NVIDIA Enters the Licensing World: My Quick Analysis

Subject: General Tech | June 19, 2013 - 09:51 PM |
Tagged: Volta, nvidia, maxwell, licensing, kepler, Denver, Blogs, arm

Yesterday we all saw the blog piece from NVIDIA that stated that they were going to start licensing their IP to interested third parties.  Obviously, there was a lot of discussion about this particular move.  Some were in favor, some were opposed, and others yet thought that NVIDIA is now simply roadkill.  I believe that it is an interesting move, but we are not yet sure of the exact details or the repercussions of such a decision on NVIDIA’s part.

The biggest bombshell of the entire post was that NVIDIA would be licensing out their latest architecture to interested clients.  The Kepler architecture powers the very latest GTX 700 series of cards and at the top end it is considered one of the fastest and most efficient architectures out there.  Seemingly, there is a price for this though.  Time to dig a little deeper.


Kepler will be the first technology licensed to third party manufacturers.  We will not see full GPUs, these will only be integrated into mobile products.

The very latest Tegra parts from NVIDIA do not feature the Kepler architecture for the graphics portion.  Instead, the units featured in Tegra can almost be described as GeForce 7000 series parts.  The computational units are split between pixel shaders and vertex shaders.  They support a maximum compatibility of D3D 9_3 and OpenGL ES 2.0.  This is a far cry from a unified shader architecture and support for the latest D3D 11 and OpenGL ES 3.0 specifications.  Other mobile units feature the latest Mali and Adreno series of graphics units which are unified and support DX11 and OpenGL ES 3.0.

So why exactly does the latest Tegras not share the Kepler architecture?  Hard to say.  It could be a variety of factors that include time to market, available engineering teams, and simulations which could dictate if power and performance can be better served by a less complex unit.  Kepler is not simple.  A Kepler unit that occupies the same die space could potentially consume more power with any given workload, or conversely it could perform poorly given the same power envelope.

We can look at the desktop side of this argument for some kind of proof.  At the top end Kepler is a champ.  The GTX 680/770 has outstanding performance and consumes far less power than the competition from AMD.  When we move down a notch and see the GTX 660 Ti/HD 7800 series of cards, we see much greater parity in performance and power consumptions.  Going to the HD 7790 as compared to the 650 Ti Boost, we see the Boost part have slightly better performance but consumes significantly more power.  Then we move down to the 650 and 650 Ti and these parts do not consume any more power than the competing AMD parts, but they also perform much more poorly.  I know these are some pretty hefty generalizations and the engineers at NVIDIA could very effectively port Kepler over to mobile applications without significant performance or power penalties.  But so far, we have not seen this work.

Power, performance, and die area aside there is also another issue to factor in.  NVIDIA just announced that they are doing this.  We have no idea how long this effort has been going, but it is very likely that it has only been worked on for the past six months.  In that time NVIDIA needs to hammer out how they are going to license the technology, how much manpower they must provide licensees to get those parts up and running, and what kind of fees they are going to charge.  There is a lot of work going on there and this is not a simple undertaking.

So let us assume that some three months ago an interested partner such as Rockchip or Samsung comes knocking to NVIDIA’s door.  They work out the licensing agreements and this takes several months.  Then we start to see the transfer of technology between the companies.  Obviously Samsung and Rockchip are not going to apply this graphics architecture to currently shipping products, but will instead bundle it in with a next generation ARM based design.  These designs are not spun out overnight.  For example, the 64 bit ARMv8 designs have been finalized for around a year, and we do not expect to see initial parts being shipped until late 1H 2014.  So any partner that decides to utilize NVIDIA’s Kepler architecture for such an application will not see this part be released until 1H 2015 at the very earliest.


Sheild is still based on a GPU posessing separate pixel and vertex shaders.  DX11 and OpenGL ES 3.0?  Nope!

If someone decides to license this technology from NVIDIA, it will not be of great concern.  The next generation of NVIDIA graphics will already be out by that time, and we could very well be approaching the next iteration for the desktop side.  NVIDIA plans on releasing a Kepler based mobile unit in 2014 (Logan), which would be a full year in advance of any competing product.  In 2015 NVIDIA is planning on releasing an ARM product based on the Denver CPU and Maxwell GPU.  So we can easily see that NVIDIA will only be licensing out an older generation product so it will not face direct competition when it comes to GPUs.  NVIDIA obviously is hoping that their GPU tech will still be a step ahead of that of ARM (Mali), Qualcomm (Adreno), and Imagination Technologies (PowerVR).

This is an easy and relatively painfree way to test the waters that ARM, Imagination Technologies, and AMD are already treading.  ARM only licenses IP and have shown the world that it can not only succeed at it, but thrive.  Imagination Tech used to produce their own chips much like NVIDIA does, but they changed direction and continue to be profitable.  AMD recently opened up about their semi-custom design group that will design specific products for customers and then license those designs out.  I do not think this is a desperation move by NVIDIA, but it certainly is one that probably is a little late in coming.  The mobile market is exploding, and we are approaching a time where nearly every electricity based item will have some kind of logic included in it, billions of chips a year will be sold.  NVIDIA obviously wants a piece of that market.  Even a small piece of “billions” is going to be significant to the bottom line.

Source: NVIDIA

TSMC's 28nm process is going to be around for the long haul

Subject: General Tech | March 20, 2012 - 01:12 PM |
Tagged: TSMC, nvidia, amd, southern islands, kepler, 28nm, maxwell, llano

TSMC's 28nm process has been in the news for a long time, sometimes this was a good thing but more often it was not.  Back in May of 2009 the first announcements of TSMC's brand new 28nm process hit the news with major production slated to start in early 2010.  That didn't happen on time, much to several companies dismay as Josh unhappily discussed towards the end of 2010.  This set a trend for TSMC's 28nm process for a while, for instance AMD did not quite meet their promise of readily available 28nm GPUs in 2011, though a late December launch for the HD7970 did meet the spirit of the agreement.  The delays and issues on TSMC's 28nm lines had a variety of causes, perhaps one of the worst being TSMC's overly optimistic attitude about their production capabilities especially when AMD had a surprise for them.  Add to that the long line of woes during the development and production of NVIDIA's 28nm Kepler GPU as well as the recent shutdown of the production line, and you can see why TSMC's 28nm process has spent a lot of time being maligned in the news.  It almost makes you forget about the 40nm process woes, but that is ancient news.

All that effort is not going to waste as DigiTimes reports that TSMC is planning on expanding their 28nm capacity this year and expects that process to account for 10% of their 2012 revenue.  The next question on most peoples minds is the progress on TSMC's 22nm process which in 2010 they announced would be ready by Q3 2012, something which NVIDIA's Maxwell team is probably anticipating with great anxiety.


"With current capacity for 28nm processes filled up, Taiwan Semiconductor Manufacturing Company (TSMC) is likely to expand the leading-edge process capacity later in 2012, according to industry sources.

TSMC reportedly is running at full capacity at its 12-inch fabs due to strong orders for 28nm as well as 40nm and 65nm designs. In order to avoid orders to rivals such as United Microelectronics (UMC) and Samsung Electronics, TSMC will have to speed up the pace of its leading-edge capacity expansion in particular its 28nm capacity, the sources said."

Here is some more Tech News from around the web:

Tech Talk


Source: DigiTimes

As expected NVIDIA's next generation GPU release schedule was a bit optimistic

Subject: General Tech | July 8, 2011 - 12:37 PM |
Tagged: nvidia, amd, 28nm, kepler, maxwell

TSMC's 28nm wafer yields are having a negative effect on NVIDIA's scheduled release of their next generation of GPUs, no matter what the PR coming out of NVIDIA might suggest.  That news is coming from graphics card manufacturers who were hoping to release cards but have since seen NVIDIA's scheduled releases delayed by a year.  While it may be true that TSMC is partly to blame for the delay there is also talk about the chips performance being lower than was expected and is needed to challenge AMD.  The news for NVIDIA gets even worse as DigiTimes confirms that AMD is still on schedule with it's 28nm chips.  This may seem like a bit of deja vu, as we saw similar production problems from TSMC's initial 40nm chips; though that effected both major GPU makers more or less equally.

nvidia jensen.jpg

"Despite Nvidia CEO Huang Jen-hsun previously saying that the company is set to announce its new 28nm GPU architecture at the end of 2011 and 22/20nm in 2013, sources from graphics card makers have pointed out that Nvidia has already adjusted its roadmap and delayed 28nm Kepler and 22/20nm Maxwell to 2012 and 2014.

The sources believe that the delay is due to unsatisfactory yield rates of Taiwan Semiconductor Manufacturing Company's (TSMC) 28nm process as well as lower-than-expected performance of Kepler.

TSMC originally expected its 28nm capacity at Fab15 to be available in the fourth quarter of 2011 and was set to start pilot production for its 20nm process technology in the third quarter of 2012.

However, TSMC's other major client Qualcomm, currently, still has not yet adjusted its 28nm process schedule and is set to launch three new products, 8960. 8270 and 8260A using dual-core Krait architecture in the fourth quarter of 2011.

Meanwhile, AMD will follow its original schedule and enter the 28nm era in the first half of 2012. The company's next-generation graphics chips Southern Island as well as Krishna and Wichita processors, which will replace the existing Ontraio and Zacate processors, and will all adopt a 28nm process from TSMC."

Here is some more Tech News from around the web:

Tech Talk

Source: DigiTimes