Future thoughts

Mephistopheles · Jul 15, 2003

Why not 64 bit? I think it's most likely.

Also, 6Ghz might be a little too slow for clock rates. 4x400Mhz is good enough, though.

Mephistopheles

varghesejim · Jul 15, 2003

-----------------------------------------
Why not 64 bit? I think it's most likely
---------------------------------------------
Me too.I never told I am not expecting 64 bit

imgod2u · Jul 15, 2003

Both radeon 9x and geforce fx are 256 bit GPUs.I desktop computers there is not a 512 bit GPU yet.

Uh, yes they are. If you follow the standard convention for naming "bitness" of a GPU (traditionally, this has been the total width of the rendering pipeline), the NV3x and R3xx series *are* 512-bit by definition. NV30 and NV35 have quad-128-bit (4x 128) vertex pipelines and an 8-way 128-bit (8x 128-bit) rendering pipeline. In pre-DX9 software, it uses a legacy part of the card which is still 256-bit, however, the GPU was built for DX9 and under the DX9 standard, it has a 512-bit vertex shading pipeline and a 1024-bit rendering pipeline.

Similarly, the R3xx series is 8-way 96-bit for the rendering pipeline. Its vertex pipeline is still a 4-way 128-bit vector pipeline which makes it 512-bit (unless you want to go by the traditional means of counting the rendering pipeline, which would make it 768-bit).

Either way, the card is not 256-bit. Only half the rendering pipelines may be used for pre-DX9 software, but the hardware is still there. The vertex pipelines are all 512-bit and the rendering pipelines are even more than that.

"We are Microsoft, resistance is futile." - Bill Gates, 2015.

DrJeckyl · Jul 15, 2003

Check <A HREF="http://news.nationalgeographic.com/news/2003/02/0224_030224_DNAcomputer.html" target="_new">it</A> out...

--the Doc

slvr_phoenix · Jul 15, 2003

Well AMD will definately have a 64-bit CPU. I mean hell, they have one now. Heh heh.

But Intel ... I don't know. I'm really not convinced that Intel even cares about moving the desktop platform to 64-bit. And if Intel doesn't care, will there even be enough software for the A64 concept to catch on? I'm doubting it.

It's pretty much in the bag that the A64 performs about the same as an AXP of equal clock at 32-bit. It's pretty much in the bag that the NorthwoodC already kicks the pants out of an AXP of equal rating. Toss Prescott into that mix, which has even more IPC enhancement. Intel will rock the 32-bit performance world. AMD definately won't.

Likely, AMD's only hope for x86-64 to have even close to the performance of a Prescott is if lots of software developers write special x86-64 versions of their softwares. Intel's best hope of keeping that from happening is simply not supporting x86-64 in its own processors.

Why write two versions of every product when the A64 runs 32-bit code and when it runs so much better on the Intel anyway? That's what I imagine the mentality will progress to.

And lets face it, there's hardly any reason to switch to 64-bit computing for home use. So my expectation (not hope, but expectation) is that Intel will refuse to aknowledge x86-64 and will remain at 32-bit. Software developers will mostly target their development at the largest customer base (Intel's 32-bit processors) and AMD's A64 just won't catch on. So three years from now, I expect that Intel won't have a 64-bit CPU and AMD just might give up on one for SOHO use and aim x86-64 squarely at servers and nothing more.

I readily admit that I could be wrong. I mean what isn't a wild guess at best here from any of us? That's just what I expect to happen.

"Yeah, if you treat them like equals, it'll only encourage them to think they ARE your equals." - Thief from <A HREF="http://www.nuklearpower.com/daily.php?date=030603" target="_new">8-Bit Theater</A>

Mephistopheles · Jul 15, 2003

Intel's best hope of keeping that from happening is simply not supporting x86-64 in its own processors.

Bingo. I think so too. So if they have any 64-bit support in Prescott, of course it should be disabled! This would make AMD's life now that much harder... Well, we'll see...

Mephistopheles

HardWareBoss · Jul 16, 2003

I want one of those slow 6Ghz Chips for my puter.

david__t · Jul 16, 2003

I think that chip design will have to take a new direction pretty soon - I mean they can only keep reducing the die size so much and change the fab process so much before you reach the physical limit of production costs / possibility. I mean Intel has gone from no heatsink on a 486 through passive heatsinks to 60mm fans and now 80mm fans as standard. We are not far away from the 200 watt dissipation barrier. Cooling is definitely going to have to improve over the next few years.

4.77MHz to 4.0GHz in 10 years. Imagine the space year 2020

Syndil · Jul 16, 2003

Good point. Eventualy there is going to be a barrier that cannot be crossed, and to progress, chip technology will have to change dramatically. It could be just as simple as changing from silicon to some other material, or we could start to see 3-dimensional chips, with layers of circuits stacked on top of each other. Of course, that would present a cooling nightmare.

And then there are <A HREF="http://www.cs.caltech.edu/~westside/quantum-intro.html" target="_new">quantum computers</A>. Not exactly light reading, but VERRRY interesting.

_________________________________________________
Armadillo[TcC] at Lanwar and MML

imgod2u · Jul 16, 2003

Current MPU's are already layers of metal stacked on top of each other. The Athlon XP (T-Bred B stepping and Barton) uses 9 layers of metal interconnects on top of a layer of silicon transistors. The Pentium 4 uses 6 metal layers of metal interconnects on top of a layer of silicon transistors.

"We are Microsoft, resistance is futile." - Bill Gates, 2015.

slvr_phoenix · Jul 16, 2003

Current MPU's are already layers of metal stacked on top of each other. The Athlon XP (T-Bred B stepping and Barton) uses 9 layers of metal interconnects on top of a layer of silicon transistors. The Pentium 4 uses 6 metal layers of metal interconnects on top of a layer of silicon transistors

Exactly. Not only that but the smaller the process the less voltage you need. The less voltage you need the less heat the CPU produces. So we can go all the way down to a process as small as nanotubes and keep the same method.

Plus if AMD or Intel really wanted to, they could just make a CPU with extra layers who's sole purpose is to take up as much space as possible on those extra layers without being connected to any voltage and to run vertical paths through gaps in other layers. You'd make the die bigger and more expensive, but you also increase it's ability to transfer heat considerably. Pair that with a good heatsink and you're doing peachy-keen on a Sunday afternoon.

"Yeah, if you treat them like equals, it'll only encourage them to think they ARE your equals." - Thief from <A HREF="http://www.nuklearpower.com/daily.php?date=030603" target="_new">8-Bit Theater</A>

DrJeckyl · Jul 16, 2003

I'm surprised you didn't comment on my link Phoenix...nanotech may be closer than you think :evil:

--the Doc

reever2 · Jul 16, 2003

What is with you people and these huge fsb's, it is totally unneccesary to have anything more than quad pumped and DRAM is not going to move any faster in the future than it is moving today

slvr_phoenix · Jul 16, 2003

I'm surprised you didn't comment on my link Phoenix...nanotech may be closer than you think

So much to do, so little time. Sorry. I've been preoccupied with preparing to build my new system. I read the link, I just forgot to post. Heh heh.

Nanotech will be nice if used for good. (And downright evil if used for even just not so nice things.)

DNA computers though kinda don't seem like all that useful. Quantium computers, sure. Nanotech, absoutely. An organic computer that runs on chemicals ... maybe. More likely cross typical nanotech hardware concepts with DNA computers and make a tech/organic hybrid. That'd have potential.

In any event, as much as I'd like nanotech and that kinda stuff, I figure that the world has got to end or something before any of this happens or else we're going to enter an age where humans are gods and that'll get messy. So something funky will have to happen before and/or at the beginning of because I'm a pessimist and just don't think man can handle the responsability that'd come with that kind of power.

Though I still want a direct neural interface to a PC and/or a computer in my head.

That'd just be fun.

Speaking of weird DNA things, I heard that Japan (I think it's Japan) has just gotten itself a new fish. A zebra fish crossed with jellyfish DNA so that it glows in the dark. Wicked. I want one.

"Yeah, if you treat them like equals, it'll only encourage them to think they ARE your equals." - Thief from <A HREF="http://www.nuklearpower.com/daily.php?date=030603" target="_new">8-Bit Theater</A>

Syndil · Jul 17, 2003

Current MPU's are already layers of metal stacked on top of each other.

I stand corrected! I was not aware. Wonder if we will ever see enough layers to make a cube-shaped processor?

_________________________________________________
Armadillo[TcC] at Lanwar and MML

Mephistopheles · Jul 17, 2003

I thought that would be the case sometime... Cube-shaped processors. Three-dimensional matrixes of components, maybe layered planes. Interesting idea...

Mephistopheles

DrJeckyl · Jul 17, 2003

Actually, even though it says DNA computer, it is much more than that. The data pathways present in the human body are thousands times faster than even the fastest computer out now. The idea behind this research is to find an organic alternative to current data pathways that would allow for an amount of "bandwidth between hardware" and "processing power" we could never achieve with current means.
I work in a bio-medical R&D lab and we have mostly been working with protein research recently, but we have done a lot of research on nerve tissue conductivity and regeneration. To put it more into perspective without getting into the actuall tech. details, think about having terabytes transfered over your FSB, PCI-bus, etc. There is almost no resistance over these pathways, allowing for almost an infinite scaleability. There is practically no thermal radiation, signal loss, or signal degredation over nerve axons encased in myelin.

Anyways, just some "Future thoughs".....

--the Doc

imgod2u · Jul 17, 2003

A cube shaped processor would not be very practical. Not only will the packaging be much more difficult, core integrity (easy cracking) and not to mention cooling would be a problem. I'd still say the best way to scale a processor is to use architectural methods to facilitate signal propogation.

"We are Microsoft, resistance is futile." - Bill Gates, 2015.

Syndil · Jul 17, 2003

Actually, check this out:

<A HREF="http://www.discover.com/apr_01/feattrap.html" target="_new">http://www.discover.com/apr_01/feattrap.html</A>

A processor that runs on trapped photons, aka light, instead of electricity.

"...in photonic integrated circuits, information would flow more rapidly and copiously than it does today— much more rapidly. An optical computer that processed information as light rather than as electricity could process trillions of bits per second. That's thousands of times faster than the one-gigahertz microprocessors in the most advanced computers today."

_________________________________________________
Armadillo[TcC] at Lanwar and MML

Flinx · Jul 17, 2003

I was thinking that electrons already travel at the speed of light.

Anyway there is hope!

The loving are the daring!

DrJeckyl · Jul 17, 2003

yeah, cool stuff Syndil!!
The idea of optical, fiber optic pathways, or even light traveling in a vaccuum has been looked into. This would be much faster than current computers, but light travels as a wave and a partical, still causing resistance, unless in a vaccuum and under ideal conditions. Of course the amount of resistance on a photon (light particle) can hardly even be measured. I think this form of advancement would come before organic pathways.
As for the question about electron speed, they do not travel at the speed of light, but they travel pretty damn fast...hehe. Now the energy released during an orbital change does travel at the speed of light. There are differences in speeds of light due to the different wavelengths in each spectra. But this is a computer forum, not a quantum mechanics forum, so we'll assume we're talking about the combined spectrum (white light).
Very cool ideas though, combining organic and non-organic materials; combining physical and nonphysical materials.
....looking forward to seeing what the future brings.

--the Doc

varghesejim · Jul 17, 2003

-----------------------------------------------------------I mean it might be possible to multiply computing power radically with a machine that uses more than two states.

---------------------------------------------------------
It might be possible.But such a development should take more than 10 years to be in the mainstream

varghesejim · Jul 17, 2003

But Intel ... I don't know
---------------------------
The reasoning is OK.But if there is some software and AMD is really showing its power with it?Defenitely games will benefit from a 64 bit CPU(If every software is optimised for it).

varghesejim · Jul 23, 2003

And with the introduction of 3GIO and highbandwidth technologies,I really would like to see solidstate devices only in the computer.Eagerly waiting for the replacement of harddisk......

Mephistopheles · Jul 23, 2003

I was thinking that electrons already travel at the speed of light.

Ooo, that's a subtle issue. Electrons have mass, and therefore cannot be accelerated to the speed of light. See, actual electrons only travel at a very few centimeters EVERY HOUR or so. However, the information they transport is effectively passed on with the speed of light. It's like a water flow: if you have a long water pipe filled with water and the water at one end moves, then you'll notice movement at the other end, even though it's not the same bit of water moving... Thing of a wave travelling on a string, that works too.

I hope you got what I meant...

Just a curiosity from physics.

Mephistopheles

Future thoughts

Distinguished

Distinguished

Distinguished

Distinguished

Splendid

Distinguished

Splendid

Distinguished

Distinguished

Distinguished

Splendid

Distinguished

Distinguished

Splendid

Distinguished

Distinguished

Distinguished

Distinguished

Distinguished

Distinguished

Distinguished

Distinguished

Distinguished

Distinguished

Distinguished

Share this page