英伟达
直播中

h1654155275.5835

7年用户 210经验值
私信 关注
[问答]

GRID 2.0许可混淆

大家问候!
在我的工作场所,我们正在探索为VDI实验室提供一些GPU支持。
正如我的同事和我开始研究K1和K1之间的差异。
K2(网格1.0 ??)和特斯拉(GRID 2.0)存在一些实质性差异。
所以你现在用GRID 2.0告诉我,我必须购买硬件(Tesla M60.M6)和使用GPU的软件许可?
如果这个假设是正确的,那么由于许可成本,GRID 2.0的总体拥有成本将增加30%。
当我们比较K1&
K2,我们不必担心许可模式,但是当涉及到GRID 2.0时,我们必须以并发使用模式和年度支持许可我们的用户?
他们可以为教育机构提供折扣吗?
当我们添加数字时,对于具有NVIDIA许可模式的VDI中的任何GPU都没有意义。
我误会了什么吗?

以上来自于谷歌翻译


以下为原文

Greetings everyone!

At my workplace, we are exploring offering our VDI labs some GPU support.  As my co-worker and I start looking over the differences between the K1 & K2 (Grid 1.0??) and Tesla (GRID 2.0), there are some substantial differences.

So you are telling me now with GRID 2.0, I have to purchase the hardware (Tesla M60. M6) and the software licensing to use GPU ?  If that assumption is correct, the TCO for the GRID 2.0 is 30% more all because of the licensing costs.  When we both compared the K1 & K2, we didn't have to worry about licensing model but when it comes to the GRID 2.0, we have to license our users on a concurrent usage model and yearly support ? is their a discount available for educational institutions ?   When we add the numbers up, it doesn't make sense for any GPU in a VDI with the licensing model NVIDIA has in place.  Am I misunderstanding something ?

回帖(1)

李皓茗

2018-10-10 17:07:47
你好
是的,许可模式的引入并不是最受欢迎的事情,我认为没有人会否认这一点,特别是因为物理GPU也比它们的前辈更贵。
你是对的,M60和K2之间存在很大差异。
对于初学者来说,一个是在Quadro板上构建的,另一个是特斯拉。
然后是会话密度和性能的相当大的增加。
你可以用更少的物质资源做更多的事情,这是我们都想做的事情,对吧?
如果您想最大化M60投资,显然需要使用RDS / XenApp,而不是纯VDI。
更好的密度和整体硬件利用率。
关于EDU的折扣,请与GRID经销商联系以获取最新信息;-)
只是一些个人建议......我强烈建议您不要在此阶段购买K1 / K2 GPU,未来的驱动程序开发将受到限制,您将遇到Hypervisor版本限制。
XenServer 7.1和vSphere 6.5是支持K1 / K2 GPU的最新Hypervisor版本......如果您现在购买这些版本,除非它们用于R& D或某种Dev,否则您将浪费资金。
问候

以上来自于谷歌翻译


以下为原文

Hi

Yes, the introduction of the licensing model is not the most popular thing to have been done, I don't think anyone will deny that, especially as the physical GPUs are also more expensive than their predecessors. You're right though, there are substantial differences between an M60 and a K2. For starters, one was built on a Quadro board, the other is a Tesla. Then there's the rather large increase in session density and performance. You can do (a lot) more, with less physical resource, which is what we're all trying to do, right?

If you want to maximise your M60 investment, obviously subject to use case, rather than pure VDI, look to use RDS / XenApp. Much better density and overall hardware utilisation.

Regarding discounts for EDU, please speak to a GRID reseller for the most up to date information ;-)

Just a bit of personal advice ... I would strongly recommend you do not purchase K1 / K2 GPUs at this stage, future driver development will be limited and you will run in to Hypervisor version limitations. XenServer 7.1 and vSphere 6.5 are the last Hypervisor versions to support K1 / K2 GPUs ... If you purchase these now, unless they're for R&D or some sort of Dev, you'll be throwing your money away.

Regards
举报

更多回帖

发帖
×
20
完善资料,
赚取积分