NVIDIA Unveils Industry's First Cloud-Based GPU

[btarunr]

Engineers and design professionals will be able to work anywhere on virtually any device and still have access to the computing and graphics performance of a GPU-powered workstation, with today's launch of the cloud-based NVIDIA VGX K2 GPU. Built on the NVIDIA Kepler architecture -- the world's fastest, most efficient GPU architecture -- VGX K2 adds unprecedented workstation capabilities to the NVIDIA VGX platform, which was announced earlier this year.

The NVIDIA VGX platform utilizes the GPU to allow enterprises to efficiently deliver virtualized workstation performance and capabilities -- including rich multimedia and 3D graphics -- to users on smartphones, tablets or PCs. Its VGX K2 board, which includes two workstation-class GPUs, enables enterprises to increase user density without sacrificing performance or application compatibility.



Other key benefits of NVIDIA VGX K2 include:

• Faster interactivity: With 4 GB of graphics memory per GPU, VGX K2 ensures that graphics-intensive design and content-creation applications run with ease.
• Low-latency remote display: Patent-pending remote display technology minimizes the lag traditionally associated with virtual desktop computing.
• Data center power efficiency: Using SMX, a revolutionary streaming multiprocessor, VGX K2 provides unsurpassed performance per watt for enterprise data centers.

"The VGX platform has been developed to bring rich, interactive graphics to all enterprise virtual desktop users," said Jeff Brown, general manager of the Professional Solutions Group at NVIDIA. "With VGX K2 in the data center, designers and engineers who create the core intellectual property for their companies can now access their IP from any device and still enjoy workstation-class performance."

NVIDIA and Citrix Expand Enterprise Presence of Workstation-Class Virtual Desktops
Citrix XenDesktop with HDX 3DPro uses NVIDIA VGX technology to deliver a faster, more interactive user experience. In combination with Citrix XenDesktop and Citrix XenApp, the NVIDIA VGX K2 board provides users full compatibility and performance of all their graphics and GPU computing-intensive applications. In addition, the ecosystem of Citrix Ready partners around HDX 3D will be able to use VGX K2 to provide workstation-class performance.

"In our increasingly global and mobile economy, companies are looking more and more to desktop virtualization as a critical solution, and we're working closely with NVIDIA to offer the most advanced virtualized user experience on the market," said Krishna Subramanian, vice president of marketing and partner engagement at Citrix. "XenDesktop with NVIDIA VGX acceleration aims to deliver workstation-class performance to the most demanding users in enterprises, while also increasing user density in their data centers."

Supporting Partner Quote
"NVIDIA's GPU virtualization technology builds on and accelerates Cisco's vision of delivering virtual desktops and rich-media applications through the cloud, and helps to enable enterprise customers to deploy simple, scalable and highly secure virtualization solutions," said Satinder Sethi, vice president at Cisco, an NVIDIA technology partner.

Availability
The NVIDIA VGX K2 platform for virtual workstations is expected to be available from leading server OEMs starting in early 2013. More information is available at www.nvidia.com/vgx.

View at TechPowerUp Main Site

 

[Scatler]

It's all handy dandy but i believe the price tag will be as always sky high.

 

[eidairaman1]

It's all handy dandy but i believe the price tag will be as always sky high.

its technically a workstation card, enterprise level.

 

[hardcore_gamer]

When the gaming moves to cloud, I'll quit gaming.

 

[LTUGamer]

In my opinion cloud technologies is just a toy for adult kids with bald head. Needs to stop playing this shit and come back to serious job...

 

[[H]@RD5TUFF]

on an industry scale this is a brilliant idea, for consumers, aside from little actual need for this, with ISP's attempting to give you less and less data and charge you more for the privilege, this just seems like one more way to hit the cap faster.:shadedshu

 

[remixedcat]

on an industry scale this is a brilliant idea, for consumers, aside from little actual need for this, with ISP's attempting to give you less and less data and charge you more for the privilege, this just seems like one more way to hit the cap faster.:shadedshu

this is for companies with cheap "client" PCs that want to offload rendering to a server farm rather then someone wanting to game on a phone or whatever.

 

[Disparia]

Looks like Citrix bought the PR spotlight for this release. If you go the link in the OP, looks like RemoteFX/Server 2012 is ready for VGX with VMware support coming soon.

this is for companies with cheap "client" PCs that want to offload rendering to a server farm rather then someone wanting to game on a phone or whatever.

Yes, because products are never used in unintended or unexpected ways

However I agree in a sense that if I wanted to start a business renting virtual PC's to customers, I'd probably load up on cheaper GTX 680 or HD7970. Those 4GB/6GB models would have a purpose then, lol.

 

[T4C Fantasy]

http://www.techpowerup.com/gpudb/1699/NVIDIA_VGX_K1.html

http://www.techpowerup.com/gpudb/1700/NVIDIA_VGX_K2.html

 

[TheoneandonlyMrK]

When the gaming moves to cloud, I'll quit gaming.

+1

And fullly totally BS, Amds S9000 and S7000 do the same job and were out first, and allready support Vmware and other such things inc remote gfx.

dosnt matter though eh as its PR bumph

 

[HumanSmoke]

And fullly totally BS, Amds S9000 and S7000 do the same job and were out first, and allready support Vmware and other such things inc remote gfx.

Thanks fanboy for the update.

S9000 / S7000 are passive versions of the W9000 and W7000 FirePro

Nvidia VGX K2 TOTAL power envelope <225W

AMD W9000 (w/out S7000) = 275+ W

So, unless performance per watt is irrelevant in the server class enviroment, it would seem that AMD are in a bit of a quandary...which might also account for the lacklustre reviews and pro market reaction to the new FirePro cards.

 

[TheoneandonlyMrK]

so you agree with the fact i stated then Amd had already done this so the first is unfounded.

and fanboy, factboy more like

check yourself btw stat boy ,your not lookin entirely unbiased with your hyper defensive malarkyins

 

[HumanSmoke]

so you agree with the fact i stated then Amd had already done this so the first is unfounded.

Technically, Nvidia's C/D/S870 were doing "this" (compute/ECC compliance/1:2 DP:SP) almost five years ago.

Just for the record, I didn't argue that the W/S9000 debuted before Kepler Tesla- so maybe thats a comprehension fail on your part. But given that you're supposedly from Manchester I can fully understand that English as a language may not be familiar to you. The point I noted was your " fully totally BS" kneejerk remark.

So, please, talk up the ability of an AMD card on an Nvidia thread...a board whose power usage (i.e. the chart) and real world performance parameters (i.e. the reviews) don't make it the slam dunk that you seem to think it is. Of course, once AMD wrest control of the HPC GPGPU market away from Nvidia and Kepler proves to be a flop against the mighty Tahiti feel free to resurrect the thread...since the VGX system is "fully totally BS" as you put it.

 

[Fourstaff]

In my opinion cloud technologies is just a toy for adult kids with bald head. Needs to stop playing this shit and come back to serious job...

Looks like this cloud user doesn't fully understand what cloud computing is. We are at a point where almost all online users are using some form of cloud computing.

A refresher: http://en.wikipedia.org/wiki/Cloud_computing

 

[temp02]

I have a simple setup-box (from a 3-play provider) running attached to my room TV, it's not something very powerful (because it was build to play videos and just that), there is some games you can play on it, simple ones (like checkers and the like), but the process of playing games on these kind of devices is very idiotic (or clever, because of the very low capabilities of the device): I click on the remote (IR) to perform some action on the game, the command gets sent to the ISP server where the actual game is running, and the box gets the next frame (video) for the game. These kind of boxes also have 1 or more USB ports, so wouldn't it be good to play, lets say NFS:MW, by just plugging in a remote (USB Controller) on the box and pay for 30min of game renting instead of having to buy a console, the game and the extra electricity?

Just my 2 cents.

 

[Fourstaff]

I have a simple setup-box (from a 3-play provider) running attached to my room TV, it's not something very powerful (because it was build to play videos and just that), there is some games you can play on it, simple ones (like checkers and the like), but the process of playing games on these kind of devices is very idiotic (or clever, because of the very low capabilities of the device): I click on the remote (IR) to perform some action on the game, the command gets sent to the ISP server where the actual game is running, and the box gets the next frame (video) for the game. These kind of boxes also have 1 or more USB ports, so wouldn't it be good to play, lets say NFS:MW, by just plugging in a remote (USB Controller) on the box and pay for 30min of game renting instead of having to buy a console, the game and the extra electricity?

Just my 2 cents.

Onlive?

 

[temp02]

How exactly will I access Onlive on my/a settop box? It's required that the ISP/content-provider arranges some way of doing so (like adding an extra channel or functionality) and since most (if not all) of these devices are working on a "walled garden" thats not going to happen.
Whats needed is either a fast shared/for-rent "processing and video rendering" cloud or cheap "video rendering devices for a cloud" that ISPs/content-providers can buy and setup as their own.

 

[Fourstaff]

How exactly will I access Onlive on a settop box?

Well that is the logical next step for onlive: to integrate everything

 

[TheoneandonlyMrK]

Quote:
Originally Posted by theoneandonlymrk
so you agree with the fact i stated then Amd had already done this so the first is unfounded.

Technically, Nvidia's C/D/S870 were doing "this" (compute/ECC compliance/1:2 DP:SP) almost five years ago.

so that was a yes, I did not in anyway big up ,promote or argue the favour of Amd , wind your neck in and actually read my posts before retorting, i finished the first post adequately if you had read it, you ranter

dosnt matter though eh as its PR bumph

im not biased I just want acurate headlines, as that one missleads.

 

[HumanSmoke]

im not biased I just want acurate headlines, as that one missleads.

As do a great many press releases. I notice you don't generally refrain from putting the boot in with the exception of one company

Proof then as if it were required that this Pr bumph has a point

But then of course, every A10-5800K is 6.5G capable, so your comment holds truth. I stand corrected.


I had actually thought that the point of the release was to highlight a single board solution where previously the same 3D render/display and co-processor duties were the province of seperate add-in cards.