poslathian - may I encourage caution on spreading resources across too many bases?
I am very pleased to know that you folks investigate alternatives and competition.
I'm very interested in the combination of normal/sane 32-bit processor + extra 'oomph', so Oak, Cypress PSoC 5, Actel, Xilinx, are the biggest interest. After that Fast analogue+DSP, ARM+DSP, and ARM+FPU are interesting.
To respond to llAndy:
Unless their is a very concrete piece of information to suggest significant change is approaching, I disagree with spending any resources to mitigate the specific risk "... just in case something does happen to STM and they lose their access to their chips."
I think there is limited value in spending resources on this without good evidence. Many companies fail because they lose focus, or spend their limited, hence precious, resources on things which bring little value. IMHO, LeafLabs are likely to be agile and flexible enough to react more effectively to concrete threats than hypothetical, or unbounded risks.
1. Apple buy ARM
What do we imagine would happen to STM's Cortex-M3 products?
Will Apple be able to prevent STM making Cortex-M3 for the next few years? I think that's unlikely.
I believe the investment that STM (or NXP, Cypress, TI, etc.) has to put into developing a product line like STM32F/W/L is so large, that they are very likely to have contracts which ensure they have licenses to make Cortex-M3 for several years, or more, into the future.
Would Apple want to prevent STM making STM32F? After all, it's likely profitable to ARM. If Apple did buy ARM, and did, for whatever reason, want to stop STM making Cortex-M3, they may well want to stop all Cortex-M3 licensees, not just STM.
IMHO, Cortex-M3 user will likely have several years to migrate, and a purchaser of ARM might much prefer to continue licensing to existing, and new, manufacturers simply to make a profit from existing investments.
There is a second problem with the premise that LeafLabs can't get STM32F.
If someone else bought ARM, and did want to stop STM making Cortex-M3, IMHO under such a scenario, it seems more likely that they would want to stop several Cortex-M3 licensees, or maybe even all ARM products in a particular sector, and not just STM. So to truly mitigate this risk, the alternative may have be to be non-Cortex-M3 products. So the fact that there are other Cortex-M3 vendors only seems to deal with one of several potential risks. In the absence of other information, I don't believe STM is more likely to stop making Cortex-M3 than any other Cortex-M3 licensee.
2. Intel has been looking for non-PC revenue for many years. Intel used to maker ARM processors. I had a palmtop with one in, in 2002. What is the risk that causes LeafLabs to lose access to STM32F but still have access to other Cortex-M3 products? I don't understand. llAndy, you'd need to expand on this one for me to respond.
3. Microchip did try to buy Atmel. I don't think anyone believed ATmega168/328 was going to stop being available within a couple of years. I didn't, and I don't remember anyone seriously suggesting that would happen. Why would someone like Microchip buy a company with a competing product line (and hence acquire customers and market share), and rapidly kill the products? No reason that I can think of. They would buy the company in order to make more money. It is very hard to do that by killing products quickly. On the same basis, if someone bought STM, I think STM32F would be available for several years, even if the new owner introduced an easy migration path. If a migration path is introduced, its probably best to wait and see how well it works.
I suggest reacting to the problem if or when it becomes real.
This might take a 'crash program'. To reduce the time to react, most companies analyse and experiment with alternatives so that selection of a viable path can be based on concrete experience. This is pretty much a natural business process for companies in this area of business because it is a natural way to evaluate competitors, and develop new products.
I believe Poslathian has explained that LeafLabs do this as a normal part of their activities.
To go further than the normal research and product development activities, it needs to be funded, Where would the extra profit come from? A version of a Maple with an STM32F vs a version of Maple with NXP's LPC17xx? That has risks too.
It would make sense to have alternative sources of STM32F, but that is definitely not what I interpret from llAndy's "Leaflabs should seriously consider an "alternate vendor" just in case something does happen to STM and they lose their access to their chips."
Summary
IMHO is unlikely that any event could happen to STM or ARM, which would cause a problem for LeafLabs which is so urgent and rapid that LeafLabs are better off "ensuring an "alternate vendor" just in case something does happen to STM and they lose their access to their chips".
IMHO, it is more effective to accept their may be a small risk with companies of the scale, longevity and success of STM, and mitigate it by being flexible, agile and informed. All these attributes are likely to be valuable no matter what happens.
If I were looking to spread my risks, I would look at a complementary, but significantly different product like Oak, and put effort into getting those out as rapidly, and cheaply as practical. I'd do that by applying the in-house experience I've got, and avoiding any distraction away from my strongest options, unless there is significant upside.
As I was told, many years ago "Go early, go dirty" - it is better to have a single product, with imperfections into the market, than try to remove imperfections, and risk missing opportunity.
It sounds like STM32F is currently the strongest option for near-term products. Wouldn't it be great, in three years time, to have to wean 200,000 users off an STM32F-based Oak, knowing what to move them towards?
IMHO, continuity of STM32F supply is likely to be a risk worth mitigating much more than STM stopping production of STM32F. Further, I think continuity of supply is likely to be a lower cost risk to mitigate.
