Slashdot Mirror
Moore's Law for Motherboards
An anonymous reader writes "VIA CEO Wenchi Chen revealed a business card-sized motherboard billed as the 'world's first industry-standard form-factor for PC/phone convergence,' at Computex this week. The mobile-ITX" board measures 3 x 1.8 inches. It's half the size of pico-ITX, which was half the size of nano-ITX, which, in turn, was half-the size of mini-ITX — which was already small. It's not clear whether VIA will make these tiny motherboards available to end users, or if they will only be sold directly to device makers, but generally all of VIA's tiny motherboard formats have spread around to other suppliers and become widely available."
Wake me when they have an equivalent System on a chip.
I always thought of Creationism as the Raving Right's version of the Loony Left's Anthropogenic Global Warming-brightmal
Imagine building itty-bitty robots...
Or digital picture frames...
Case-modding an Altoids tin...
You can't talk about Wikipedia's flaws on Wikipedia
TI already has your system on a chip. It's called the DM6442, DaVinci.
The society for a thought-free internet welcomes you.
The embedded world has had complete computers on "motherboards" this small for quite some time. Check out gumstix sometime.
The fundamental problem with PC based motherboards has always been heat dissipation and interface connectors. Heck, the back panel of my desktop uses more area for the connectors than exists on this board. There are processor heatsinks bigger than this thing!
PC's have always been about cheap computing power, not low power dissipation or form factor. I remember a time when the power of your desktop was considered commensurate with the size of the box - we had friends putting regular motherboards into server towers so they could "impress" fellow geeks.
Not that I would mind x86 in the embedded world, but it seems to me that this is going nowhere fast. The problem isn't technical - it's business. Most embedded systems run some sort of ARM variant, which would mean that code would have to be ported to x86. Furthermore, there's no way this would make it into a cellphone - primarily because of the fact that it is x86, and the carriers are adamantly opposed to the prospect of the consumer being allowed to run unauthorized code on their cellphones.
Linux already runs on the ARM, and you still aren't seeing a proliferation of ARM-based general purpose computers. While this would be nice for a sub-notebook, the problem is that sub-notebooks, while a personal favorite of mine, typically have not done well in the marketplace. Consider the HP Jornada, which was discontinued after a few short years. And it seems today that that trend is toward larger, not smaller, laptops.
The society for a thought-free internet welcomes you.
Here's my first suggestion for what to do with this thing. I want a handheld (something the size of an battery-powered face shaver or large cell phone) written language translator. On the out-facing side is a mini-scanner and on the other side is a graphics LCD screen.
Suppose you are in some place where you can't read the language (it does happen in the age of 500-seat 6000km airliners). You get a newspaper, wave the scanner over the text, and within about three seconds, the scanned writing appears in English on the LCD. You can either pop in small memory cards for different language families or have their programs stored on your laptop for downloading.
Another feature would be a built-in microphone with a program that has been trained to your voice so that you can speak into this mic and have your words translated into the written form of the local language and displayed on the LCD.
I'd be willing to pay about $100 US for this device. I'll bet a lot other people would also. Anyone interested in developing it? Would we have to pay royalities on a language? Could such a device be built on this new miniature PC card? Am I just dreaming?
Can anybody even imagine how amazing it would be if cell phone networks became like wifi? You pay a monthly fee for access and you're on. Devices like this motherboard would really really open up the possibility for a homebrew-cellphone market. What would be very interesting to see is cell phone carriers become more like ISPs. You get some bandwidth from them, and you get to use it for pretty much whatever you want.
NewslilySocial News. No lolcats allowed.
They're going to run out of prefixes pretty quickly, since they're usually applied to powers of 1,000 rather than 2. And whatever happened to micro? milli (okay, mini), micro, nano, pico, femto, atto. I don't think there's anything past atto.
Don't piss off The Angry Economist
Yup, we're reaching the size limits regarding standardized, interchangable interfaces.
Thankfully, we're already tossing out electrical connections for low-speed I/O with the introduction of Bluetooth, ad there's the portential for making high-speed I/O wireless with beefy 802.11n and later revisions.
The big problem is power. Power is going to kill innovation in the wireless devices field unless we can come up with some impressive storage capacity improvements. You can improve the efficiency of the transmitter and receiver, but you can't really reduce transmission power much futher than we've already done. That would require reducing the noise floor, which is impossible.
I don't expect any of these small concept PCs to pan-out precisely for this reason: many people want flexible interfaces in their devices so they can communicate.
Man is the animal that laughs.
And occasionally whores for Karma.
OK.... lets look at your claims and your gripes.
1) The C7 core runs a full speed in-order ALU and FPU. Unlike the C3, where design constraints required a half-speed FPU, this one has it at full speed. The ALU has a full 24 cycles to complete simply 32 bit operations (should only require 4 cycles, at best). Let's not even mention the 64KB 4-way associative L1 cache with only 3 cycle latency. Even at 1Ghz, this would indicate to me that the clock speed is not, in fact, ALU bound, but more likely FPU or L1 cache bound. Where did you get your ALU claim? It sounds made up to me.
2) The C7 core benchmarks show it to be about 40% slower, clock-for-clock than the new Celeron M. Presumably the ULV version may be a hair slower, leaving it at half the performance clock-for-clock. Considering that today's Celeron M is far better per clock than the original K8, it would lead me to believe that this chip would be roughly equivalent to a 500Mhz K8 processor, but using only about 1/10th of power.
This chip uses a 90nm SOI process, and VIA's process is up there with the best in the world. I'm not exactly sure what the basis for your gripe is...... except it sounds hollow.
Also, the (as you put it) "P4 escapades" are only one of many examples where clockspeed != performance. Take the old Intel 386SX or the AMD 486DX4-120 or the old AMD K6-2, which were all total dogs "per clock" compared to the much more efficient (at the time) 386DX, Pentium and Pentium 2.
Stew
There are 10 kinds of people in the world. Those who understand binary and those who don't.
You have a peculiar definition of "way more". The claims I'm familiar with is that the C7 performs comparably to a Pentium III at half the clock frequency, except on many cryptographic benchmarks, where the C7 outperforms the Pentium III by more than a factor of ten. My recollection is that the K8 is a highly optimized three-issue design (decode and retirement rate limited to three instructions/clock), whereas the Pentium III is a middle of the road three-issue design (more interlocks and stalls, more unnecessary opcode splitting). The difference could range anywhere from zilch (where the instruction sequence has no liabilities on the P3) to about 50% faster on the K8. I can't recall the last time I said "woo hoo" over 50% some of the time.
The C3 was appallingly slow at floating point. I've read the C7 is vastly improved, but still far from great. Still, your concept of "way more" seems to better apply to a comparison of a 1GHz C3 to a 500MHz K8. No one would argue there.