Home computers

[Azat]

I was just reading the following article in PC World(http://www.pcworld.com/news/article/0,aid,106415,tk,dn102802X,00.asp) on the new processor that is coming out in a week and I started to wonder where we will be in the near future with technology.

 

Yesterday - 1993

Intel in 1993 introduced the Pentium Processor. It was a 60/66 MHz processor with 3.2 million transistors on it.

Most high end PC of that day had a single 500-1000 MB hard disk in it. 16 MB of RAM. And they had a 256 KB video card. 14.4 Modems were common and some lucky people had the 28.8 modems to connect to CompuServe or to AOL or Prodigy or some bulletin boards.

 

Today - 2002

In a week(Nov 14th) Intel is going to release the 3.06GHz Pentium 4 processor. The processor has 55 million transistors on it.

Most high end computers of today come with a minimum of 120 GB hard drive, 512-1024 MG of RAM and 56KB Modem and a 64MB video card. This is the technology that is available to you and I at teh local store.

 

And lets not talk about price as I think I still owe money on that first $4000 Pentium 150 that I got in 95.

 

Tomorrow - 2011

So who cares to guess where we would be in 9 years from now?

[Sip]

Very simple! Just follow Moore's law ... double performance every 18 months. All indications from intel is that at least in the next decade, that will not be a problem. They are even working on a 10 nm manufacturing process for the chips ... right now, the best chips are somewhere around 180nm (0.18 micron)... that is the "feature" size ... basically the dimension of each transistor.

 

So by around 2009, you should expect to see 10-15GHz desktop pc processors readily available on the market. Disks also seem to follow Moore's law so we may see close to terabyte disks.

 

Dave Tennenhouse, the head of intel research was actually giving a talk today at UCLA... he has the feeling that disks (mechanical) are going to slowly disappear and be replaced by solid state devices (memories). And I don't think he is that far off ... we have a lot more computing devices without physical disks than we have ever had before.

 

The other thing is, we now have somewhere between a few to tens of processors per person (in the west). That number will soon grow to thousands of processors per person so that is going to be a HUGE change.

 

It's called ubiquitous computing and it's definitely here

[Sip]

Oh by the way, the 3.5 Ghz processors that you see, can actually be clocked maybe up to 7 Ghz! The problem is (1) heat and (2) manufacturing imperfections that will hinder the performance being that high. So as intel slowly matures each manufacturing process (the different steppings of each chip), the clock frequencies keep going higher and higher.

 

So it's not too uncommon for people to buy a 2Ghz chip and run it at 3Ghz with some fancy cooling devices successfully (overclocking).

 

The other thing that amazed me is that there are actually two ALUs (arithmatic logic units) inside each pentium 4 that run at twice the speed of the chip!!! So if you have a 3.5Ghz Pentium 4, those ALUs add/subtract integer numbers at 7Ghz!!! that's 7 Billion addition/subtractions per second, done in parallel, meaning potentially 14 BILLION addition/subtractions per second on ONE P4 chip.

 

The problem they have is that in most "systems", memories can not deliver data nearly that fast to keep those data hungry processors satisfied! So those ALUs sit idle (at least one of them) most of the time.

 

[ November 05, 2002, 10:05 PM: Message edited by: Sip ]

[Azat]

Sip, it is true we have followed Moores Law pretty closely for both CPU and Storage(I think actually storage has been a little faster), but I wonder how much more we can follow that law. I have heard that 7-9 micron might be the smallest we can get on the chip as anything smaller we start to get shorts as the electricity starts to travel on the outside of the transistor. I too have heard about 15GHz and about 1 billion transistors on the Intel chip of 2009.

 

BTW: Simple Moores law applied to CPU and Storage for the next 9 years would mean that the CPU would be running at about 200GHz and storage would be 7.7 TB and 32GB RAM. I find it hard that we would be there.

[Sip]

Oh yah ... having 10Ghz PCs on the market means probably 100Ghz ones are on the drawing boards and 200Ghz ones not too far away.

 

Ok, there is no doubt that there has to be a physical limit. Unfortunately, I am not that far down in the design/development process to be able to give you an authoritative answer. But as I said, at least in the next decade, all signs indicate that Moore's law is not going to be a problem.

 

The problem is that even with a 1Ghz processor, your PC is idle more than 99% of the time (that's an optimistic guess .. usually, resource utilization is much lower than that!). So there is really not going to be a need to push processors too much anymore. They are advancing much too fast for everything else to keep up (i.e. applications, peripherals, I/O devices, memories, disks, etc).

 

Having a 200Ghz CPU, although possible, is going to be useless if everytime you need data from DRAM you hit a 6ns pentalty!

 

Just imagine, only 6ns to fetch something from memory and already your CPU wasted 1.2 Trillion cycles just sitting and waiting!!!! Obviously that is not the power efficient thing to do ... power is the KEY Factor in the design process now. You don't want a 200Ghz cpu that is going to eat up more power than your toaster

 

So I think you will see much more sophisticated low power designs starting to show up. There of course will always be a small market for ultra-high performance stuff, but the mainstream is not going that way in my opinion (games maybe). But how many MegaFlops do you need to check your email and surf the web?

[Sip]

Oh by the way, in the 10nm process that intel is working on, the insulator on the chip is 5 atoms thick!!!!! Don't ask me how they get those atoms to stay there.

[Guest Fadi]

I am very skeptical of this 10-15 Ghz stuff, and if that is practical... what I mean, is just look at intel 64 Bits designs and the comming Hammer AMD processor, they have a lot more units working in parallels, I just guess that in future they will find it easier to just add the units in the processors and make them work in parallel, rather then reducing the process to less than 9 micron and multiplying the frequency.

 

5 atom thick ? That sound smaller I think then the switch on the new light microscope working on the principale of Vanderwall's attraction...

 

[ November 08, 2002, 10:32 AM: Message edited by: Domino ]

[MosJan]

do we even have Software / wich will work at this speed ??? each and every Software has it's own speed - jisht em te sxal ???

[Guest Fadi]

Mosjan, if you would have made that offer on eBay 2 years ago, Bill Gates would have made an offer, just because its expensive, it would replace his Lionardo De Vinci's notes.

[Guest Fadi]

I've made a fool of myself again(I think you guys are quite habituated of this )

 

My post above was supposed to answer that thread.

 

/cgi-bin/forum/ultim...c;f=30;t=000070

 

About the Hyeforum Collection.

 

[ November 08, 2002, 06:54 PM: Message edited by: Domino ]

[Azat]

An article from one of my favorite Technology researchers Jeff Harrow from The Harrow Group. He use to be one of the research scientists at Compaq prior to opening up his own consulting business.

 

Here is the link to the article: http://www.theharrowgroup.com/articles/200...16/20021216.htm

 

Speaking of Molecules...

 

 

From Diode to Diode

 

I remember, all too well, the old vacuum tube-infested, two-way radio transmitters that I maintained as part an early job. They were typically hidden away in South Florida belfries and rooftop construction shacks, laboring away in inhumane working conditions. Of course THEY didn't mind this steam room treatment (the sun cooking them from the outside, and their tubes cooking them from the inside), other than to fail (a lot) more often than they otherwise might. (Air-condition these spaces? You must be kidding...)

 

So everything I could do to reduce the internal heat generated by these boxes improved their longevity, as well as my disposition (I DID mind working in those conditions). A pretty good incentive.

 

Which is why I clearly remember when the first solid-state power diodes became available to replace the large and VERY hot 5U4 rectifier tubes.

 

http://www.theharrowgroup.com/articles/20021216/20021216_files/image004.jpg

 

Their new solid state replacements (shown below)

 

http://www.theharrowgroup.com/articles/20021216/20021216_files/image006.jpg

 

fit entirely within the BASE of a 5U4 tube, saving significant space. They generated virtually no heat compared to the tubes they were replacing. And, perhaps most impressive from my perspective, they almost never failed. That combination of less heat and essentially unlimited life meant that I didn't have to visit these ovens nearly as often. Which was a Good Thing.

 

Of course, not all of the then-newfangled solid state diodes were used as power rectifiers; some, like the legendary 1N34A,

 

http://www.theharrowgroup.com/articles/20021216/20021216_files/image008.jpg

 

were smaller than a grain of rice but did yeoman's work in receivers and many other types of circuits.

 

 

 

Fast Forward To Today: The Diodes Get Smaller. Way Smaller...

 

Unsurprisingly from today's vantage point, Moore's law continued to dramatically shrink diodes (and their cousins, transistors), now packing millions (and soon billions) into our unimaginably-complex chips.

 

Yet this gets better (or worse): Reader Dana Hoggart brings our attention to the work of chemists at the University of Chicago who now claim to have created a diode "...from a single molecule ... about 2.5 nanometers in diameter." (http://www.nanoelectronicsplanet.com/nanochannels/research/article/0,4028,10497_1474851,00.html)

 

Inventors Man-Kit Ng and Luping Yu indicate that they can now "...mass produce molecular diodes with relative ease." Yu believes that synthesizing molecular transistors (which are essentially two diodes back to back) will also be amenable to mass production.

 

 

 

...To Molecular Calculators!

 

But molecules are 'so big,' being made up of all those atoms -- surely, they can do more?

 

Indeed. An article at http://www.nytimes.com/2002/10/25/technolo...75cc7b2ea47bce9 (per reader George Daszkowski), describes how Dr. Eigler, the same scientist at IBM who first spelled out "IBM" in xenon atoms in 1989,

 

http://www.theharrowgroup.com/articles/20021216/20021216_files/image010.gif

 

has now demonstrated an entire working logic circuit (a "three-input sorter") that is 260,000-times smaller (12 x 17 nanometers) than the equivalent logic circuits used in today's integrated circuits!

 

To put this in perspective, 190-billion of these logic circuits would fit on the top of a pencil eraser. (One-nanometer, or one-billionth of a meter, is the size of 5 to 10 atoms lined up next to each other.) This is "small." Consider this picture from IBM's press release (http://researchweb.watson.ibm.com/resources/news/20021024_cascade.shtml) that shows the molecules involved.

 

http://www.theharrowgroup.com/articles/20021216/20021216_files/image012.jpg

According to IBM (http://www.research.ibm.com/resources/news/20021024_cascade.shtml), each logic block is made up of a few carbon monoxide molecules placed very precisely on a copper surface. The copper surface, at the molecular level, looks a bit like an egg crate, covered with tiny dimples, and the very bottoms of the larger carbon monoxide molecules rest in the "egg cups." But since most of the carbon monoxide molecule sits above its cup, a "cue ball" type of operation can cause all of the molecules to do a Chinese fire drill across the surface which, by some magic means, yields the answer to a calculation.

 

(You can view a fascinating animation of how this tiny logic block works at http://domino.research.ibm.com/Comm/bios.n...scade_small.wmv . The animation begins with an Atomic Force Microscope moving a molecule into just the right depression, and then shows the "cascade" of molecules that actually processes the information. )

 

It's not quite that simple, of course -- at this point, the whole ball of molecules has to be a couple of degrees above Absolute Zero and in a hard vacuum. Then, an Atomic Force Microscope has to laboriously place each molecule, initiate the molecular cascade, and then read-out the answer.

 

Even though we don't (yet) know how to turn this demonstration into practical molecular computers, it does join the "Ah Ha"s that continue to open scientists' minds to what they might next accomplish. According to Dr. James Heath, chemistry professor at UCLA,

 

"It's just a really interesting demonstration of how small you can get and still manipulate information... It's a beautiful piece of work. Don Eigler and his group work at the boundary between art and science."

 

Yes, there are many problems yet to be solved to enable the commercialization of these molecular logic gates, yet this dye may have already been cast. One day in the not too distant future, the idea of using anything bigger than single molecules to perform such a function might seem as funny as, well, using a large, ultra-hot, power hungry vacuum tube to turn AC into DC. Or of using mechanical relays as the components for logic gates. (Both of which were the case only 30 years ago...)

 

Similar future 'paradigm shifts' are going to be SO interesting...!

[Azat]

10.20GHz Intel Nehalem slated for 2005

 

Future Desktop Roadmaps Tejas to reach 9.20GHz, Prescott 5.20GHz

 

By Mike Magee: Wednesday 29 January 2003, 10:38

 

IRONY ALERT This story is paid for by INTC.

INCLEMENT WEATHER ON this side of the Atlantic ocean threw a turquoise parakeet off course today and a note it was holding in its beak fell into the INQUIRER's back garden.

 

The contents of the note appear to reveal future plans for future Intel desktop processors right up until 2005.

 

By then, according to the note, Intel will be able to deliver 10.20GHz desktop CPUs codenamed "Nehalem" and produced using 65 nanometer technology.

 

If Intel manages to migrate away from the 90 nanometer technology it will introduce towards the end of this year, by then the "Prescott" core will deliver at least 5.20GHz using the 800MHz system bus.

 

The immediate successor to Prescott after it tops out at 5.20GHz will be the "Tejas" core, also produced on a 90 nanometer process and delivering 5.60GHz using a 1066MHz system bus. That's slated to start appearing towards the end of 2004.

 

Tejas will increase in steady increments which appear to be 6GHz, 6.40GHz, 6.80GHz, 7.20GHz, 7.60GHz, 7GHz, 8.40GHz, 8.80GHz and topping out at 9.20GHz.

 

The first Nehalem is supposed to appear at 9.60GHz before Intel succeeds in its goal to produce a 10GHz+ chip, the Nehalem, and using a 1200MHz front side bus. µ

 

http://www.theinquirer.net/?article=7481

[Sip]

oooooooooo yaaaah

[Azat]

Sip, I sure hope that all those hearts are for Intel and not for me.

[Azat]

Check out the price of this ultra lite laptop.

 

It's worth getting one just for the heck of it.

 

http://info.lindows.com/mobilepc/mobilepc.htm

 

[ February 24, 2003, 09:46 AM: Message edited by: Azat ]

[MosJan]

Azat jan i have used a unity like this - but made by Dell - it was ok but - charcharanqa external Cd - ~~~~ ehh normal@ amenalavna -

 

http://pawnplex.com/Computers/x200/MVC-002F.JPG

 

[ February 24, 2003, 11:34 AM: Message edited by: MosJan ]