First AMD 4x4 Mobo Spotted

[adrian]

Source

Intel may be getting all the attention with its quad core and Core 2 Duo blitzkrieg, but AMD's not one to rest on its laurels. Its forthcoming 4x4 platform will place two dual-core CPUs on one mobo for a total of four cores. Asus' L1N64-SLI WS will be the first mobo to support the 4x4 architecture and this here is an overview of what you can expect. It'll have 4 memory slots (2 memory sticks dedicated for each processor), 12 SATA ports, and 4 PCI Express x16 slots. The two CPUs will be connected via AMD's Direct Connect architecture. Spec-wise, the board is fully loaded. Our only concern will be pricing, as something with this many features can't possibly be cheap and could potentially be more expensive than the Intel alternative.

Geeks, your porn is here!!

[craigwatanabe]

Re: First AMD 4x4 Mobo Spotted

You know I get the feeling that in time motherboards won't sport any cpu's at all but will tend to mount them on removable trays (like in the old days) to suffice more powerful cpu's on available platforms.

This thing about multiple cores is beginning to look more like those disposable razor commercials with not two or three but five blades for a smoother shave! Heck why not 200 for that matter then!

The sad truth about more computing power is the demand for electrical usage. And here we are trying to conserve energy. Five years ago a 500-watt power supply would be generous. Now days it's almost par for the course with new gaming systems sporting 750-watt to 1Kw powersupplies.

1000-watts per hour! Multiply that by a night on HT (maybe 4-hrs) and that's enough energy to cook a turkey for Thanksgiving!

[GeckoGeek]

Re: First AMD 4x4 Mobo Spotted

You know I get the feeling that in time motherboards won't sport any cpu's at all but will tend to mount them on removable trays (like in the old days) to suffice more powerful cpu's on available platforms.

I worked for a dealer of a computer like that. Didn't really work out. Faster CPU needs a faster motherboard, faster memory. An upgraded m/b was slower then a new m/b. Then there are the different buss that shows up. AGP showed up when PCI was too slow for video, USB was added, etc. Not to mention that system sold at something of a premium over regular ones.

Bottom line, there was a reason that approach was abandoned.

[craigwatanabe]

Re: First AMD 4x4 Mobo Spotted

I worked for a dealer of a computer like that. Didn't really work out. Faster CPU needs a faster motherboard, faster memory. An upgraded m/b was slower then a new m/b. Then there are the different buss that shows up. AGP showed up when PCI was too slow for video, USB was added, etc. Not to mention that system sold at something of a premium over regular ones.

Bottom line, there was a reason that approach was abandoned.

But buss speeds has nothing to do with circuit traces (okay some) but more with the circuitry that controlls the buss itself. The whole point of an integrated motherboard was to combine more function on one board (hence the word motherboard) to allow for a smaller computer case or form factor.

I think it would be great if I could use another graphic chipset over another if I could simply unplug and insert a different chip. Right now we can do that with processors but even that requires compatible mobos for certain cpu's.

One thing I do understand is that the wiring for a cpu be minimized for faster processing speeds, however more processing is used for things like graphics. And with today's PCI-E SLI graphics cards you don't need a faster processor because most of it is being done onboard the graphics card anyway.

As we go for multiple core mobos, full tower ATX cases will become mainstream again and pretty soon we'll be looking at mainframe sized computers again.

[GeckoGeek]

Re: First AMD 4x4 Mobo Spotted

But buss speeds has nothing to do with circuit traces (okay some) but more with the circuitry that controlls the buss itself.

Well, yes, but are you saying that the motherboard should just be a backplane? I don't think that will work (see last paragraph). The problem with the computers I worked with was that the memory was on the m/b while the CPU was plug in. So a upgraded CPU got slow memory. Stuffing a 486 into a 286 m/b just didn't work. When I got my 200MHz MMX I thought I was set. Then AGP came out, USB went to 2.0, drives got faster, etc. When all was done, after a few years all I could recover out of that machine was the CD ROM and the floppy. Had I been a little bit lucky, I could have reused the case and power supply, but the that's when everyone went ATX.

But the bottom line, is that when you change the CPU beyond just the speed, usually everything else is in need of a replacement as well. Two years or so have gone by and something better is out for most everything. Pick any two years in the past and see if it would have worked out for the average buyer (not the gamer). The marketplace has found it cheaper to sell the whole package then to sell parts.

And no, I don't think computers will be getting much larger. At 3GHz, anything 10cm away is on the last clock cycle. That's how far the speed of light can get in that time period. You'd have to break the speed of light to get faster and larger.

[craigwatanabe]

Re: First AMD 4x4 Mobo Spotted

I'll let you in on some chit chat I heard in 1980 while in the Air Force mapping out circuit boards to determine their schematics then determine what the heck these boards did. Soviet electronics they told me, yeah right the Soviets were 10-years behind us in electronic technology they should have be doing this to us.

I heard some Air Force techies mention Base 5 level computing. What the heck? As fast as we can process information we are still at Base 2. I couldn't imagine the processing speeds of a Base 5 CPU.

But getting back to mobos. Eventually I think there will need to be a standardized buss. And yes you'd have to replace virtually all components to accomodate the differing CPU's, but in this world of recycling electronic circuit boards, the circuit tracings would be one less item on a mobo that would otherwise go to the landfill.

[GeckoGeek]

Re: First AMD 4x4 Mobo Spotted

I heard some Air Force techies mention Base 5 level computing. What the heck? As fast as we can process information we are still at Base 2. I couldn't imagine the processing speeds of a Base 5 CPU.

And where are the Soviets now that they need some money? I'd say it's not doing too well. I'd have to understand more about how base 5 was implemented, but I'm not sure as I see an advantage with a discrete analog approach.

Eventually I think there will need to be a standardized buss. And yes you'd have to replace virtually all components to accommodate the differing CPU's, but in this world of recycling electronic circuit boards, the circuit tracings would be one less item on a mobo that would otherwise go to the landfill.

The problem is as soon as you've set a standard, you've locked in your performance. Nobody out there is saying "things are fast enough now". I'm not sure as I understand your last part. Are you saying we will go back to stuffing chips into sockets? I don't think so. Sockets add to the wiring problems. If you're not stuffing chips into sockets, then you've still got PC boards to recycle. The whole toxic disposal issue will have to be dealt with regardless what happens with motherboards. I'm sure it will end up being cheaper to change out boards then to go any other route.

Besides, the route we're headed for is whole computer on a chip. I think the next generation of CPU has the memory on it. Main memory today is too slow because it's external of the CPU package. We've covered it up with caching, but that's got it's limits. I'll bet we end up with a dual level virtual memory. Main memory in the CPU is paged out to slower memory used today which in tern is paged out the hard drive.

[craigwatanabe]

Re: First AMD 4x4 Mobo Spotted

There's nothing analog about Base 5. It's just four more logic levels above base 2. How you incorporate five logic levels into digital is beyond me. But Base 2 is just 1's and 0's and even our fastest supercomputers rely on serially turning shift registers on (1) or off (0) at some breakneck speed to get your CPU frequency. Base 5 allows not only 1's and 0's but 2's, 3's and 4's.

When I asked one of those guys in the chow hall all he could say with a smile was , "Confidential...need to know", and as he walked away, he said slyly, "Quantum Physics...research that one" and walked out. Remember, this was back in 1980.

[GeckoGeek]

Re: First AMD 4x4 Mobo Spotted

There's nothing analog about Base 5.

But you can't describe how it works either. The reason we use Base 2 is because it's VERY simple to represent. It's either on or off. High voltage, or low voltage. Either/or. Five is an unusual number since it it's not a power of two. Base 8 or 16 math is popular with computers because that's just 3 or 4 bits of Base 2. Only way I can see to transmit a Base 5 signal from one device to another is stepped voltage. It's either 0, 1, 2, 3, or 4 volts. But that's just a guess.

Let's just say that the CPU world is highly competitive. If there was an advantage, someone would start using it. The Soviets where in many ways behind. So they might have tried some tricks to bridge the gap, but I see no evidence they produced anything superior, even if the approach was clever.

[craigwatanabe]

Re: First AMD 4x4 Mobo Spotted

But you can't describe how it works either. The reason we use Base 2 is because it's VERY simple to represent. It's either on or off. High voltage, or low voltage. Either/or. Five is an unusual number since it it's not a power of two. Base 8 or 16 math is popular with computers because that's just 3 or 4 bits of Base 2. Only way I can see to transmit a Base 5 signal from one device to another is stepped voltage. It's either 0, 1, 2, 3, or 4 volts. But that's just a guess.

Let's just say that the CPU world is highly competitive. If there was an advantage, someone would start using it. The Soviets where in many ways behind. So they might have tried some tricks to bridge the gap, but I see no evidence they produced anything superior, even if the approach was clever.

No it wouldn't be voltage it would be states of on-ness or offness. There is actually three states to a binary bitstream. Zero represents the off state but there is also a quiescent state where it's neither on or off.

Something tells me Base five as it relates to quantum physics allows for polychronic computation for any observed or preconceived notation. In other words the computation has already happened and is awaiting the moment in time for it to be realized. Five levels of computation occuring at the same time, interacting at any given moment as the point of observation allows.

This would allow for concluded computation at the point of inquisition. You can't get any faster than that, and this would happen at five levels. Right now the closest thing to this scenerio is parallel computing but even that is in the binary domain.

Regarding base 8 or 16 that's not exactly correct. We do have Hexidecimal which is just placement values for a Decimal or base 10 system (A-F) but the computation is still done in binary thru Hexidecimal to Decimal or Hex to Binary decoders. It just allows for 25-bit words for a larger "Byte" in a bitstream. Typically in Binary one word is 8-bits or one Byte. But in Hexidecimal or Machine language you can attain 25-bit words using hexidecimal 0-9 then A-F.

[craigwatanabe]

Re: First AMD 4x4 Mobo Spotted

Oops base 16 is hexidecimal...my bad

[GeckoGeek]

Re: First AMD 4x4 Mobo Spotted

There is actually three states to a binary bitstream. Zero represents the off state but there is also a quiescent state where it's neither on or off.

Yes, there is a tri-state "state", but it doesn't represent anything. It's simply a means to allow someone else to control the buss. Data-wise everything is 0 and 1. Voltage-wise, well the real world is always analog, with reactance, impedance, voltage, current, rise-time, etc. But we try to push everything into one of two states to make circuit design easy.

Something tells me Base five as it relates to quantum physics allows for polychronic computation for any observed or preconceived notation. In other words the computation has already happened and is awaiting the moment in time for it to be realized. Five levels of computation occuring at the same time, interacting at any given moment as the point of observation allows.

I don't see what that has to do with base 5 (digits 0-4). Whatever, it doesn't sound like anything we're going to see anytime soon.