Taipei (Taiwan) - In very simple terms, all chip manufacturers have the same business goal : To sell more chips every year. If your core market limits your growth, you will have to look for new markets. Intel has recently done this with its Atom processor, which targets low-cost computing devices and handheld computers. Nvidia’s Tegra hits the same market segments and expands the company’s reach into a new market that currently has a demand of more than 1 billion processors per year. It could be a game-changing move for Nvidia - not just in terms of growth opportunity, but also in its rivalry with Intel.
It is known that Nvidia organizes its product families under very few umbrella trends and new brands typically indicate the launch of a major new product. The company started out with its GeForce line of graphics cards (which also include nForce chipsets) and eventually extended its strategy with its professional Quadro products. In late 2006, the company launched the Tesla high-performance computing line and now we have a fourth major brand, Tegra. And at least from what we have seen so far, this could be a big one for Nvidia.
What is Tegra ?
To clear out some confusion, let us first stress that Tegra is not a CPU. Neither is a GPU or a combination of both with one part dominating the other.
Instead, Tegra is a "system-on-a-chip" (SoC) or "computer-on-a-chip" (CoC). Tegra consists of an ARM11 CPU core, a GoForce (renamed into GeForce ULV) GPU, an image processor (digital camera support), a HD video processor (PureVideo for handhelds), memory (NAND Flash, Mobile DDR), a northbridge (memory controller, display output, HDMI+HDCP, security engine) and a southbridge (USB OTG, UART, external memory card SPI SDIO, etc).
In short, Tegra includes the whole shebang : CPU, graphics and what you traditionally find on a motherboard are squeezed onto a single silicon die. What is particularly impressive about this device is the fact that this chip measures just 144 mm2, which is smaller than a dime and about one quarter the size of the upcoming GeForce graphics chip, which measures 576 mm2, according to our sources.
Tegra arrives in three versions with different markets in mind. Nvidia ditched the suffix "CSX" at the last moment, and decided to name its mobile computing parts "600" and "650". Strangely enough, Nvidia decided to keep the name "APX" for its cellphone chip, which is called "APX 2500". So, the complete product family at launch includes the Tegra APX 2500, Tegra 600 and Tegra 650. And yes, the lowest performing part has highest number. Go figure.
The (Tegra) APX 2500 was introduced in February 2008 in Barcelona at the Mobile World Congress. The chip is clocked at 600 MHz, and targets applications with FWVGA display resolutions (854 x 480 pixels). This version is a sub-1 watt part and is rated at a power consumption of 0.6-1 watts. Considering the fact that this chip is capable of running HD video decoding and encoding (720p only), the power consumption rating is impressive. Not surprisingly, the APX 2500 targets cellphones and PDAs with low power envelopes. If you consider the chip’s capabilities, its size and power rating, there is no doubt in our mind that this is a potentially game changing piece of silicon that is a kick in the you-know-what of sleeping cellphone giants such as Texas Instruments (TI), Freescale, STMicro, Philips and others.
The Tegra 600 is clocked at 700 MHz and targets devices with SXGA resolution (1280 x 1024). Besides the clock speed, Nvidia expanded the storage capabilities with IDE support for Compact Flash and solid state disks (SSDs). This model aims to attract customers in the GPS segment and automotive applications such as dashboard systems or central entertainment systems. The power consumption is in the 1 Watt range, which is still a very low level. Nvidia demonstrated this chip (Tegra 600 + HDMI, USB, stereo jacks) consuming 1 watt while running 720p video decode through HDMI.
Both the Tegra APX 2500 and 600 integrate low-power DDR memory clocked at 166 MHz.
The Tegra 650 is a beast, the "GTX" of the handheld and sub-notebook world. Nvidia has clocked this part at 800 MHz, while keeping the power consumption in a 2.5 - 4 watt envelope. This investment delivers 1080p video decode at 30 fps (24 and 29.99). The Tegra 650 uses embedded LP-DDR memory clocked at 200 MHz.
Tegra on the desktop, notebook
One of interesting applications for Tegra might be add-on video and the connection of processors to GeForce, Quadro and Tesla boards. Why would Nvidia spend transistors inside a GPU when it can use Tegra instead of NVIO chips and support all video functions while decreasing the thermal design power of a board ?
For GPU computing, future Tegra parts (Tegra 2 is scheduled for 2009, Tegra 3 for 2010) will inevitably include SATA support, which would allow users to put a SSD array directly onto a Tesla GPGPU card and eliminate the need for a motherboard and Intel/AMD CPUs. Indeed, applications of this "jack of all trades" chip with a ridiculously small TDP can deeply impact the already existing desktop, notebook and HPC markets.
Read on the next page : Competition, Intel, disadvantages
At least for now Nvidia is "competing" with Intel in powerpointery wars (see our gallery), while we have note that Intel dies not yet play in the handheld industry yet. That will certainly change soon, but as of now, both companies are just preparing their entry into this market. In a way, Nvidia is going head-to-head with Intel if we look into the MID (mobile Internet devices) category of products, but Tegra has a lot of overlap into other segments.
If we take a look at Nvidia’s 10-K filling from March 21 of this year, Nvidia defines its Tegra competition as follows :
- suppliers of application processors that support PMPs, PDAs, cellular phones or other handheld devices intellectual property such as AMD, Broadcom, Fujitsu Limited, Imagination Technologies Ltd., ARM Holdings plc, Marvell Technology Group Ltd, or Marvell, NEC Corporation, Qualcomm Incorporated, Renesas Technology, Samsung, Seiko-Epson, Texas Instruments Incorporated, and Toshiba America, Inc.
- suppliers of application processors for handheld and embedded devices that incorporate multimedia processing as part of their existing solutions such as Broadcom, Texas Instruments Inc., Qualcomm Inc., Marvell, Freescale Semiconductor Inc., Renesas Technology, Samsung, and ST Microelectronics.
It is obvious that Nvidia is elevating its game and wants to play with the big boys. Tegra is the company’s first advance into a highly competitive market that has a market potential of more than one billion cellphones and more than 200 million smartphones and connected PDAs.
What about Intel’s Atom ?
Make no mistake. Intel’s Atom and Nvidia’s Tegra basically aim for the same market segments. But there are differences and each of these products may go after different segments first, building their defenses over the next several months.
The first obvious difference is that Tegra is a SoC and Intel’s Atom comes in two parts (CPU + integrated graphics chipset called System Controller Hub - SCH). While Atom is a marvel of technology, Tegra is significantly smaller when you add up the size of the processor and chipset. Compared to Tegra’s 144 mm2 package, Atom’s package by itself is 182 mm2 and the SCH runs at an additional 484 mm2. The fact that Atom does not include the chipset and graphics functionality, but has to rely on huge chipset is the main reason why Intel can’t compete in the smartphone space at this time - Atom is just too large. Tegra, on the other hand, is one step ahead of Atom and can aim for iPhone-like devices. Intel says that its 2010 Moorestown chip will be small enough for integration in smartphones.
Another distinct difference is the fact that Tegra has an ARM core while Atom has an x86 core. Atom’s software stack is compatible with Core 2 Duo, which means means that you can install any x86 operating system on an Atom computer. It also means that all those applications that run on your PC today will run on any device with an Atom CPU. Intel believes that this fact will give it an advantage over non-x86 devices (such as ARM), which typically require developers to modify their applications for one specific CPU core. It is unclear how this scenario will play out, but there is no doubt that Nvidia will have to come up with a software strategy for its Tegra processor family.
You can read all the details about Atom in our launch article.
The obvious downfall of Nvidia is the fact that the company has to take its battles very carefully, since it has only 5000 employees and somewhat limited resources. Most of its new rivals have much larger employee counts. While Nvidia may be more nimble on the execution side, these products relies not only on the number of engineers that will support the product over the course of its life, but also on engineers that developer future products. Nvidia will need several hundred FAE (Field Application Engineers) dedicated to Tegra alone, and the software team has to keep up with more than one operating system, especially since the company is dealing with a non-86 CPU.
Supporting Windows Mobile is great, but not supporting upcoming operating systems such as Google’s Android and the dominant player (Symbian) are signs of weakness. Competitors in this field have to support everything that platform vendors want - and we are only talking about cellphones at this point. GPS vendors such as Garmin, Magellan, TomTom might sell millions of Tegras, but Nvidia has to be able to support their specific demands. Pitching Windows Mobile as the platform of choice is not the way to go : At least two out of these three don’t want to hear about Microsoft.
The automotive industry is a different ballgame. In order to qualify as a processing platform for a major car manufacturer, Nvidia needs to dedicate 20-50 engineers for a single project alone. And if Tegra consumes too much of Nvidia’s engineering power, future projects such as PlayStation 4 and NV70 (GT300 ? GT400 ?), NV80, NV90 might suffer.
Also, bear in mind that current Detonator drivers (ForceWare, GeForce) exceed 15 million lines of code, or are about the size of Windows NT 4.0. Microsoft has 120,000 employees, Nvidia has just over 5000. We do not imply that Nvidia suddenly needs to double in size. But Nvidia will have to hire a lot of staff to support Tegra. We are hearing numbers that suggest that Nvidia may have to add about 10% to its workforce.
What does this mean for NVDA ?
Given the design of the part, it is obvious that Nvidia wants to capture substantial market share in the mobile space. And we know that the company, just like Intel, isn’t playing to come in second. There is a clear potential to ship more than 150 million Tegra chips within three years. Just the cellphone market could see more than 100 million units, if this new platform is successful.
It is still too early to say what kind of market share can Nvidia take from Texas Instruments, Philips and Infineon, but if the company captures 10% of global cellphone share with its Tegra APX and a decent portion of mobile computing markets, then we are looking at a market opportunity of somewhere between $1 and $2 billion. Add to that a growing GPGPU business and it is not hard to see that Nvidia may be able to double its current annual revenue of about $4.1 billion within the foreseeable future - if everything goes according to plan.
And that is something that should get every single one of its rivals concerned. Even if Tegra is not just aiming for Intel, it is certainly a sign of Nvidia growing up : Tegra is Nvidia’s first product that will directly challenge Intel and reveal how strong Nvidia has become. Time will tell if Nvidia is strong enough.