Arm was always a favourite...
Of anyone doing embedded.
ARM Holdings' chief executive Warren East is stepping down after nearly 12 years leading the British tech success story. East, 52, is retiring on 1 July, 2013, and will be replaced by 45-year-old ARM president Simon Segars, processor core designer ARM confirmed this morning. East said in a statement: “It has been a privilege …
Of anyone doing embedded.
Reminds me of the story I read yesterday about Michael Foot being made commissioner for nuclear disarmament or something. The Times's headline was "FOOT HEADS ARMS BODY"
It's great to see decent headlines returning after the seo madness subsided.
And then there was also that Bulwer-Lytton competition runner-up from 2010 (Detective category), which is what immediately sprang to my mind:
As Holmes, who had a nose for danger, quietly fingered the bloody knife and eyed the various body parts strewn along the dark, deserted highway, he placed his ear to the ground and, with his heart in his throat, silently mouthed to his companion, "Arm yourself, Watson, there is an evil hand a foot ahead."
I think it even got read out on Countdown. Make of that what you will :)
..........................reporting the retirement of ARM Holding's CEO?
Not at all.
The guy was in charge when Microsoft announced the equivalent of an industry seismic shift.
Obviously, the shift being managed by Microsoft it ends up looking like a wet firecracker but I can see the M$ angle.
OK, I have to toe the line and hand it to you, but what part of the male anatomy are you waving?
"it wasn’t until the success of the Apple iPhone and iPad that the ARM name finally went mainstream"
I thought the same! It's not like apple give any of their supplies any credit at all (think corning gorilla glass). Other companies would use it as a selling point but thats not Apples way (which is fine, it's their game), it's special apple glass. Apple may have brought volume but so did RIM (although they did use some intel 386 chips for a while?) with xscale chips.
Every mobile phone, surely, not just Apple?
Every hand held computer? (from PocketPC onward)?
The router on your broadband.
Your TV? Your set top box, surely.
And so on.
If it needs a little computer, it's probably an ARM.
I haven't seen much sign of ARM in the "tough" end of embedded systems where mil-spec PowerPC and the like used to be fashionable. Is there a presence or are these folk just slow to catch up (for understandable reasons)?
RIM may have used some Intel chips I think, but ARM-based ones rather than x86 I believe (Intel was an ARM licensee at one stage, but sold that business to Marvell IIRC).
"I haven't seen much sign of ARM in the "tough" end of embedded systems where mil-spec PowerPC and the like used to be fashionable. Is there a presence or are these folk just slow to catch up (for understandable reasons)?"
In the US historically the USAF 1750A architecture has been the one a lot of people used (I think it's still running most ULA launches) but the rad hard versions of a POWER PC are where they've gone.
In Europe they seem to prefer the SPARC architecture, also implemented in some rad process with majority voting registers. Sun wanted to get more mfg making SPARC chips so I think their licensing was pretty lenient.
As both processor have enough power to host linux implementation they can leverage the whole tool chain.
The bottom line is ARM could be an instruction set that was used in this market if a company that wanted to get into the business licensed the instruction set and found a foundry to do the rad hard implementation.
The "problem" with ARM is that it's hidden, no equivalent or "Intel Inside". And it seems to be hidden in just about everything that uses electricity! Whist our desktop PCs don't use ARM as the main CPU, they've probably got a couple of ARM chips in there maybe in the hard disk controller or graphics card. When the price of the most basic versions of ARM is pennies then it becomes the best choice for anything needing a bit of logic. My stockbroker said it was time to take profit (after 2000% growth) on my ARM shares recently but then totally blew his credibility (like the author of this piece) by saying it was a risk because Apple may stop using ARM...
And when I say "problem" - well is it? I probably own a few dozen ARM chips in phones, computer preipherals, GPS, (car? Fridge?), I don't know and it doesn't matter, like I don't know the manufacturer of the mains cable in my house, I don't need to, leave that kind of decision to the guy responsible for wiring up the house. The only device where I do explicitly know it's ARM based is RaspberryPi. I'm fairly sure my mobile phone is too but my criteria were not "what chip" but the end-deliverable features - things like speed and battery life.
Keyword: mainstream. That's not tech-world mainstream, that's the wider-world mainstream.
"if a company that wanted to get into the business licensed the instruction set and found a foundry to do the rad hard implementation."
Thanks, but I think there's more to this than meets the eye so far.
I'm vaguely familiar with (for example) OnSemi. They have licences for various flavours of PowerPC *and* for ARM, amongst other products and technologies. They produce mil-spec (including rad hard) stuff and commercial stuff.
I can entirely see why existing PPC and SPARC designs might well stay that way; apart from anything else, the certification costs would be unpleasant (although in at least one company I know of, they're now suggesting source-level testing is sufficient and that something like a change from 68K to PPC would therefore not need sw to be re-tested using the appropriate toolchain and binaries! Nothing could ever go wrong in the toolchain, could it).
I currently struggle to see why any clean sheet or new-generation design would want to be on anything other than ARM, where there is a massive (and massively succesful) set of support software (free and paid-for) and a massive installed base and a relatively large base of clued-up embedded folks to leverage, not to mention the benefits of ARM-style system-on-chip technologies where applicable. And, where applicable, Linux on ARM as a target and/or Linux (on anything) as a development host.
"Thanks, but I think there's more to this than meets the eye so far."
What I've downplayed is the bill the foundry will charge you for the work.
I'd guess both ARM & and foundry will set one off charges and royalties based on projected volumes and the volume for such systems is just not that large.
That would include proving the chip design has the necessary radiation resistance, which AFAIK is not cheap either.
"although in at least one company I know of, they're now suggesting source-level testing is sufficient and that something like a change from 68K to PPC would therefore not need sw to be re-tested using the appropriate toolchain and binaries! Nothing could ever go wrong in the toolchain, could it)"
On the surface this looks reasonable because the software will run on the same hardware and only the code generators and debuggers are going to need to be different,right?
But IRL both of those are pretty big chunks of software in their own right and I doubt they will be alone.
The fail is strong with this idea.....
"I currently struggle to see why any clean sheet or new-generation design would want to be on anything other than ARM, w"
Agreed. The question is how often people have that opportunity.
Biting the hand that feeds IT © 1998–2017