Bye bye Keil 166 and 8051 (??)

I think IBM should be given some industry award for burying them, then.

Is there a good reason to look beyond the obvious?

• True cliche: "ARM is the 32-bit 8051"
• Thousands of customers representing billions of dollars in volume are looking to migrate upwards
• 8051 vendors using a low-cost (royalty free?) core not keen to push ARM, since they have to pay more in wasted IP fees to use it
• Since there is not much "push", ARM seeks to create "pull"

Wouldn't you love an ARM compiler from Keil that could import an 8051 project, retarget it to a specific ARM part, and "just go"?

I concur with this.

Yes that would be nice. But I think what Charlie is getting at is that regardless of the push/pull to larger processors, the need for cheap 8bit

Bo

Didn't IBM recently open-source the Rational "process"?

Talk about a poisoned chalice :->

cheers, Rich,

I'm hesitant to say this, because of course creative accounting can prove anything, but I'd say that Keil's '51 tools are the "gold standard" in the field and very profitable. I'd say ARM's accountants would not want them to kill those products. However - now they own Keil, they can make sure that the Keil ARM offerings are:

a) up to date with "insider" support for the latest ARM developments before those developments are publicized,

b) optimized by direct contact with the engineers who work on the core, and

c) designed to provide an easy migration path from '51 to ARM.

Hence, ARM will be perfectly positioned to direct upward-migrating customers presently using Keil's '51 tools towards ARM-cored parts.

Tool vendor acquisition is always a worrying event, but I think this particular one will not involve any massive upheaval.

There are no GCC tools for 8051. I think that may have been your problem.

Charlie wrote On 11/02/05 16:26,:

You are correct. Its not good.

Lets face it, GCC and other freeware offerings have beat the compiler industry down to a shadow of its former self. As with any such dying business, the independents are simply deciding that it is not profitable to remain independent. The CPU makers are willing to subsidize compiler writing in a captive division, but not independent makers.

This is actually fine. These are market forces. These captured companies divest themselves of CPU support other than their new parents quite quickly. Metrowerks actually dropped non-PowerPC support before being bought by Freescale, perhaps to make themselves more attractive to Freescale.

Judging by the track record, Keil will drop non-arm support sooner or later, my bet is sooner. To paraphrase Bruce Springsteen, these compiler companies are going, boys, and they ain't coming back.

By the way, this happened to our company's processors a long time ago, the Sparc. Today there is virtually only our own compiler and GCC.

What do you do with anything that is valuable that you don't need? You sell it. Metrowerks sold off their x86 technologies. Keil will sell their non-arm stuff.

In article , Scott Moore writes

So where to we go for quality compilers? Gcc ain't it.

Slightly different case - x86 was worthless to Metrowerks, but the Keil C51 user-base IS valuable to ARM. My prediction: Sell-it, no; milk-it, yes.

-jg

Chris Hills wrote On 11/03/05 13:01,:

Same place you go for quality operating systems and quality word processors :-) :-)

Seriously. Its hard to beat free. People made a choice, and it seems to be that the majority want "free", and don't care about quality.

Compilers are not easy to create. Its easy to create easy compilers, which tends to make novices believe that compilers are a non-issue. But making a first class compiler used to be equivalent to making a moon rocket. Now, the only ones who really have the incentive to invest this kind of time and money are the CPU makers themselves.

If I had to guess, I would say the future is that you will pick a CPU by the quality of the tools the vendor provides for it, which is pretty much what I suspect ARM, Freescale and others had in mind when they bought those companies.

Our company (sun) is a good example. We make first class compilers. Not super shiny and neat IDEs, but our compilers wring the last %1 of performance from the Sparc CPU. Some groups here at sun use the GCC compiler anyways, just because they prefer to be %100 compatible with what everyone else is using. However, GCC is never going to achieve explotation of all the features of the Sparc CPU that our own compiler group is capable of. There's no way GCC could. Our compiler group gets the silicon when it is still warm from the fab. They even have a say in what features the next Sparc will get.

Chuckle. You slay me, Chris, really.

To Franklin?!

Now then - he's facing a nasty, painful change of business model.

Maybe use all the expertise developed certifying compilers to offer a certification service for gcc? Or did Cygnus do that?

Come to think of it, how do ACT manage to sell the Gnu Ada Compiler?

cheers, Rich.

Hello Scott,

Exactly. In my experience tool quality and tool cost has been a major factor in deciding which product gets designed in. There is a reason why TI lands so many design-ins with their DSP. Their tool set isn't free but reasonably priced. An engineer who wants (and needs) a tool set has an easier time walking into his boss' office with that cap ex request if the number at the bottom of it isn't in the high four digits.

Regards, Joerg

We just added the tagline

to our website a month or so ago, looks like we may have amend it to read "..in the USA and the World!" soon :-O

I do look at this deal as a possible opportunity for ImageCraft. People have always bug us for a 8051 toolset, but it didn't make too much sense before. With Keil being bought up, it may make some marketing sense for us to do so now. Then we will cover most of the popular 8 bits embedded targets except for PIC16 (AVR, HC08/11/12, even PSoC)

There may be a greater value in killing it off.

If killing Keil's 8051 compiler increases ARM's market share by just a couple of percent it will be worth it for ARM. After all, a couple of percent of a huge market is still a nice big number AND it doesn't mater who's ARM processor you buy, ARM still gets license fees.

Another interesting possibility is selling the 8051 toolset to a specific chip manufacturer. Then watch that manufacturer modify it's 8051s and compiler so that the compiler works only with its varient.

Regards Sergio Masci

LOL :-)

Go for it, Richard!

Meindert

What do you folks think of the present and future of Altium-owned Tasking?

BTW, they do support 8051 and 166. (See

Wayne

----== Posted via Newsfeeds.Com - Unlimited-Uncensored-Secure Usenet News==----

----= East and West-Coast Server Farms - Total Privacy via Encryption =----

In article , diggerdo writes

I don't think that would ever happen. It's a long story.