by Kris_Tuttle on February 11, 2010
I avoid talking about every technical burp and tweek we review from the major technology firms including Intel, AMD, Nvidia, and the like. (Same goes with every ebb and flow of performance ratings and twists and turns of patent disputes.)
However the recent release of Nvidia Optimus tecnology is another small reason we prefer Nvidia to companies like Intel and AMD in the computing space.
There are several pages of technical description of Optimus available (this is a good one) but we’ll net it out as an elegant way to offer a computer with the power of a discrete GPU engine that is only used when it is needed, thus saving lots of power and helping to lessen the big tradeoff between speed and power consumption which bedevils mobile computing.
The most familiar analogy is the all-wheel drive in cars. In the old days most all cars were 2WD but for a very few 4WD like the Jeep. The drive train was permanent, there was no switching. Then cars started to come with the ability to lock and unlock the 4WD feature when the car was parked and you had the time and willingness to change the hub settings on the wheels. This got better with a simple lever but in most cases you still had to stop the car.
Now 2WD/4WD systems are automatic and unless you do something to prevent it, the car senses and uses whatever drive parameters are needed to maximize performance. The same is true with hybrids that shift between battery and internal combustion engines.
You get the idea.
Similarly it has been possible to manually turn off a GPU inside a laptop but it’s complicated and requires a reboot – so nobody ever does it.
The Nvidia Optimus technology allows a GPU to turn on only when an application that leverages the power of the extra processors is invoked. At the same time it turns off when it’s not needed. So when you are reading email the GPU is off and the power consumption is minimized. When you are done and fire up a video or graphically intensive game the GPU kicks in and delivers the power needed for a great experience.
This sort of “hybrid computing” has been around for some time but what’s important about this is the implementation. An important difference this time is that incorporating the design doesn’t involve extra effort and costs on the part of the device maker. So this will become a standard feature right away. After all any laptop with a GPU and without Optimus is at a major disadvantage in terms of expected battery life which is a big factor in use.
I don’t want to make too much of it but it’s another good datapoint with respect to Nvidia maintaining the leadership in driving the GPU into the fabric of general purpose computing (which BTW is a certainty in our research view, see our other research notes on it.)
Also refer to an earlier post (Nvidia Turns the Crank) for more links.
[Disclosure: The R2 model portfolio has a long position in NVDA.]
by Kris_Tuttle on February 4, 2010
We’ve been on the Nvidia case for some time now. See some of our past research publications to get caught up.
Semiconductor specialists The Linley Group recently posted some details around the performance of the Nvidia Tegra 2 which we will see in numerous devices this year.
(There’s been plenty of coverage of what’s inside the A4 so I’m not sure why they are playing possum on that one. See other sources on the A4.)
There are so many races in the mobile Internet device space (smartphones, netbooks and now tablets) that it will be a free-for-all. Add to it the fact that even the TV in the living room is going Internet this year and the demand for new processor architectures and capabilities will be mushrooming.
These are interesting times not just for Nvidia but also for Intel, AMD, Qualcomm, and ARM. It’s hard to get excited about Intel here given the risks to their growth and margins unless they can change their stripes and embrace the idea there there is more to life than the CPU.
Our positive views on Nvidia are rooted in the fact that computing is shifting from processing words and numbers into not just video but games, rich content and interfaces that behave with physical properties and the deeper content that 3D provides.
It’s also true that consumers are expecting to be able to do many of these fairly complex things at once; all on a hand-held device with hours and hours of battery life that costs around $500.
Architecturally this suggests we move into a hyper-multi-core kind of processing to be able to parallelize the processing to meet the demands all this software has on the hardware.
Performance per watt remains a major issue in the mobile space and the one area where we think Nvidia still has some strides to make. This is an area where ARM shines and they are making some improvements in their graphics processing ability.
Our conclusion is that investors will want to play this disruption in computing architecture and we find Nvidia to be the most attractive in terms of execution, design wins and software momentum.
If Nvidia could understand and really consolidate their position in software the company and the stock would be a huge winner. However based on all our research we don’t quite see that happening. Our IV on NVDA leads us to expect a share price this year of $19-20 this year.
[Disclosure: The R2 Model Portfolio has a long position in the shares of Nvidia and Qualcomm.]
by Kris_Tuttle on December 17, 2009
The last couple of months have been marked by a broad set of positive developments for the GPU makers, especially Nvidia and AMD. Some of the highlights have been:
Application acceleration goes mainstream. – The leading creative application providers, Adobe and Autodesk, are embracing the GPU and driving the requirement for one into space of generalized computing. This actually started with prior generations of the software but we’ll see more of this in products like Adobe Creative Suite Version 5 but even Flash will take advantage of the GPU to improve performance and the experience.
Internet video gets serious. – Video is one of the leading new applications for the Internet today and it is shifting from the short grainy YouTube content of yesterday to high definition, in some cases even 3D, content. This level of digital video demands more processing power to produce and even to view properly. Everybody want’s video, more so than even text and browsing, and it’s another good reason to have a GPU in computer, especially mobile Internet devices.
GPU in the cloud. – Until recently it wasn’t clear if GPU computing power would be available in the cloud. Given the success of generic on-demand offerings from Amazon, Google and others it seemed only matter of time before we would begin to see it. Both AMD and Nvidia are working closely with the major players and we will see GPU processors on-demand and in the cloud pervasively in 2010.
Cloud-based GPU applications are shipping. – Mental Images has announced and is providing Reality Server which represents the high end of digital image rendering in the cloud. Some companies are using it today with their own cloud servers. Application platforms like OTOY are enabling the highest end of gaming even on mobile devices like the iPhone using cloud-based GPU software and acceleration. Here is a link to a Techcrunch post on OTOY back in June of 2009. We also covered a demonstration of OTOY in our short research report on the 2009 Gilder Telecosm Conference. It’s available in the sample publications section of our website.
3D content has arrived. – The cinema industry has embraced it. Avatar is in theaters this week. A big difference today is that the cinema industry creates content in native digital form which means that content can be moved rather than redeveloped into other formats for games, short videos, images and other forms of consumer entertainment content.
Standards are emerging. – Although there is still plenty of bickering over PhysX versus OpenCL and DirectX versus others, these technologies are settling into mainstream software like applications, browsers, and operating systems. In cases were two important standards make it, it looks like they will both be supported. This makes it possible to deliver high-end graphics and 3D applications to a large market.
We expanded on the mobile 3D theme further in a report published yesterday by GigaOM called: 3-D Untethered: A Look at Mobile 3-D Technology (GigaOM Subscription).
On top of all the industry adoption Intel settled with AMD and paid them $1.25B, scrapped their Larrabee-based plans to enter the discrete GPU market and is under further FTC investigation for trying to screw Nvidia.
Put simply the industry is shifting to video, 3D, and mobile and right now this leaves Intel mostly out of the action. It’s great news for Nvidia, AMD, ARM, Imagination Technology, Qualcomm and a host of other semiconductor makers.
When it rains, it pours.
[Disclosure: The Research 2.0 model portfolio has positions in both Nvidia and Qualcomm.]
by Kris_Tuttle on December 7, 2009
Just as we published a research note highlighting some key takeaways from the recent Gilder Telecosm, we see the news that Intel has suspended their immediate plans to enter the discrete GPU market. From the day we published our Nvidia note on January 29th “Transition is Opportunity for Nvidia” there as been a measurable increase in the rivalry between Intel and Nvidia over the increasingly-important GPU market. Since we have been highlighting Nvidia as the leading company we’re pleased to see the news but it’s also just a blip on what is becoming a monster technology growth and investment area.
In fact our note highlights the fact that the RealVR and Cloud Computing themes are already converging. Taken together this represents probably more than 100% of technology market growth in the next decade (although Digital Power and Biological IT will help too.) Inside this new mega-theme we include all forms of consumer and enterprise cloud computing and the mobile Internet.
Intel still dominates the integrated graphics chipset market so the news is more interesting for AMD and Nvidia. AMD now is unique in offering both mainstream x86-based CPU options and their own quite powerful GPU chips as well. Nvidia leads in the mobile and embedded space by a wide margin and has been racking up design wins with Tegra and the ION platform. Both AMD and NVDA are working with partners to bring GPU computing into the cloud in a serious way. We are seeing early applications of this technology already.
Although our RealVR Cloud observation steals the show in our Gilder Telecosm update there are other important implications including: much better prospects for investments in alternative energy, particularly Digital Power, now that the unrealistic expectations have been burned off, and more disruption is in store for the very large and stodgy education market.
[Disclosure: At the time of this writing Research 2.0 holds shares of NVDA in their model portfolio.]