Single-Chip x86 Chipsets Around the Corner?

An anonymous reader writes "Kontron, a giant among industrial single-board computer vendors, yesterday revealed a credit-card sized board apparently based on a single-chip x86 chipset that clocks to 1.5GHz and supports a gig of RAM. It targets portable devices — not x86's usual forte. Kontron isn't saying whether the board uses a Via or an Intel chip(set) — both vendors reportedly have single-chip chipsets in the works, part of their respective missions to drive 'x86 everywhere.'"

Comments

Great idea

[nurb432]

If they can find a market for it. Its going to be hard to unseat the arm.

"generic" embedded devices come to mind. ( but you have the pc104 standard there already..

Re:Great idea

[truesaer]

AMD makes a whole line of embedded x86 processors. http://www.amd.com/us-en/ConnectivitySolutions/ProductInformation/0,,50_2330_9863,00.html

Re:Great idea

[Calmiche]

Yah, but current ARM processors max out at about 700-900 mhz.

If they can really pull off a good, stable, low powered chipset in the 1.5 ghz range.. I would be very interested.

I am still waiting for a revival of the handheld computers. UMPC isn't going anywhere, Palm is getting out of most hardware.

HP is FINALLY getting back into the handheld market, but it's WAY late for it's projections and dosen't seem to be doing any advertising at all for it's new line.

Re:Great idea

[Colin Smith]

Yah, but current ARM processors max out at about 700-900 mhz.

If they can really pull off a good, stable, low powered chipset in the 1.5 ghz range.. I would be very interested. Right. Because more gigahertz means faster.

Re:Great idea

[spirit of reason]

I am still waiting for a revival of the handheld computers

You mean something like the Pandora?

Also, more information here.

While it's technically meant more for a gamer market like the GP2X, the arm + linux + wifi + usb host + decent resolution screen might make it a more general purpose machine.

Re:Great idea

[bombshelter13]

Yes. That's ~exactly and exclusively~ what more (giga)hertz means: it's faster.

Now, what it doesn't say anything about is whether it's higher performance.

Re:Great idea

[nurb432]

Arm is still lower power. That is pretty important.

But i do agree there is a large x86 codebase out there. ( but then again, there also is a decent sized codebase for ARM and other embedded processors )

Re:Great idea

[canuck57]

Right. Because more gigahertz means faster.

That is a fallacy big time.

One game is to just clock up the frequency and make you think you have more. Put a divider in the middle and I could give you a 20GHz CPU. It is about throughput. How much can I get don in n cycles. For this, benchmarks are where it is at. Pick a benchmark(s) that is similar to the anticipated loads and work from there.

Re:Great idea

[afidel]

When you're comparing a superscaler, out of order, multipiplined x86 chip vs an ARM? Yeah I think the 2x faster chip will win, in fact even at the same mhz the x86 part would most likely be faster. Now the question is, is it capable of more MIPS/WATT, which is what matters almost as much as absolute performance in the embedded space. There the answer is possibly but unlikely.

Re:Great idea

[hitmark]

read up on cortex A8 and A9...

x86 cores?

[heroine]

It would be huge if x86 or x86_64 was available as a core like MIPS & ARM. Life would be much easier for the set top boxes.

Re:x86 cores?

[0100010001010011]

Just like a monoculture of GSM has hurt the innovation of mobile phones in Europe.

God bless the USA where competition between GSM, CDMA, & what ever sprint uses has increased innovation such that the USA always has the best cellphones out of any civilized country.

Not that I don't think in

And Theo's quote can be found here: http://kerneltrap.org/OpenBSD/Virtualization_Security

"x86 virtualization is about basically placing another nearly full kernel, full of new bugs, on top of a nasty x86 architecture which barely has correct page protection. Then running your operating system on the other side of this brand new pile of shit. You are absolutely deluded, if not stupid, if you think that a worldwide collection of software engineers who can't write operating systems or applications without security holes, can then turn around and suddenly write virtualization layers without security holes."

Who's in charge of code names?

[KevinKnSC]

From the article:

Codenamed "John," the processor will integrate CPU, northbridge, and southbridge...

That was the best code name they could come up with? Seriously?

Re:Who's in charge of code names?

[User 956]

"Codenamed "John," the processor will integrate CPU, northbridge, and southbridge..."

That was the best code name they could come up with? Seriously?

Given what they probably had to do in the area of patent licensing, calling it a "John" is pretty polite, if you ask me.

Re:Who's in charge of code names?

[gendusoa]

Via's code names are almost always biblical:

* Luke
* Esther
* Nehemiah

and I'm sure the others I can't remember off the top of my head are biblical names too.

Re:Who's in charge of code names?

[brit74]

Via's code names are almost always biblical:

* Luke
* Esther
* Nehemiah

I'm still looking forward to the Satan and Whore of Babylon chipsets.

Power consumption please?

[4D6963]

It targets portable devices -- not x86's usual forte

Yeah, that's not x86's usual forte because x86s are more power thirsty than say MIPS or ARM, which is why it would be interesting if the article could mention how much this new thing is supposed to drain.

Re:Power consumption please?

[RightSaidFred99]

This isn't true because of some "law" that says x86 must be power hungry, it's true because nobody's really sat down and done an x86 processor for the embedded market, or at least not donw well. Check out Silverthorne, it has power use comparable to MIPS/ARM.

'x86 everywhere.'

[geekoid]

What did we do to you to deserve this?

It's VIA

[RzUpAnmsCwrds]

If the chip is codenamed John, as the article claims, it's a VIA chipset. VIA uses biblical names for their CPU codenames.

Previous VIA CPU codenames:

Samuel
Esther
Nehemiah
Ezra

Note also that VIA combined a C3 CPU and a northbridge into a single package - it was codenamed "Luke".

Re:It's VIA

[TubeSteak]

If the chip is codenamed John, as the article claims, it's a VIA chipset. Not to bruise your karma, but I suspect people are going to get the wrong idea from your post.

Here are the two quotes FTFA:
1. "It is based on an unnamed "highly integrated chipset" from an unnamed silicon vendor."

2. "Via has long planned to bring out a single-chip part in its CoreFusion line. Codenamed "John," the processor will integrate CPU, northbridge, and southbridge into a single x86-compatible SoC (system-on-chip)."

AFAIK, Via & Kontron have nothing to do with each other.
TFA is not stating that Kontron is using Via's chipset.
TFA later goes on to say that Via may eventually be a competitor to Kontron.

Sounds like a bad idea to me

[Vthornheart]

"'x86 everywhere.'"
Can I pass on that? The x86 architecture may be POPULAR, but it's inefficient, forced into backwards compliance with horribly outdated standards, and has been horseshoed for the past 20 years into a full architecture chip when the initial design was never meant to become like this.
If a realm of computing has x86 as the non-dominant chipset, I think that's a blessing and it should remain that way. You can't do anything about the PC market at this point, for example... but I think the motto should be "x86 only where it already exists" rather than "x86 everywhere."

Re:Sounds like a bad idea to me

[pla]

The x86 architecture may be POPULAR, but it's inefficient, forced into backwards compliance with horribly outdated standards, and has been horseshoed for the past 20 years into a full architecture chip

The "x86 architecture" doesn't exist. x86 merely describes an ISA exported by the microcode of whatever underlying architecture a given chip really uses. An ARM chip could look like an x86 chip. A PPC chip could look like an x86 chip. The Core2 or Athlon64 could just as well export a traditional Motorola ISA as the chosen x86 - and with modern chips, they could do so with a microcode patch at boot time, you wouldn't even need to buy a new chip!

Thus, any holy wars regarding its efficiencies or inefficiencies must remain firmly rooted in the ease of actually using it for coding. I do so, and find it for the most part adequate. It traditionally lacked enough GP registers, but even that doesn't hold true these days (at least for AMD's version - Not 100% sure about the Core line). And for that matter, very few coders even bother with ASM anymore... Even firmware development (which I also do) uses C almost exclusively nowadays.


Not to say I want to see it everywhere, but we can't really hold the flaws of ancient hardware with no current connection to the ISA against it.

Re:Sounds like a bad idea to me

[Windom Earle]

Well, this discussion appears to be about an 'x86 Chipset,' not an x86 processor. Which means more than just the processor. So there is an 'x86 architecture.' If there wasn't, this topic would be meaningless.

Re:Sounds like a bad idea to me

[pla]

Well, this discussion appears to be about an 'x86 Chipset,' not an x86 processor. Which means more than just the processor. So there is an 'x86 architecture.' If there wasn't, this topic would be meaningless.

My earlier post may have sounded more caustic than I intended it, but I meant what I wrote literally.

A "x86 chipset" just describes the supporting chips (usually memory, bus, and I/O) that let the CPU-which-happens-to-speak-x86 do its thing in a way familiar to programmers and users of non-embedded PC hardware. Although the PC world has a somewhat standardized set of these, by no means do all x86-speaking CPUs require similar enough supporting hardware as to justify the overly broad generalization.

Simple example - The earliest x86 chips manually controlled DRAM refresh. Then that moved out into the chip-set (a large collection of single-purpose chips each doing their own thing - including a memory controller as one of them). Then a handful of those merged into a single-chip solution that handled all the critical non-CPU functionality (but had basically nothing we would think of as actual features by itself). Then as more and more I/O tasks also merged into that one chip, it eventually split into the Northbridge (bus and memory) and Southbridge (I/O). And now, we have CPUs with integrated memory controllers once again (though they have dedicated hardware for it rather than needing to tie up actual processing cycles for the job). Which of those would you call "the" x86 chipset?


Perhaps you could more accurately say that this provides for a single-chip PC architecture, since most of what the chipset does has little to no relevance to what language the CPU speaks.

Re:Sounds like a bad idea to me

[truesaer]

x86 is actually quite efficient and flexible. Variable length instruction sets have a lot of benefits, especially in embedded applications where memory is at a premium. You'll get a lot more data into your cache and memory with x86 code as opposed to a fixed length instruction set. Most of the disadvantages (complex decoders, etc) are borne by the chip designer. And for embedded there's no reason they need to support legacy stuff.

This common belief that x86 is the devil is simply absurd. It sounds good but it doesn't really match with reality. As long as the chip designers are willing to make them, they're a great choice for software and system developers.

x86 programming

[neapolitan]

It's a shame that the x86 is such a complex instruction set; this means that the age of the handheld computer as an easy programming platform for hacking is over.

When I was programming for Apple //e, I had a good majority of the opcodes for the 6502 chip memorized, laying out assembly by hand. I later learned 68k assembly, and again, it is very "understandable" to a person just sitting down in front of the computer looking at an assembly printout. In the early 90's, pretty much x86 dominated and I stopped doing pretty much all assembly programming.

In 1996 I was delighted when the palm pilot came out, using a 68328 (68k instruction set). It was like a renaissance, again programming in assembly and hacking other things for fun. Now, once again, it appears this will be dead!

As a question to the slashdot community, is it possible to program "naked" x86 assembler? I have never really put in the time to learn it, but it just seems exceedingly complex and tedious to program for this chip without use of a higher level crutch (C compiler...) I am sad that once again everything I know is becoming outdated... :)

Re:x86 programming

[Waffle Iron]

but it just seems exceedingly complex and tedious to program for this chip without use of a higher level crutch (C compiler...) I am sad that once again everything I know is becoming outdated... :)

You sure are outdated. Today's "higher level crutch" is Python.

What would it take?

[Zaphod-AVA]

What would a chip have to include for VIA to codename it Jesus?

Re:What would it take?

[Kagenin]

A built-in WINE environment.

Re:What would it take?

[makapuf]

Flash memory (or some storage). Remember JESUS SAVES !

Re:What would it take?

[Anonymous Coward]

Nah, Just water will do, Jesus can do the rest...

More info for x86 in embedded dev. at arstechnica

[IYagami]

You can find articles about the use of x86 in embedded devices at arstechnica, from Jon Stokes:

Return of the Son of Pentium in 2008? Intel's new ultramobile processors

Intel's low-cost "Diamondville" CPU to power OLPC/Eee PC mobile category

And a very interesting article why processor makers want to extend their architecture to other realms: Beyond the BlackBerry crowd: life in a post-32nm world

Crap idea

[EmbeddedJanitor]

Single chip x86: Geode etc are a crap idea. The idea has been done to death and has never caught on. There's no real benefit in them. In the past there was some appeal in x86 because of good, cheap compilers etc. Now there's gcc for everything this advantage has long since disappeared.

ARM, and at a push MIPS, PowerPC and SH4 own this space. x86 needs to offer something huge to get back in the game.

Re:Crap idea

[cnettel]

GCC targets everything and still is more heavily optimized for x86. Despite that, it's FAR from the best x86 compiler around, performance-wise. The pool of people with a "can read, cannot write it good without great pain" grasp of x86 assembler is also damn huge.

Other x86-specific assumptions inherent in code (like atomic writes of different sizes, context switches limited to instruction boundaries) means that a platform porting of seemingly good multithreaded code can cause very subtle bugs. It's even possible to write Java code that is almost impossible to turn into a race condition on x86, but where you might do it on other platforms. You might argue that it's rare or that the code is "bad" and incorrect in the first place, but it's still there.

Re:I hope not.

[Sillygates]

It's already obsolete(in terms of size), and this is not news :(
pico itx is already on the maket, the mainboard is about the same size (1.5ghz [like that means anything], upto 1gb ram):
http://en.wikipedia.org/wiki/Pico-ITX

and the transmeta crusoe processor (which implemented x86 in software) has been out for almost a decade now. The sony picturebook has a credit card sized motherboard along the left side of it's case:
http://images.google.com/images?svnum=10&um=1&q=sony+picturebook&btnG=Search+Images

How about a better summary first?

[mr_mischief]

Yeah, because there haven't been 386, 486, and other systems on a chip and Via doesn't have a 1-watt processor anywhere to be found. This is not the first 1-chip chipset for all of the x86 line. That's bullshit. An SoC is even more integrated than just having the chipset as one chip. Somebody never read the old Computer Shopper before it slimmed down. SoC solutions for x86-compatible systems have been around more than a decade. The summary is bad, because TFA does not say this is a first for the x86 line.

You're right that even low-powered x86 chips like the C7 and the Geode line are generally no match for ARM and XScale. MIPS I'm not as familiar with for power usage purposes. It'd be nice if that question was answered, but I'm afraid it'd be summarized incorrectly too.

2005 article on anx86 SoC
another 2005 article about a different x86 SoC
2004 product page for an already obsolete x86 SoC
Linux Devices list of x86 SoC solutions, some dated to 2000
2000 Register article about the year since Cyrix released an x86 SoC
Chipslist page showing availability of AMD processor with 80188 features plus DMA, watchdog timer, serial ports, and I/O pins in 1995
article on the National Semiconductor Geode (the owners of that line before AMD bought it) thin client system-on-chip

And the best proof of all: an archive of a 1996 story on the AMD Elan,which featured a 386, ISA bus, serial UART, memory controller, power management, and PLL hardware ON ONE CHIP

x86 should be like slavery in the 1820

[kiyoshilionz]

x86 has its market, the personal computer, but its legacy architecture should not be allowed to spread anywhere it has not already tainted. Remember Why Do We Use x86 CPUs? I thought x86 is something we want to eventually move away from (Remember VAX?), not something we want to spread.

Re:x86 should be like slavery in the 1820

[evilviper]

I thought x86 is something we want to eventually move away from

You were wrong. x86 isn't particularly impressive, but it's just a CPU, not a war crime.

It's pretty much inevitable that x86 will move into new areas, as embedded systems need more and more processing power for multimedia, x86 vendors spend more and more of money reducing power consumption, and the economies of software development more and more favor reusing x86 software, rather than spending time on optimizations for the other architectures you use.

Since Intel can't seem to make money on any architecture other than x86, they've eliminated their StrongArm/XScale line, and are replacing it with ultra-low-powered (sub-1watt) x86-based CPUs. VIA has long be trying to make inroads in the high-power, higher-performance embedded market with their own CPUs as well.

Didnt Cyrix try this a decade ago?

[DJRikki]

With the MediaGX (I think) range? Integrate everything you can think of into the die including sounds and graphics.

Re:Why no x86 microcontrollers?

[Windom Earle]

I think you're asking about the 80186 and the 80188, which have had a long history in embedded use. They're not particularly popular anymore. They'd match your laptop if you still use a Toshiba 1000SE, of course.

Similar parallel offerings from Intel were the 80196 line.