Battling with a Heavy Hitters

“We few, we happy few, we rope of brothers…” — Henry V, Act IV, Scene iii

What’s it like to be a underdog? To be a tiny guy, a “other” choice, a alt-weekly opposite a large daily newspaper, a Toledo Mud Hens instead of a San Francisco Giants?

It’s honest work and it pays a bills, though we know you’ll never make it to a large leagues. Does that daunt a engineers and unman a salespeople?

“It’s been a blast!” says Dave Jarmon, Senior Vice President of Worldwide Sales and Business Development during VeriSilicon. He peddles a Vivante GPU, VeriSilicon’s answer to Arm’s Mali or Imagination’s PowerVR. So, a tiny fish in a large pond.

So many of a attention is a duopoly. CPU decisions mostly come down to Arm vs. Intel; a GPU is possibly Mali or PowerVR. Which OS do we want: Linux or Android? Sure, there are other specialty choices, though a leader(s) tend to browbeat a category. How can a tiny actor wish to compete, and what does that tell us about a prospects for new startups?

If Vivante is any indication, a opinion is good. Although shadowy deeply by a shade of a dual better-known GPU competitors, Vivante is though renouned – like, hundreds of millions of units popular. It’s not mostly that shipments strike 9 total and you’re still deliberate a bit player.


Granted, “shipments” is a deceptive judgment when you’re in a IP business. But Vivante has been protected by large players like Texas Instruments and Freescale (now NXP), and a GPU appears in an unholy series of a latter company’s i.MX 6 processors. Those, in turn, have been designed-in to a series of automobile dashboards, creation Vivante a #1 GPU in automotive applications, according to Jarmon, forward of nVidia, ATI, Mali, or PowerVR. Take that, reality.

Nobody starts out formulation to be a tiny business. It usually works out that way. The thought wasn’t always to permit a IP rather than to make chips, but, even then, it was apparent that Arm and Rambus were sucking all a income out of a IP industry. Everyone else was usually scraping by. Maybe if Vivante could obstacle one or dual large pattern wins – maybe even a dungeon phone? – it could strike a large time. Hey, it’s happened before (cf. Intel 8088 and IBM Model 5150).

Cell phones were a bullseye on each IP vendor’s dartboard. Get one or dual of those, and you’re in clover. But that business also started to bifurcate, with PowerVR and Mali holding a lion’s share. VideoLogic (creators of PowerVR) morphed into Imagination Technologies and acquired MIPS Technologies. The outcome was that Imagination and Arm incited into approach competitors, both with a high-end CPU, a high-end GPU, and an collection of midrange support IP, all going after a same sockets. The dual superpowers battled it out, withdrawal no room for eccentric vendors. When elephants dance, a mice had best scurry, lest they be trampled.

In hindsight, Jarmon says a Arm v. MIPS conflict was good for Vivante. “It forced us to differentiate.” As Arm and Imagination hurdled uncontrolled toward faster and some-more large GPUs, it left an opening for some-more medium GPUs for confidence cameras, bureau equipment, and automotive uses. It didn’t harm that Vivante was also utterly a bit cheaper than possibly Mali or PowerVR.
Most business learn Vivante indirectly, by some other device with a company’s GPU inside. NXP, STMicroelectronics, and TI have been good for Vivante in that regard. Developers who’ve used an i.MX6 or a identical MCU though are prepared to pierce adult to their possess SoC have a bent to hang with a Vivante GPU they already know. So, they find out a source and permit it directly. One patron begets another.

“GPUs tend to be flattering sticky,” says Jarmon. That is, once you’ve coded for a sold GPU, we tend to hang with it unless there’s some constrained reason to switch. That works both ways: Mali and PowerVR business tend to be flattering loyal, even if Vivante competence be smaller and cheaper.

The growth roadmap is another cause in a decision. Most first-time business are meddlesome in Vivante’s devise for destiny generations. The usually ones who aren’t, says Jarmon, are startups. They’re economically encouraged – that is, they’ve got no income – so they simply select a cheapest GPU that meets their needs. Follow-on products (if there are any) can worry about themselves.

So… how inexpensive is cheap? Are Vivante’s terms unequivocally that reduction dear than a other guys’? That all depends. IP chartering is a sleazy eel, and a chartering terms mostly seem to change with a tides. This is where Vivante’s primogenitor company, VeriSilicon, comes in.


Like any good wholesaler, VeriSilicon offers apportion discounts. Buy a tiny some-more from Column A and we’ll give we a bonus on Column B. VeriSilicon offers ASIC-design services – in fact, it’s many of their business – so they’re generally skilful during stirring in a IP they’ve already designed. Ironically, VeriSilicon offers PowerVR and Mali GPUs, as good as Cortex processors, right alongside a in-house alternatives. It’s all about patron choice.

What doesn’t change is a simple NRE-plus-royalty chartering indication that’s common to many IP vendors. Charging a one-time upfront cost is candid enough, though business hatred profitable royalties. (This is loyal of any IP vendor, not usually VeriSilicon.) Royalties cut directly into a customer’s margins and, thus, a product’s profit. It’s like handing a square of your income tide to an IP businessman that helped we out dual years ago. But, as any IP businessman (or author, screenwriter, recording artist, or photographer) will tell you, royalties are how we compensate a bills.

How does a destiny demeanour for Vivante, and, by implication, for primogenitor association VeriSilicon and other IP vendors? Jarmon thinks it’s still a good business – usually not one for beginners. He compares it to a progressing days of a EDA industry, when “boutique” vendors with specialized pattern collection got snapped adult by a bigger players, until a marketplace was eventually dominated by usually a really few EDA vendors. The large EDA companies can cost specialty collection some-more attractively than a smaller actor could; they can radically bury a cost within their incomparable portfolio, since a standalone apparatus businessman needs to tarry on income from one product alone.

Similarly, IP vendors should demeanour to merger for survival, says Jarmon. Develop something singular and valuable, and afterwards let someone else take over to make it economically viable. And what forms of IP would that be? Not digital design, he says. That’s too tough to differentiate. His insider tip: Focus on routine technology.

Or, as “Wee Willie” Keeler says, “Keep your eye on a round and strike ’em where they ain’t.”

Posted in
Tagged . Bookmark the permalink.
short link blacxbox.com/?p=14063.