Last week, Nvidia launched the first graphics processing unit (GPU) designed for the cloud, dubbed Kepler. Supporting vendors include a who’s who of server providers, such as HP, Dell, Cisco, and IBM — all of which will have products on the market shortly.
The whole concept behind these servers is to serve up a desktop experience from the cloud. This means delivering games, applications, utilities, and media to any device that will run the client: iPads, iPods, Android tablets, smartphones, and even cars and smart TVs. As this technology comes to market, it will increasingly not matter what you are using — you’ll be able to get your stuff on it as long as it is connected with decent bandwidth.
Let’s talk about some of the results.
Gaming from anything
On stage at its GPU Technology Conference in San Jose, Nvidia had one person on an iPad gaming head to head with another on a new LG TV using a service called Gaikai. The demo game was Hawken, a mech-oriented title that isn’t even in market yet. These two were gaming on hardware that couldn’t hope to run top-line graphics intensive game locally. Yet both where pounding away at each other, and the amazing thing was, the guy on the tablet was winning, showcasing that screen size didn’t matter as much as gaming skill.
This is often the problem with games: If it comes out on one platform and you or your friends don’t have that platform, not only can’t you play the game, the developer gets a fraction of the available revenue. But if games were delivered like streamed movies, then they could go everywhere. You could play from your connected AV system in your car, your iPad, or your TV in the home.
This is truly cloud computing, though Nvidia calls it GeForce Grid.
Windows on an iPad
I was out to breakfast the other day, and I have a nasty habit of listening in on the conversation at neighboring tables if it has to do with tech. The guy talking had been a recent convert from Windows to the Mac, and was talking about switching back because the Mac sucks. (His words not mine, no desire to peg the hate-mail meter this week.) He was complaining because he was going to have to dump his near-new MacBook Pro for an Ultrabook, and he was going to lose on that investment.
Well, what if you could run Windows on a Mac, or an iPad, or anything that would host a tiny client? If you like Apple hardware but hate the Apple platform, you can still run Windows. If you want to run Windows on your big smartphone or tablet in an emergency, you can do that, too.
Citrix demonstrated new hardware that could scale to support 100 desktops off one tower that looked smaller than my (admittedly rather large) PC.
This is the freedom to run what you want wherever you want. To not be tied to Apple or anyone else. To have software delivered like it was electricity. Someone else worries about malware, and backups, and making sure a catastrophic event doesn’t destroy your digital life along with your real one.
Galaxy-class performance
One of the most fascinating demonstrations had to do with modeling galaxy-class events. No I’m not referring to something out of Star Trek (the Enterprise was a Galaxy Class Starship). What Nividia showed was the progress from its existing Fermi platform, which can model the birth of the universe, to the Kepler platform, which can model what’s going to happen in a few short years when the Andromeda Galaxy runs into our own. Granted, a few short years in galaxy-class events is 3.5 billion years, so no need to jump under a table (not this would do you any good, mind you). As you can imagine, the scale is massive, and the capability is a magnitude (10 times) greater than what it was with the older hardware.
We often get excited about 20 percent performance leaps, so 10 times the performance is amazing. If this level of advancement keeps up, heck, we’ll be obsolete in a few years.
Universal robotics
You may think I’m joking on this last one, but one of the other Nvidia presenting at the show was Universal Robotics. This is the company bringing to market thinking robots that can respond to sensor-based events. In short, they can see and change their actions based on what they see. I’m hoping the eventual result is more like Robbie the Robot than Terminator, but I have my doubts. In any case, at the Nvidia conference, we once again saw major progress with regard to what you can do in the cloud, and even what machines will be able to do in the near-term future. Granted, they may be the only thing that is left of us in 3.5 billion years to say “oh crap” when the galaxies do collide.
And on that festive note, I’ll leave you to ponder our near, and far, future.
Guest contributor Rob Enderle is the founder and principal analyst for the Enderle Group, and one of the most frequently quoted tech pundits in the world. Opinion pieces denote the opinions of the author, and do not necessarily represent the views of Digital Trends.
Awesome new GPU architecture..cloud based games and apps for mobile and any device will be great..post PC world may be here sooner than later..
Yes but my point is this allows you to run Windows on anything that will host a the small client. Bootcamp doesn’t work on Android PCs, iPads, iPhones, or Smart TVs. And it is fully supported neither Microsoft nor Apple support Bootcamp.
He was complaining because he was going to have to dump his near-new MacBook Pro for an Ultrabook, and he was going to lose on that investment.
Why?
That is my criticism of your article is that you fail to clarify what is new, instead make pointless claims and offer zero explanation of what this thing is. Is it software, is it hardware?
Ah, NVIDIA, Kepler, boy I kind of thought folks here would know they are a chip company and I did say “server companies supporting it”, did you really need me to say “and these server companes are making servers?”. For the software/service I call it out (Gaikai) and provide a link and call out Citrix.
Ok, my point was that you talk about this guy needing a new computer because he is not smart enough to be able to do the 15 minute install that is bootcamp and then complain aout lack of support.
In this case what does supported even mean? Apple provides drivers and an easy install path and will help if you get stuck, any modern mac is a fully functional windows machine as well.
I just don’t understand what you were trying to say. You lacked clarity. I read through this 3 times before I got the point about the Citrix thing and then realized you were just talking about more remote access software for mobile platforms.
What is different about the hardware platform that makes remote access faster and more efficient than today’s hardware? Numbers? ETA? Basic manufacturing process changes that led to it?
How is the remote software to be different than the other Windows on demand software or other well established remote desktop solutions?
Have you never heard the term “thin client” previous to this? This kind of thing has been around for as long as there have been mainfraims.
The “love apple hardware but hate the os” argument hasn’t been valid since before the intel switch as there were PPC versions of Linux that could be installed on mid 1990′s Macs.
I don’t know a single technicallly inclined person who wouldn’t try to save their buddy a few bucks by just saying “have you tried boot camp” if they were unhappy with OSX.
I miss the good old days when people who wrote about technology cared about and understood it and understood what a tech news reader would care about. Now it’s just “hey look, I went to this conference, this is stuff I saw, I don’t get it, but here you go!”
Wow thin clients have been around as long as mainframes. That explains a lot. Google is your friend.
http://www.thocp.net/hardware/mainframe.htm mainframe 1942
http://en.wikipedia.org/wiki/Thinclient thin client 1993
Listen if you just want to be critical in an attempt to look smarter than I am, feel free.
If you actually want to try to understand this drill down on SMX.
http://www.nvidia.com/object/nvidia-kepler.html
http://www.nvidia.com/content/PDF/kepler/NVDSTeslaKComputeArchMay2012LR.pdf
Which ever you decide, good luck!
Back in the day, the thin client was effectively the terminal you worked at. It had no computing capability, but it had all the bits that make a thin client. They are just implemented in software now. I should have said NEARLY as long , but I am just commenting, not submitting an article for publication, so I am not as inclined to triple check my work. Like a pro should be.
Any way, answer the damned question about your buddy.
Feel smarter?
Nope, I was looking forward to serious answers to questions, instead I get a petulant troll of a writer being precious about an article that was obviously written in about 3 minutes.
Just let it go man, you’re really getting too upset about this. You published a weak article, it happens. Let it go.
There are and have been for many years ways of remoting into your home pc or Mac. Also ANY Mac made since the intel switch can run Windows natively, it is called boot camp and it’s trivially easy to use. You’re a tech writer that doesn’t know about RDC, VNC, virtualization, or basic things like go2mypc?
Ben, see Robs response above.
I did, I was just trying to figure out what was different about this type of thin client as compared to existing solutions. There have been solutions that let you remote into any machine from any machine for as long as there have been computers. This is unnews. This is like talking about the web as a new technology today because a new browser comes out.
@Ben, I don’t think you really get it. Of course terminals and thin clients have been around since the beginning of computing. However, NVIDIA is bringing several things to the table that have not been here before. Firstly, they have a GPU that can render multiple desktops at high resolutions/settings simultaneously — it is extremely power-efficient. This is a huge boon to the server companies. Secondly, they have made the virtualization of the GPU work “properly” with the overall VDI architecture — eliminating software bottlenecks and latencies.
You can’t play games properly over VNC, but can over GeForce Grid. You can’t make money as a server or service company with power hungry dedicated GPUs, but with Kepler the math starts to make sense.
Anyway, the “it’s been done before” comment sounds smart but actually shows ignorance about what is actually being offered.
Thanks for the clarification! I was just trying to figure out what advances this offered over onlive and the related services as well as of course trying to suss out what he meant about the iPad options. I just got annoyed when he started quoting Wikipedia (incorrectly).
Also, I wasn’t saying “it has been done before” I was more ASKING “hasn’t this been done before ?” which is a bit different. His lack of explanation and obvious lack of comprehension of what he was writing about just frustrated me.
Okay, fair enough. True the article didn’t really explain why this time would be any different, and it is also fair to be skeptical generally of any “next big thing”.
Exactly, and his attitude generally showed not just a lack of understanding ofnthisntech on his part, but an arrogance protecting that ignorance. I may have phrased things less than clearly and I definitely came across as more confrontational than I meant, but a writer should be able to answer basic questions about an article without getting snippy or having to resort to Wikipedia as a “see what I mean?”. Plus, Wikipedia??? Really?
I hate to break up a good arguement..but..I need to get in on the cloud side! That way I can make a fortune on all information (what you do..when you do it..how..yada yada!) I don’t know about any of you..but..I like my privacy a little bit more than that! But as a money maker..or data miner..I bet google and the rest will be in on the ground floor!! LOL!
The solutions are from Dell, Cisco, and IBM, and Amazon and all are vastly better than Google with regard to security and privacy. Right now Google isn’t doing this and they will likely develop their own hardware if they do. Gaikai is likely similar to Blizzard with regard to privacy and security so no huge issues but you are right to ask the question.