPU cards while deploying a VM with GPU enabled.
> Benefits of Tesla M60:
> =================
> 1. The Tesla M60 card supports latest NVIDIA GRID 2.0 technology.
> 2. The Tesla M60 can run more VMs per server. It supports a maximum of 128 VMs on a XenServer 6.5 SP1 host.
> 3. Now It can run on a blade server and supports vGPU on Linux VMs.(Only on M60-8Q, M60-4Q, M60-2Q, M60-1Q, M60-0Q)
> 4. The Tesla M60 card supports a maximum resolution of 2560 X 1600.
> 5. Tesla M60 has a maximum CUDA core of 4096 on both the GPUs (2048 each).
> Tesla M60 Virtual GPU Types:
> =======================
> ||Physical GPUs||GRID Virtual GPU||Frame Buffer (MB)||Virtual Display Heads||Maximum Resolution per Display Head||Maximum vGPUs per GPU||Maximum vGPU per Board||
> |2|M60-8Q|8192|4|4096×2160|1|2|
> |2|M60-4Q|4096|4|4096×2160|2|4|
> |2|M60-2Q|2048|4|4096×2160|4|8|
> |2|M60-1Q|1024|2|4096×2160|8|16|
> |2|M60-0Q|512|2|2560×1600|16|32|
> |2|M60-1B|1024|4|2560×1600|8|16|
> |2|M60-0B|512|2|2560×1600|16|32|
> |2|M60-8A|8192|1|1280×1024|1|2|
> |2|M60-4A|4096|1|1280×1024|2|4|
> |2|M60-2A|2048|1|1280×1024|4|8|
> |2|M60-1A|1024|1|1280×1024|8|16|
> Reference:
> ========
> [1] https://www.citrix.com/blogs/2015/11/06/nvidia-m60-support-on-xenserver/
> [2] http://hcl.xenserver.org/gpus/28/Nvidia_Tesla_M60
> [3] http://images.nvidia.com/content/grid/pdf/GRID-vGPU-User-Guide.pdf
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)