What Moore Can we do?

[X-Rated]

http://www.wired.com/thisdayintech/2...op+Stories+2))
Moore's law states that the number of transistors will double every year in a processor. Well, okay. Every two years now. But this was made on April 19, 1965 and it is holding pretty much true to date after 45 years.

What will the future bring? How will Bill Gates figure out what to do to eat up that transistor power generated by the hardware side of technology?

The world may never benefit from Moore's law. It seems to me that issues with computers have not improved that much over the last 20 years. I used to spend hours updating my computer, I spent many many seconds to bring it up and to shut it down. It is the same thing today. Nothing changes. The hardware is multitudes more impressive than what it was 20 years ago, but the software is moronic and wasteful to resources compared to what it was 20 years ago.

[KC2UGV]

I agree. It seems boot up time for computers is an inverse law to Moore's: Double the transistors, and double the boot times.

Tandy CoCo III Boot time: 0.5 Seconds
Today's computer boot time: 45-60 seconds

Some strides are being made in that arena, with Google actually leading the way, and they got it down to 6 seconds for booting.

A big area to trim down is much of the out-moded interfaces that are still brought online in your run-of-the-mill PC. I ask, why is the BIOS concerned with ISA cards, and starting the ISA interfaces (Which none exist on the board)?

If the board has no LPT ports, why check out the LPT interface?

Artifacts. Nobody has gone over the firmware for these boards yet and gut the stuff that is not needed. Just add more!

Same goes with OS's. Don't gut old stuff, just add more.

Dungeons Of Daggorath, a 99-level game, that was very addicting: 7K. Less used to program the entire game than what a JPEG occupies. It even had game-saving (Rudimentary DOS).

Why does an OS, which is only used to activate programs taking up 8GB of space? 8GB of API and eye candy? There is waste there. Much of it.

[n2ize]

I agree. It seems boot up time for computers is an inverse law to Moore's: Double the transistors, and double the boot times.

Tandy CoCo III Boot time: 0.5 Seconds
Today's computer boot time: 45-60 seconds

Are you talking about time required for the BIOS to boot the hardware into a usable state or, the total time for both HW and operating system to boot ?

http://Some strides are being made i...LPT interface?

Can't you turn these off in the BIOS ?

Artifacts. Nobody has gone over the firmware for these boards yet and gut the stuff that is not needed. Just add more!

Same goes with OS's. Don't gut old stuff, just add more.

Backwards compatability ? To enable certain OS's to run older hardware. The OS can be streamlined by not loading stuff that you don;t need.

Dungeons Of Daggorath, a 99-level game, that was very addicting: 7K. Less used to program the entire game than what a JPEG occupies. It even had game-saving (Rudimentary DOS).

No doubt, media files have grown HUGE. Start playing around with video formats and watch how fast a 500 gig or 1 Tb drive gets consumed.

Why does an OS, which is only used to activate programs taking up 8GB of space? 8GB of API and eye candy? There is waste there. Much of it.

Fancy desktops with all sorts of bells and whistles. In the old days we did everything from the CLI. :D There are some issues with software design. Not all software is well designed. There are some "programmers" who are good at "spaghetti code" but cannot design efficient algorithms nor analyze run times.

[KC2UGV]

I agree. It seems boot up time for computers is an inverse law to Moore's: Double the transistors, and double the boot times.

Tandy CoCo III Boot time: 0.5 Seconds
Today's computer boot time: 45-60 seconds

Are you talking about time required for the BIOS to boot the hardware into a usable state or, the total time for both HW and operating system to boot ?

Both. The BIOS boot to OS time is around 10-15 seconds. 30 seconds to BIOS Handover to Usable interface.

Some strides are being made in that arena, with Google actually leading the way, and they got it down to 6 seconds for booting.

A big area to trim down is much of the out-moded interfaces that are still brought online in your run-of-the-mill PC. I ask, why is the BIOS concerned with ISA cards, and starting the ISA interfaces (Which none exist on the board)?

If the board has no LPT ports, why check out the LPT interface?

Can't you turn these off in the BIOS ?

Some you can, some you can't. You can't disable the non-existent ISA interfaces from getting checked. Others you can (LPT). But, the point is, if it doesn't exist on the board (COM1, LPT1, etc), why even have it in the BIOS?

Artifacts. Nobody has gone over the firmware for these boards yet and gut the stuff that is not needed. Just add more!

Same goes with OS's. Don't gut old stuff, just add more.

Backwards compatability ? To enable certain OS's to run older hardware. The OS can be streamlined by not loading stuff that you don;t need.

There comes a point where backwards compatibility can be shed. I know, keep the drivers there for ISA sounds card, 10-Base2 NICS, etc. But there is no reason to load them by default on boot-up.

Dungeons Of Daggorath, a 99-level game, that was very addicting: 7K. Less used to program the entire game than what a JPEG occupies. It even had game-saving (Rudimentary DOS).

No doubt, media files have grown HUGE. Start playing around with video formats and watch how fast a 500 gig or 1 Tb drive gets consumed.

Oh, I agree. But there is hardly any need for huge media files to be built into the OS, is there?

Why does an OS, which is only used to activate programs taking up 8GB of space? 8GB of API and eye candy? There is waste there. Much of it.

Fancy desktops with all sorts of bells and whistles. In the old days we did everything from the CLI. :D There are some issues with software design. Not all software is well designed. There are some "programmers" who are good at "spaghetti code" but cannot design efficient algorithms nor analyze run times.

And I think, with some clean coding work, on all fronts (Drivers, OS, BIOS, etc); a whole lot of speed can be saved.

[n2ize]

Some you can, some you can't. You can't disable the non-existent ISA interfaces from getting checked. Others you can (LPT). But, the point is, if it doesn't exist on the board (COM1, LPT1, etc), why even have it in the BIOS?

Even if they don't exist on the board can't they be added on, i.e. via a plug in board that provides that function ? Or , in that case would the emulation be provided soley by the add-on hardware and not be required to be loaded by the BIOS ? I'm not entirely sure on that one.

There comes a point where backwards compatibility can be shed. I know, keep the drivers there for ISA sounds card, 10-Base2 NICS, etc. But there is no reason to load them by default on boot-up.

It depends on the OS you are running. If you are using Linux you can usually remove a lot of this stuff from the kernel and reconfigure and recompile it. If you are on a prebuilt binary OS or Windows you're probably SOL.

Oh, I agree. But there is hardly any need for huge media files to be built into the OS, is there?

I agree. But it seems these days people expect lots and lots of fancy widgets, animations, and assorted jazz.

And I think, with some clean coding work, on all fronts (Drivers, OS, BIOS, etc); a whole lot of speed can be saved.

True. Although sometimes developers say otherwise. It also depends on the ability of the coder/designer to analyse/understand the run times for what he is designing. Does it run in O(1), O(n), O(n^3}. O(log(n)), etc.

A while ago a friend of mine was including a straightforward linear recursion (not iteration) into a sub-routine. He was actually encoding the recursion itself, i.e. to find f(n) the processor needs to find f(n-1), f(n-2), f(n-3), ... , f(0) which obviously for very, very large values uses lots CPU cycles. I showed him how to express the particular recursion in "closed format" (a freshman problem) , i.e. so if you want f(n) you plug n into a closed formula and crank out a value. Much easier on the CPU. Much easier on his customers when the product goes to market. I'm not saying all solutions are that simple but, the important thing is... don't overlook the obvious or the simple. Another problem I see is spaghetti code. Stuff being linked together with little regard for the efficacy of what is being spaghetti'd together. Of course project deadlines and project managers who drink too much caffeine don't help much either. LOL

[kf0rt]

The PC world went to hell when Marketing got involved.

You can quote me on that.

Almost everything in the mainstream is "bloatware" these days. Do we really NEED a new version of Word after all these years? Today's feature creep trumps cleaning up yesterday's disaster every time when the measure of excellence is the bottom line.

[KC2UGV]

The PC world went to hell when Marketing got involved.

You can quote me on that.

Almost everything in the mainstream is "bloatware" these days. Do we really NEED a new version of Word after all these years? Today's feature creep trumps cleaning up yesterday's disaster every time when the measure of excellence is the bottom line.

Uh, oh. You're treading into the area of "Unix Philosophy" :) Do one thing, and do it well.

[WØTKX]

We're going quantum, boieeee!

http://en.wikipedia.org/wiki/Photonic_computing

From opals to optical chips...

As electronic devices shrink and operating speeds increase, silicon chips are running out of room. Photons – light particles – are an obvious replacement for electrons, because they can carry more information in the same space.

Communications technology has been revolutionized by electro-optical devices based on the semiconductors gallium arsenide (GaAs) and indium phosphide (InP), optical fibers, and even all-optical amplifiers. But as PHAT spokesperson Gudrun Kocher points out, these devices tend to be much larger than the components needed to make computer chips. GaAs and InP are also expensive materials, and integrating them with silicon brings extra complexities. As a result, she says, most researchers agree that it will be 10-15 years before we see all-optical chips based on conventional (silicon) technology.

This is where photonic crystals come in. A combination of 3D photonic crystal optical switches and 2D waveguides could yield devices that are 10 or even 100 times smaller than those made at the moment. These could be used to assemble all-optical chips made entirely from silicon.