Imagination Technologies IP licensing and execution

[silent_guy]

Everyone "cares", but there's a chicken-egg problem for us to solve first.

I'm really wondering what kind of market this would be. For an IP company to really make money, the market needs to be huge. I said multiple times that smartphones are already reaching the point of getting overbuilt in terms of GPU (not that it will stop anytime soon, mind you), I think we can all agree that nobody is waiting for a smartphone or tablet with ray tracing? So that immediately removes one of the biggest users of IP...

A tad short sighted, but if you're just talking about phone and tablet APs then I can see where you're coming from. We're going to try and build something compelling there.

Yes, there's a huge world of chips out there that aren't phone and tablet and they don't care as much or at all about the instruction set. But still there's the question why, other than legacy reasons, they aren't using ARM?

Is it purely because of price (essential bottom feeder, which sounds a bit harsh) or are there also technical reason? I have no idea about the latter: I've lost count of the number of ARM variants that are out there, but I don't think they are missing an obvious segment.)

I haven't observed cost being the dominant reason for MIPS licensees to take the IP, but those reasons aren't something I can comment on publicly.

It would be interesting to hear eventually.

I've seen to many IP companies die or being gobbled up for next to nothing by Synopsys to stand by my point that it's fundamentally a shitty business. But I hope IMG can make it work.

(As for the stock being down right now: for employees who receive shares, there's nothing more profitable than a highly volatile stock price… as long as it gets back up eventually. )

 

[Ailuros]

I'm really wondering what kind of market this would be. For an IP company to really make money, the market needs to be huge. I said multiple times that smartphones are already reaching the point of getting overbuilt in terms of GPU (not that it will stop anytime soon, mind you), I think we can all agree that nobody is waiting for a smartphone or tablet with ray tracing? So that immediately removes one of the biggest users of IP...

Except maybe Apple itself. As for something being overbuilt at the moment the GK20A GPU in TegraK1 is OGL4.4 and supports FP64 and ray tracing amongst other things.

Yes, there's a huge world of chips out there that aren't phone and tablet and they don't care as much or at all about the instruction set. But still there's the question why, other than legacy reasons, they aren't using ARM?

Is it purely because of price (essential bottom feeder, which sounds a bit harsh) or are there also technical reason? I have no idea about the latter: I've lost count of the number of ARM variants that are out there, but I don't think they are missing an obvious segment.)

It's not as harsh as my comment where I suggest Apple should license its Cyclone CPU right back to ARM in order to show them how you actually develop an efficient ULP SoC CPU. Right next to it comes Qualcomm's Krait.

Don't you find it somewhat weird that those two custom processor families hold the majority of the market and everyone else so far for the greatest majority has been using the extremely power efficient ARM A7 cores? I haven't seen ONE mobile developer to date to "celebrate" the existence of an A15 core in a smartphone. Yes they will get there soon but if it takes N hw revisions to get there, there is something rotten in the house of Denmark.

It would be interesting to hear eventually.

I'm not sure what they're selling at exactly, but from the so far public IMG financial results it sounds like their margins are even slightly higher that those at ARM but I could be wrong there. In any case so far - since they are still separating MIPS results - it seems like the CPU IP department is just a tiny bit over the break even point.

I've seen to many IP companies die or being gobbled up for next to nothing by Synopsys to stand by my point that it's fundamentally a shitty business. But I hope IMG can make it work.

(As for the stock being down right now: for employees who receive shares, there's nothing more profitable than a highly volatile stock price… as long as it gets back up eventually. )

10 years ago or so when they were just a couple of hundred employees filing continious losses with a share price of 10-15 pennies concerns like the above would be way more justified. We're at the point where around 1b smartphones are sold worldwide on an annual basis and in the longrun there are prospects for that volume to triple. Would you expect IP to stop making sense all of the sudden in that market? I'd rather say the exact opposite.

This year they'll sell over 630Mio IP cores from the IMG side and about as much from MIPS; since the market is continuing to explode they'll most likely reach their 1b chip goal soon (without MIPS obviously) which I thought was insane when they first stated it.

In reality I'm rather worried about those IHVs that haven't been able to reach as much ground in the ULP market as many expected or wanted to. Maybe it is time for the entire landscape to change and my only other hope is that we won't be any further monopolistic tendencies. And yes I'm that naive and no it's definitely not realistic.

Here's some financial/marketing wash with colourful pictures: http://www.imgtec.com/corporate/pre...f-Year-Results-December-2013-presentation.pdf

 

[tangey]

While I have no clue what IMG is planning, I hope the first IP wave at least will be standalone RTU IP. I can't imagine what I'd need as a user RT in a smartphone.

I think it'll start higher up and work lower down as process size advances, and other things advances. I'm expecting first appearance to be within integrated TV for UI or possibly with an integrate games system. Then down to tablet...then phone. If you can create the market in higher end, non- portable devices, then the hope would be that an RT requirement would start to exist in portable devices. (all the above is pure guesswork).
Apple is one of the few companies that could create a market where none exists.

IMGs RT solution seems to require a lot of memory, which will also be a significant factor in determining relevant target segments.

 

[Lazy8s]

Some graphics effects of a future API might be most efficiently implemented through ray tracing.

Integrating ray tracing hardware as additional/complementary functionality with a future mobile GPU design may make the most sense.

 

[CarstenS]

The point here is if someone like NVIDIA has a hard time to actually beat on any metric so far any PowerVR GPU IP why should I be convinced that Apple with no GPU experience, no related patent portofolio whatsoever can overnight create something that would beat its current solutions by a sizeable margin.

Let me add a thought here: Apple might have a chance since, as it stands, they're probably the only company that can afford to design a processor that can be comparatively expensive to manufacture (I don't need to elaborate why that's the case, do I?). Thus, they can compensate for a lot of things with pure die space.

 

[Rys]

I'm really wondering what kind of market this would be. For an IP company to really make money, the market needs to be huge. I said multiple times that smartphones are already reaching the point of getting overbuilt in terms of GPU (not that it will stop anytime soon, mind you), I think we can all agree that nobody is waiting for a smartphone or tablet with ray tracing? So that immediately removes one of the biggest users of IP...

We're targeting games as a primary market for the technology, which is a pretty big market even in single digit percentage terms for portable devices. You say nobody's waiting, but I'm not sure that's the best way to measure potential success. Nobody's really waiting for a bunch of things that vendors are delivering to customers, but they're pleased when they show up nonetheless.

The chicken-egg problem is mostly a content one. We need to prove to game vendors that it's easy enough to integrate software support for the technology if it is there and that there's a compelling reason to do so, and I don't mean lock, stock and barrel replacement of traditional rasterised graphics. The future is a complementary one, where the "RTU" augments the GPU rather than pushes it completely out of the socket.

Yes, there's a huge world of chips out there that aren't phone and tablet and they don't care as much or at all about the instruction set. But still there's the question why, other than legacy reasons, they aren't using ARM?

Is it purely because of price (essential bottom feeder, which sounds a bit harsh) or are there also technical reason? I have no idea about the latter: I've lost count of the number of ARM variants that are out there, but I don't think they are missing an obvious segment.)

It would be interesting to hear eventually.

Just from a purely objective level it's not much of a stretch to think that all of the common sense reasons pretty much apply. MIPS customers might variously believe we're better at one or more of: PPA, support, integration, commercials...

The PPA reasons are increasingly true, and they're the most important. It's not that ARM have gaps in the segments they address, more that their products aren't unbeatable. At the very least the competition for ARM is good for the consumer.

I've seen to many IP companies die or being gobbled up for next to nothing by Synopsys to stand by my point that it's fundamentally a shitty business. But I hope IMG can make it work.

Me too. Objectively speaking, we have very competitive cores right now, plus a roadmap that doesn't tie the customer to certain design choices our direct (and indirect) competitors are fixing into their IP. I'm biased of course, but I believe it all stands up objectively.

(As for the stock being down right now: for employees who receive shares, there's nothing more profitable than a highly volatile stock price… as long as it gets back up eventually. )

Sooner rather than later please!

 

[tangey]

(As for the stock being down right now: for employees who receive shares, there's nothing more profitable than a highly volatile stock price… as long as it gets back up eventually. )

Many employees get share options that are only exercisable if the share has reach x price by y date. Those that got such options at £4 or £5 are looking too pretty right now, for example.

SP volatility isn't necessarily good for employees

 

[silent_guy]

Many employees get share options that are only exercisable if the share has reach x price by y date. Those that got such options at £4 or £5 are looking too pretty right now, for example.

SP volatility isn't necessarily good for employees

All big companies in Silicon Valley switched from giving out stock options to restricted stock many years ago. Even for companies like Facebook, only the first 100 or so employees were given stock options and the rest had to settle with RSUs. Stock options are basically a thing of the past except for the earliest start-ups.

This is better for the company (after tax rules were changed), better for stock holders (since the amount of RSU given is usually much lower than the amount of options), and better for the employees if the stock stays flat or goes down. But it sucks for the employees if they planned on getting rich quick due to high volatility. It definitely reduced the incentive of employees to staying with their current employer, because they're not leaving as big a potential pay-off on the table as they used to.

Either way: it doesn't change the point that high volatility is beneficial as long as it's not only on the way down: the amount of RSUs received from year to year is usually similar in dollar value so see-sawing stock price can result in very handsome returns even if the long term average price is flat.