Intel tri-gate finFETs

XP on Intel. That's a practical reality, given the apps I need to run. I don't expect that people will be running Windows on x86 forever.

Tablets are already outselling desktops. My wife's iPad is a revelation: no cables, no keyboard, no noisy box plugged into the wall, no CAT5 cables, no wait to boot up or launch apps, 9 hours of battery life, no crashes. It makes a desktop x86 computer look like the ancient power-hog kluge it is.

We wouldn't dream of using any Intel architecture in our products.

8048-types are dogs, and modern Intel chips are absurd in hard embedded systems. All our older stuff was 68K, and all the current stuff is ARM. No heat sinks in sight. We have one board that has 13 ARM processors on it.

Linux isn't good stuff? GCC isn't good? LT Spice? How about Google and wikipedia?

John

There are flash-based ARMs for under a dollar now, so there's not much reason to keep pics or 8051s around. Standardize on one tool chain.

Serious 100 MHz ARMs, with ethernet, ADC, DAC, flash, ram, internal clock, all sorts of goodies, cost like $4 these days.

John

PowerPC does all of that, but adds a pile of registers.

Different game.

*THAT* is the power of ARM. That and the fact that you have so many to choose from. Everyone makes 'em. While MPS-430 looks nice, I don't see the point.

Wow -- I never knew that. Thanks for the tip! I'll take it for all that I think your opinions are worth!

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" was written for you.
See details at http://www.wescottdesign.com/actfes/actfes.html

What are the opinions worth of one who feeds the troll? ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |

      Remember: Once you go over the hill, you pick up speed

Of course. It is so nice and convenient to curse Wintel, isn't it?

I don't expect to live forever, so I don't need a perpetual desktop either.

Tablets are the toys. Desktops are making money.

Yes, Intel is usually not the best choice for the handicraftsmen and amateurs.

However, you could find a PC x86 compatible machine in just about every ATM, gas pump or PLC.

A pile of poorly written, buggy, undocumented and mostly incompatible code.

Commercial packages are clearly better.

Sales tool. Actually, poor simulator.

Sales tools. You are paying for Google and Wiki when you buy stuff.

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

That's only a matter of time, Intel focused on their crap CPU in line with Microsoft, riding so high. ARM is good, and will also gain that or similar technology in time.

I think Intel's been emulating the old instruction set for ages now, so they cannot let go, and they not release another RISC family for mass market, they pulled Apple into the fold, using their crappy CPU too.

Needs Microsoft to support another architecture, for it to reach desktop -- MSFT already doing that with the baby computer/phones.

Interesting times. Intel have shown they can move fast, like when they followed AMD64 brown bag bolt-on within months, and let their then new 64bit languish.

Grant.

Intel tried iapx32 and Itanic, both over-intellectual architectures and expensive failures. They licensed ARM and then abandoned it.

Balmer announced that the next Windows will run on x86 and ARM. Microsoft and Intel have never really trusted one another.

ARM on new-format devices, phones and tablets, running versions of L/Unix, and the end of Moore's Law, could change the whole game. Goodness knows it's time for some sort of change.

John

^4

That's a fact that has escaped a *lot* of people. M$ would be more than happy splitting the CPU market up into a half-dozen competitors. They want to be the only ones with a piece of the computer pie. So far, things have worked out pretty well for them.

That *wasn't* all that fast. They did it, reluctantly, after Gates & Co. ordered it.

Yep, their market didn't budge, and they bent to follow.

Any reason why Intel can't produce ARMs again, if they choose?

Grant.

Uhhh... cause they sold the design and the rights to it.

They could 'license' it back. Don't see that happening.

More likely a new design that will do amazing things on near zero power.

If there has been a great mistrust, why did Microsoft drop the RISC and Alpha support after NT4 ?

Irrelevant. There wasn't a market to justify it.

Damn! The idiot got one right.

Alpha subsequently died, and now look what Apple uses.

How much market does RISC still have? (why would I ask you) (it is a test) see if you can answer without mouthing off.

What's supposed to be the advantage of a reduced instruction set computer? If you have fewer instructions to choose from, wouldn't that mean that you'd have to use more of them to get the same job done?

Thanks, Rich

The original rationale was that instead of spending silicon area on hardwired logic for rarely-used complex instructions, use it on extra registers instead.

Then people started wanting to issue more than one instruction per clock, so RISCs started getting a lot more complicated. At this point it's more of a party badge than anything--RISCs are usually register-rich and have relatively small instruction sets.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal
ElectroOptical Innovations
55 Orchard Rd
Briarcliff Manor NY 10510
845-480-2058

email: hobbs (atsign) electrooptical (period) net
http://electrooptical.net

The idea is to have a simple instruction set and execute one instruction per clock, unlike CISC architectures that take many, and variable numbers, of clocks per instruction. Fast RAM and especially cache makes that advantageous. It must be horrible to try to really optimize the pipeline for a complex instruction set like x86.

RISC architectures also emphasize having lots of registers available to minimize slow memory accesses. The only direct memory references are usually LOAD and STORE.

One thing about RISC instruction sets is that they tend to be compiler-friendly, not people-friendly.

John

That was the original rationale, as I remember it. Compilers use only a subset of a CPU's instruction set, so why not reduce instruction set and have the compilers optimise the code? And the registers, need lots of fast temp storage very close to CPU.

But Intel stuck with backwards binary compatibility for decades.

I keep hoping for another z80 like revolution in CPU design, compatible with existing CPU but adding some amazing features. But it seems ARM will sneak in by growing up to 64 bits. But I don't expect to find a

6pin ARM chip controlling some simple hardware widget. PIC will stay around too.

Grant.

Pretty good, DimBulb, from ARM to Nintendos and from Telephones, to Servers, to SuperComputers. Let's see if you're smart enough to apologize for being the world's dimmest bulb.