Much ado has been made since Ouya was unveiled and in some respect this has been done to its main processor component the Nvidia Tegra 3 processor. There have been many positive and negative comments but I will stay away from what controversy in some part as a lot has been voiced I’m here to give you a little insight into the interesting parts of this component of Ouya.
Where to begin…
The Tegra 3 is powered by 5 Cortex A9 ARM processors and… I bet most people don’t really know what that means. Let’s go back to the beginning.
Tegra is a part of the ARM series of processors which are very popular in mobile devices due to their high performance per Watt used making them perfect for battery powered devices.
ARM stands for Advanced RISC Machine… I bet you don’t know that that means either. Let’s go back further.
If you are reading this on a desktop PC there is a very good chance that you are using a machine running an Intel or AMD processor in the x86 range of processors. These chips are considered Complex Instruction Set Computers which basically the processor can accept literally hundreds or thousands of different instructions from the program to execute on a wide variety of functions. This is a design philosophy that stretches back to the very beginning of modern day electrical computing of the late 1940’s. /the idea that every function of the machine should have its own entirely unique instruction to execute came about to keep programming simple for the programmers as they could add new functions to the hardware as the software required.
While it makes it easy for programmers it does make it difficult for the hardware engineers who need to build separate hardware for each function which means that the computing devices get more complex as time goes on. There where machines in the 1960‘s and 70‘s that tried to overcome these limitations but it wasn’t until the 1980’s that a real push was made to reduce the computing requirements on engineers and programmers.
It turned out that if one was to analyze the instructions that most programs where using, only about 60 to 70 instructions where being used for the vast majority of the time, this meant that if a processor could be made to support only the most commonly used instructions and have the software conform to this limitations then processors could become significantly more efficient. This would be called a Reduce Instruction Set Computer. Any functions that where not directly supported could be constructed out of this more limited instruction set to overcome the rare occasions when something wasn’t supported.
It’s a brilliant concept and one that unfortunately never caught on in the desktop/laptop space due to the Microsoft Windows dominance of most computer platforms in the 80’s and 90’s. Despite that some companies did try and break free to make their own standards, Apple, Motorola and IBM did eventually try to build their own standard with the POWER/PowerPC range of processors which while moderately successful for Apple for about a decade eventually have fallen out of favor for various reasons. This is not where Tegra’s adventure will begin.
ARM begins with a British company called Acorn. Back in the early 80’s after making the successful BBC Micro computer they were looking to build a new range of computers to compete with everything else on the market. Having looked at every other processor available they were left unimpressed especially considering they were to try competing with Apple in the graphical interface area. They then set out to make their own processor based on some research they had read called the Berkley RISC project. They figured that if a class of students could build a better processor technology then what was commonly available then they as a successful company could as well.
Thus the Acorn RISC Machine was born, this was abbreviated down to ARM. Acorns computers where a complete failure on the market but the technology they built lived on in the ARM technology. The processors and the technology was spun off in a company called ARM Holdings where it was renamed the Advanced RISC Machine to remove any references from the failing parent company, this is the name and company we have today.
ARM Holdings great difference between itself and other processor manufactures is that it’s doesn’t actually produce any physical processors, it’s has no manufacturing side to it in any way. If you want an ARM processor, the design is licensed off ARM Holdings and then it is up to the licensee to figure out the manufacturing. Because the design is licensed this means that it can be modified to match the need of each customer. This is why Apple can have its own ARM chips and so can Smasung and in Nvidias situations it’s Tegra series of processors. If you are reading this on your phone or tablet there is a good chance you are using an ARM processor.
ARM chips are particularly useful in the mobile scene because the small instruction set allows for some excellent power management as chips can be small and scalable. They were not built for the mobile space but eventually that is where they become a real star.
Nvidia trying to make inroads into the mobile market place due to the dropping sales of desktop top processors and more importantly to them the desktop GPU market decided to start producing a mobile processor. Products like the iPhone, iPad and the Android collective has been eating away at the more profit rich areas of the computer industry and the big players are out to survive by adapting.
Tegra being based on the ARM processor made it compatible with the Android OS and later on the Windows 8 RT standard; it was also coupled with a set of GeForce Go! processors to provide the graphics. These GeForce chips have been produced by Nvidia since the early 2000’s for the mobile market but failed to gain any real major traction in terms of sales due to a lack of what many considered required functionality on mobile products.
The idea was simple provide a processor that can do it all rather than rely on other companies to provide all the other components and just hope for the best. These are commonly referred to as a System On A Chip or SOC design, a total computing system on a single processor.
As such the Tegra 3 chip in Ouya is a great example of this in action. Ouya Inc can purchase a single processor from Nvidia and have all the major component of the system taken care of without having to source multiple partners and organize many supply chains. What you have in your machine is the condensed outputs of many technologies from the last few decades sliced into a single fairly powerful product.
What we have today
So Tegra 3 is now considered Nvidia’s mid-range processor in the mobile space with the release of the newer Tegra 4 processor earlier this year. Featuring 4 main processors running at 1.6Ghz when running all 4 simultaneously or 1.7Ghz when only one is active it is a fairly powerful chip in terms of its execution performance. Funnily the CPU has more potential performance than the CPU of the Wii U console (for the most part it doesn’t due to OS/Memory bandwidth) and is a very competent core to the system. It’s not the fastest processor available in the mobile field and it is wise to not compare it to high end PC parts but this goes for all mobile parts. However for what it set out to achieve in terms of cost and performance per Watt it is an astoundingly good part.
Tegra 3 also contains an additional CPU core that runs at 500Mhz which in a mobile product is used to maintain a low power state but in terms of run time performance I would wager that this is of little to no use in Ouya’s everyday performance and usage.
Graphics wise the system has a GPU core consisting of 12 pixel shaders running at 520Mhz which is the weaker part of the system personally. At that speed it can render a Xbox 1 (not Xbox One) game at the same frame rate at 1080p but little additional in terms of effects and even then this is dependent on memory bandwidth. It is because of this that Ouya is better suited at simpler visuals rather than trying to push loads of screen based effects. This is set to improve with future Ouya revisions but I will cover that later.
When it comes to video performance the Tegra 3 is in the realm of most modern day ARM SOC designs. Program code is done on the CPU, the GPU handles all the graphics and a separate set of processors handle the video and audio. Decoding video in software can be costly to battery life and potential performance, having dedicated hardware to take care of this alleviates this issue by offloading it to chips designed to decode these streams in as little power as possible. It is because of things like this that you can get 200 hours+ audio playback on an iPad rather than doing via software and only getting 20-30 hours.
On Tegra 3 & Ouya these embedded systems are fairly powerful to say the least. It can decode a 40Mbit H.264 video stream without an issue. To put that in comparison, Ouya has enough power to decode and display a top of the line Blu Ray video, this might be excessive but it is possible. The reason we are seeing performance issue on video playback apps like XMBC at the moment is simply because the video is being decoded on the main ARM processors rather than these dedicated processors. There are various issues why this is done but I will not go into it today.
As a whole the current technology in Ouya is very impressive in terms of performance but there is definitely room for improvements… speaking of which…
The future of Tegra and Ouya
It has been confirmed many times over that Ouya is set to be upgraded on a yearly basis to keep up with the modern technology available at the time of production, this means that on a yearly basis Ouya will be boosted to the latest Tegra chip available or maybe even another range of processors but we will stick to Tegra for the sake of this post if just for the fact that Nvidia is a big supporter of the Ouya product both morally and financially.
Nvidia’s current top of the line processor is the Tegra 4 and it’s packs a fair punch compared with the predecessor, not so much on the CPU side which gets a minor upgrade in terms of clock rates and performance per clock cycle but much more on the GPU side. Instead of 12 GPU cores there are now 72 making for a big jump in terms of the on screen impact. Tegra 4 basically addresses some potential complaints that some users had put towards the platform, its CPU was more than enough to keep most people happy but getting fantastic visuals at the same time seamed unattainable until this upgrade. You see most of these mobile processor vendors will tout “console quality visuals” similar to Xbox 360 and Ps3 but rarely deliver anything close.
Most are wagering that this is the chip that will go into Ouya V2 but I’m hoping that it is not as there are other options. When Ouya was first announced in October last year the T3 was still a fairly decent chipset even though the T4 was just around the corner, now that Ouya has yet to make it to market yet in a major way I would wager that Ouya Inc isn’t crazy enough to revise a product after being on shelves for only a few short months.
I’m going to guess that Tegra 4 will be skipped in favour of the Tegra 5 processor which will be freely available in less than 12 months and that is an exciting concept.
Tegra 5 sees Nvidia ditching its current mobile GPU line in favour of Kepler based graphics processors; these are the same kinds that power their latest desktop GPU’s only at a much lower clock rate and with fewer cores to work with. If they do in fact go this route then it will be enough to put Ouya within the same ball park as the Xbox 360 and Ps3 in terms of graphics capabilities even if slightly slower. There is a leaked video of Nvidia demonstration Battlefield 3 running on this processor which is just below here which gives you an idea of the performance that will be capable on this platform if not next year but the year after. The video isn’t very clear but you get the basic idea.
Many who are complaining about the low specs today will probably be silenced very quickly when these fast chips are more abundant.
Going even further forward is Nvidia’s project Denver which is their quest to make an ARM chip that is capable of matching a modern desktop processors, this is still a few years off and one has to wonder if such a chip could be used in Ouya and keep the same form factor and price, it’s hard to speculate and because of that I shall not.
We do know that Nvidia is committed to its Tegra platform both in terms of 3rd party manufactures like the Surface RT and Ouya also it is own Project Shield is proof positive that they have confidence in the platform. So it’s up to Nvidia and Ouya to determine exactly what we get in future.
And to conclude
I wrote this for everyone so that they could at the very least appreciate a little of what goes into the Ouya and the technology under the hood. It’s also intended to show you that there is a great potential future for the platform at the very least in terms of new more powerful and flexible technology. The future is interesting even if it is not very clear.