When will desktop speed reach tera-hertz(or whatever after GHz)

[zioxide]

They're not. Processor power is going to increase by # of cores, and all apps will be able to use multi-core efficiently. Look at "supercomputers" now, they're all multiple computers connected by fiber channel, etc. Power isn't going to increase by Ghz, its going to increase by adding more cores. In ten years, you'll start seeing processors that are clocked by the number of cores, not the speed. It will be like 512 cores or something.

 

[yg17]

.

That link doesn't show pictures of the cooling setup. I'd imagine it is something completely impractical for your average user.

The thread title mentions desktop speed. By desktop, we're looking for something that you'll find in a cheap computer at CompUSSR and not overclocked either. Not some insanely fast server processor that costs the same as a small car, not something that runs so hot that it requires an insane cooling setup.

I think we'll max out at 4 GHz max. An emphasis is being placed on more cores and not faster speeds. We're being limited by laws of physics and not research and innovation.

 

[thunng8]

I had a Commodore 486 that ran at 22MHz, so yeah I remember those days

And yes I'm sure there are some old folks who can 1-up me there, but that was my first computer so

I can beat you easily. I had a IBM compatible XT that ran at 8Mhz. It had dual 360k floppy drives and 512kb of RAM date was approximately 1985.

By ~'94 I had a Pentium 100Mhz, then a K6-233 in ~'96 and ~K7-600 in '99. First mac wasn't until '03 with a 1ghz G4 ibook.

 

[thunng8]

I think we'll max out at 4 GHz max. An emphasis is being placed on more cores and not faster speeds. We're being limited by laws of physics and not research and innovation.

IBM are releasing POWER6 systems next month (most likely low or mid range first). Those systems will clock significantly higher than 4Ghz. Rumour is that max clock will be ~4.5Ghz at launch and a high end 5ghz model before the end of the year.

http://www.eweek.com/article2/0,1759,2124931,00.asp?kc=EWRSS03119TX1K0000594

 

[SMM]

I can beat you easily. I had a IBM compatible XT that ran at 8Mhz. It had dual 360k floppy drives and 512kb of RAM date was approximately 1985.

By ~'94 I had a Pentium 100Mhz, then a K6-233 in ~'96 and ~K7-600 in '99. First mac wasn't until '03 with a 1ghz G4 ibook.

In 1983, we had an IBM PC (XT's were not available yet). It had a memory upgrade from 64K to 128K. It ran at 4.77 Mhz. It had (2) 180K single-density floppy drives (I ran DOS 1.X). We needed MS to come out with DOS 2.0 before we could use the new 360K DD floppy drives. I distinctly recall it took 8.5 seconds for each new line, when you typed, C:\DIR

 

[SMM]

To answer the question "When will desktop speed reach tera-hertz(or whatever after GHz)"

the answer is, in macs, several months after PCs

Noone can deny that some PCs have better processors than the current iMacs

I hate PCs but noone can deny that PCs are further in the hardware aspect, despite Apple's outstanding OS.

Jeez - I bet you were pacing, wondering if an opportunity to throw a barb at Apple would present itself. You sure had to reach for that one. Get lost, or try to show some respect for the others who were having a great discussion, which I found very informative.

 

[SMM]

In the 'muscle car' era, the rule was 'more cubic inches', the faster the car. Indeed, the top fuel dragsters were usually ~500 cubic inches and produced ~1200 HP. Over the years, designers learned that far too many other things contributed to HP and Speed. Today, engines are about the same CI size, but they produce considerably more horsepower. A top fuel dragster (or funny car) produces in excess of 7000!

The computer industry has gone through many evolutions, since I arrived in 1983. For me, processor speed is similar to belief that cubic inches were the key to performance. I will defer to CanadaRam and the other specialists, from whom I learn a great deal from. They provide expert knowledge.

I learned Unix from a chap who is among the leaders in Virtual Reality. He took me to the VR lab. They had banks of DEC Alphas, connected as a cluster. As I recall, they had around 50% computational power of current Cray model (which had 1000 Intel i860s). It was obvious, many processors were better (or more practical) than one huge one.

Multi-core, Multi-processor computing has not began to exploit the resources available. When you factor in the OS, and finally the applications, processor speed is not the issue. I can write a piece of **** application, and look like a genius. That is because the hardware will often allow me to disregard good programming technique. The user will not know that the .05 second difference even exists. But, the cumulative effect of many programmers doing this, produces things like ( fill in the blank ).

There are so many improvements to be made in system architecture, and software design, processor speed could be viewed as a distraction, rather than a goal.

 

[fairnymph]

I don't think we'll reach that point, or if we do, the speed increase will not be significant in daily use.

I think we'll switch over to a different FORM of computing by then, either quantum, neuronal-type networking, or some new technology. As others have said, we are reaching some fundamental limitations with the current tech, so something new must emerge.

 

[janey]

That link doesn't show pictures of the cooling setup. I'd imagine it is something completely impractical for your average user.

iirc, it involved quite a bit of ln2 (fun). completely impractical except to seriously mess around OCing stuff.

However, more sane methods of cooling that are better than just air aren't completely impossible - liquidcooled computers aren't exactly rare anymore.

 

[Swarmlord]

With current technology, never. It's all going to be parallel processing units and reducing the heat load.

 

[Shadow]

We are getting to the point now that the CPU is not the most important component in terms on preformance. And what I mean by that is that a machine can only go as fast as the slowest component-and now that is generally the hard drive. The hard drive is the current bottleneck. Its not about how much processing power we have anymore (like it was in the 80s and 90s) but more of how much data we can access and how fast we can get to it (ie, hard drives).

By the time we need processors with more than 10GHz (assuming that they can only get more efficient), we will have progressed onto Quantum and DNA computers.

Of course, I may just be talking crap here .