ARM Drives Momentum In Microcontrollers With Keil Acquisition
Hello Mr. Keil and all others, I have been discussing this news the whole day on another forum, but just wanted to check http://www.keil.com before ending the day. I have great respect for you, but I do not think that your 8051 customers will see a happy future if they want to stick with you and 8051. Why would your new owners put any priority on serving the 8051 market? A new 8051 design is a potential lost ARM or cortex design. I understand you have to say "dont worry" to your customers, but we are worried. I of course dont know the details of the deal, but if I were ARM, I know what I would focus on. I felt it personally before with ST7 and Hiware when they were bought by Motorola. I guess it was similar for Metaware when they were bought by ARC. U.s.w. All the best for the future with ARM. You have done a good job for the 8051/C51 world. Best Regards Joseph
A new 8051 design is a potential lost ARM or cortex design I'd have to disagree. Some no doubt think that way, but there's a world of difference between an ARM7 and an 8051. The ARM7 is something like 15x the gate count and die area of an 8051. The ARM7 will typically require 2.5-3x the memory of the 8051 simply due to lower code density. Usually the applications requiring an ARM7 are complex enough to require a change to SDRAM and parallel flash. The wider address and data bus drives up pin count, may force a change to a more expensive package, and may increase board layer count for ease of layout. Embed large amounts of RAM on your chip, and your yield suffers unless you move to expensive high-density RAM processes. Embed flash on your chip, and the extra metal layers similarly cost you. All of these changes and others add to the "ripple effect" increase in cost of a chip with an embedded ARM instead of an 8051. If you don't need the extra power, it would be a mistake to throw in an ARM just because you can -- never mind the ARM NRE and per-piece fees which you don't pay for an 8051 core. The ARM costs significantly more. That's fine when it's solving significantly more problems for you, but if it's mostly idle, it's a huge waste of money for the customers. Even if "the CPU is only $5", then when you sell a million units, that's $5M of lost profit, or likely more like $20M increase in end-user cost. And that can indeed make a difference to the decision of whether or not even to buy the product. When it comes to embedding processors in ASICs, ARM and 8051 really aren't competing in the same market. ARM competes with MIPS and low end PPCs, not 8051s. Lots of people can eventually solve problems by typing long enough, and then they try to justify the "need" for a big processor based on their overly-complex solution. But good engineering consists of using exactly the resources needed to solve the problem, and no more.
A new 8051 design is a potential lost ARM or cortex design I'd have to disagree. Some no doubt think that way, but there's a world of difference between an ARM7 and an 8051. Drew, I think you are making very valid points from a technical point of view. My point is however that this can not be ARM Ltds point of view. Basically, both you and ARM can not be both right at the same time. I think that the big differences you describe in 8051 micros and ARM micros is the problem ARM wants to solve with their new cortex, and buying Keil is going to "accelerate" this as it reads in the pressrelease. So for ARM, who now controls Keil, every new 8051 design is a lost ARM or cortex design, and every new Keil 8051 compiler support agreement renewing is a delayed ARM or cortex design. Good or bad? It depends on what your plans are using ARM micros. My guess - if you plan to stick with 8051 for a while, dont expect ARM to provide you with good tools and support in the long run... Pessimistic? Yes, but I have seen it happen before. Best Regards Joseph
as Robert Storm Petersen (a danish humorist from the mid-1900s) said: "it is difficult to predict, especially the future" I will not venture there. However about one thing that has been in some of the above posts i have the following comment: If the ARM acqusition of Keil is going to make a difference it will be what it should be, namely that the attempts to make the '51 handle gigantic jobs will cease. The '51 projects that truly belong on a '51 will still be '51 projects whatever ARM does with Keil. When Borland killed Brief, CodeWright arose to the challenge, I have no doubt that a good '51 development package will be available for the forseeable future whatever ARM does. Erik
There's still a number of quality 8051 compiler vendors. If ARM kills Keil C51, they won't force a switch to ARM; they'll just lose those sales to another compiler vendor. I don't quite get the cortex thing from a software point of view. We have enough problems designing chips without having to design a semi-custom processor, too. And there's no way a 3rd-party compiler is going to be able to cope with arbitrarily deleted instructions or registers. Sure, it makes the core smaller, but then you have a core with no software support. When Borland killed Brief, CodeWright arose to the challenge And now they've killed CodeWright. Well, perhaps SlickEdit can rise to the challenge.
I also share the concerns regarding Keil's C51 compiler, we are probably using one of the very last versions of this fine compiler. There is no way ARM will let its new acquired assets to continue working on something other than ARM cores. If am wrong then ARM would not have acquired Keil in the first place, would it? Later...
There is no way ARM will let its new acquired assets to continue working on something other than ARM cores. If am wrong then ARM would not have acquired Keil in the first place, would it? No, they wouldn't. I am however moved by the engineering mind separation from the business mind. The show must go on. Discussing pointers and floating point speed and correctness in this forum. The "ARM owns Keil and they (ARM) want to stop you folks from using 8051" news is sort of peripheral noise. Get your project completed, don't tell your boss you're doomed in the long run... Leave the "business stuff" to someone else. I'd better stop here, I'm a cynic by nurture.
ARM can't kill use of 8051s by killing Keil. All they can do is move the users to IAR or Tasking or whoever. Note that this means that they lose money, since they won't be selling any 8051 tools. Not to say that no executive ever makes a stupid decision, but all such a move would do is cut ARM's income. And it certainly won't win them any goodwill with the engineers or executives that suddenly have to switch. People will follow the path of least resistance in such a situation, and that path is to adopt another tool vendor, not completely redesign your hardware from scratch, and particularly not with product from a company that has (a) just annoyed you and (b) just demonstrated that if you do use their stuff, you can't really count on long-term support if it stops being trendy. Keep in mind that ARM (proper) has no products. They just license IP. This means they have lots of cash pouring in but nothing to spend it on. Most companies in that position try to buy other companies for various degress of integration or diversification. ARM's own ARM development tools were pretty weak. They can now replace those with Keil, and perhaps capture some of the market that promptly leaves to buy 3rd-party tools rather than use the ones from ARM.
It is interesting to note that there have been no updates to the C51 product for over a year. I wonder if it makes any sense to renew my subscription.
"It is interesting to note that there have been no updates to the C51 product for over a year. I wonder if it makes any sense to renew my subscription." I heard a rumour that version 8 was due out soon, but I should point out that that was before the takeover.