Feeds

back to article Nvidia, VMware join to pipe high-quality 3D graphics from the cloud

Nvidia and VMware have announced that the graphics company's GRID technology is now available on the virtualization giant's desktop-as-a-service (DaaS) platform to deliver 3D graphics to enterprise desktops through cloud service providers. In addition, the two companies are working together to bring Nvidia's virtual GPU …

COMMENTS

This topic is closed for new posts.
Anonymous Coward

"Everything from Catia to 3DS Max, all the way to Battlefield 4 and Call of Duty"

But... but... why is he not telling us the thing that really matters? <insert gaming/computer graphics reference/internet meme here>

0
0

More cloud frauds going on...

This is just the same nonsense fraud claims by Microsoft and Sony for their fake next-gen ultra-slow consoles PS4 and XBoxOne ... Their marketing depts were claiming huge performance boost with cloud computing and so on... then they didn't even try selling that crap anymore..

This virtual GPU cloud thing it's not much different.

Unless Nvidia and VMWare will be connecting each customer building with no less than 1,000Gps internet connections... it's just a fraud.

No internet link either now and for the next 15 years or so will have enough bandwidth a low latency needed to achieve the claimed performance. Both for consumer and high-end enterprise products.

2
2

Re: More cloud frauds going on...

It would have been nice if you've tried it first before posting comments. The fact is that this is available now, but with XenServer 6.2SP1 and XenDesktop 7.1 instead of VMWare. The latency is noticeable, but the setup is completely usable for high-end CAD/CAM packages.

0
1
JDX
Gold badge

Re: More cloud frauds going on...

This isn't true. I've been looking into such technologies and been disappointed that while you can get DaaS, with a few niche exceptions 3D rendering was simply not provided.

It's not for everyone but definitely a powerful tool - for instance I create powerful 3D-simulators for business users who typically run basic business laptops, maybe even 5-year old business laptops with terrible onboard graphics chipsets.

This kind of approach means that rather than tell them they have to get brand new PCs for every user, I can provide an alternative.

1
3
Anonymous Coward

Re: More cloud frauds going on...

It's also available with VMware View 5.3 on vSphere 5.5 today. Both Citrix using HDX and VMware on PCoIP need decent bandwidth, 10Mb/s or so per 3D session, but not out-of-reach bandwidth. Yes, that's more than the average home broadband in the UK and US, but not ludicrous.

0
0
Silver badge

Re: More cloud frauds going on...

Stuff the (internet delivery) cloud - I can think of half a dozen clients of mine who use CADCAM stuff, but rarely concurrently under load, and having one or two beefy GPUs sharing their power in a server, rather than a dozen GPUs doing not much most of the time, would probably save them the cost of the server side hardware alone. Bandwidth/latency would not be an issue.

I'm not in a position to implement any of this stuff (I've fallen behind in the VM stuff as most of my clients just aren't of the scale to justify it, annoyingly) but I think I'll have to read up a bit...

Steven R

0
0

Re: More cloud frauds going on...

I still confirm that I have tested the abovementioned combo and it is completely usable for serious CAD/CAM work. Rendering works as expected, the vGPU driver has direct connection to the VM without the play of the virtualizing platform, in this case XenServer, the latency requirement is met by the typical Internet connections. The program reports using real NVidia driver, so no software modifications are necessary. On my laptop at home with Virgin Media 20Mb/1Mb connection it works almost as if it is on the local machine. The frame rate is of course reduced - by Citrix Receiver itself - to about 30 fps - but for CAD/CAM work this is not so important. The screen updates typically would be lossy, so they are perceived as smooth enough; you can switch to lossless, in which case the reported frame rate stays the same, but you simply see less frames displayed, so it would be jerky. All in all, there are plenty of use cases for this, but do not expect to be playing your Chrysis 2 this way any time soon at 60FPS on a 2540x1600 monitor. It's kinda funny to show ones big CAD package running just fine on a Nexus 7 tablet...

1
0
Anonymous Coward

'fraud'

This is just cloud BS showboating or at best showcasing possibilities. At no damn point do they address the real world speed we're all stuck with. Moreover they don't even address fps, controller lag, or include any ms frame statistics.

What is the lag of using such a system to even develop games? Never mind the fact that on a daily basis Unreal and Unity crash, sometimes leaving the box needing a reboot even on Win7 i.e. the sound goes out or some other sub-system problem, and this is just games development.

What happens when you have these types of crashes on a remote machine overall? Oh, but yes it will run everything and be perfectly configured! I'm rather sceptical of that. Its more likely they will decide to roll out a system wide update just when you're in the middle of a crucial build, all because it suits them...

Don't get me wrong I understand that a hundred years from now, cloud-fog may be the norm, or even 20 years from now. But to tout this rubbish now, when no one apart from theoretical users can use it, is just total crap. So you are so dead on Joerg when you talk of fraud and the Cloud Gaming hype of the consoles.

But hey, we live in a world of spin, and its guys like us that have to pick up the pieces after some bungee manger reads something in an in-flight magazine.

0
0

Re: More cloud frauds going on...

I'm already playing modern games across their GRID service using the Nvidia Shield handheld and a 4G WiFi hot spot. Plays reasonably well and I'm definitely not on a gigabyte pipe here with this setup. Seems the system is smart enough to degrade graphics a little to cope with lag issues.

0
1

Finally, but we need cards with more RAM, i.e. 16GIG????

The only users that need a workstation are the ones that really should be using a mini cluster or supercomputer time for clock cycles. All other workstation tasks, like 3D CAD are interactive and the hardware goes 99.9999% under utilized during the day.

VMware has the best bare metal performance and we are so happy with how it runs our servers with ANY load. 10 beefy servers are WAY less expensive to maintain and upgrade as opposed to 100 engineering workstations.

So we will see even more of a nose dive in the number of PC's sold per year. And consumers are quite happy with the iPad. So I guess there is no market for Windows 8.1 licenses. :-) Our users have been asking for Mac OS X in a VM as a lot of them have Mac computers at home now...

0
0

Re: Finally, but we need cards with more RAM, i.e. 16GIG????

It'd be great if Apple would license OSX on other hardware so you can legally run it in ESXi. However for now its just a pipe dream.

Its also a royal pain in the butt to get it working and the performance isn't the best.

On topic, this is a good idea but you'd need serious bandwidth. I can see its place in the market as the technology grows and improves. For now I'd still stay in house for graphic intensive VDI work as you can provide the bandwidth to the end points much easier.

0
0
JDX
Gold badge

Cost

This is my big concern. When I looked into various DaaS providers they all seemed very expensive compared to using something like EC2. It didn't seem like it would take long to have paid more than a physical PC would cost - presumably my needs don't align well with their business cases though.

0
2

You have to laugh...

...at some of the rubbish posted here about needing Terabit bandwidth and nanosecond latency to do anything meaningful. Yes, latency can impact user experience when gaming but even on a residential broadband on a more modest 720p resolution it's plenty usable. I'm already rolling this kit out to CAD users who think it's possibly the best thing ever because the many gigabyte datasets they're used to waiting for are much closer to the horsepower, which makes 10Gb and 40Gb links more accessible.

0
0
This topic is closed for new posts.