Vista bad!

[Ishmael]

What I saw of Amiga before it went underground and vaporized looked like they had a clue when it came to hardware and software. Can't say for sure because I never bothered to own one, just had one roomate in college who thought they were the cat's meow.

Never had any dedicated software support and it wasn't C-64 compatible.

Great design, suck ass marketing.

Ishmael

 

[linuxgeek]

Never had any dedicated software support and it wasn't C-64 compatible.

Great design, suck ass marketing.

Ishmael

hmmm .. sounds like a Fileserver software company I've heard of.

 

[Ishmael]

hmmm .. sounds like a Fileserver software company I've heard of.

It's a little easier these days with cross-compilers.

Ishmael

 

[linuxgeek]

It's a little easier these days with cross-compilers.

Ishmael

Novell's biggest problem has always been marketing. Is why I worried what would become of SuSe once they had it.

 

[Acanthus]

The last truly inovative hardware architecture offered to the general public was the Commodore Amiga. It took advantage of the 6800's symetrical clock to interleave CPU and Display processes from a common memory. No one really knew how to take advantage of it.

The basic PC and the Apple products have trumpeted big breakthroughs, but as far as hardware architecture is concerned they are only incremental upgrades.

The big advances have been in the O/S's and although the relative quality can be debated, that's where the rubber has really met the road. I happen to think OSX is superior to anything Billy has churned out. But the fact remains that both do what the users want them to do with enough reliability to hold market share.

Ishmael

Multi core is going to get very big in the next few years, highly parralel software is going to be an enormous increase in performance.

Intel is planning retail 80 core CPUs by 2010, and has a large project called "Mitosis" that is supposed to be able to optimise plain code for multicore on the fly with a special compiler and a mitosis enabled CPU.

They have already made a proof of concept 80 core cpu, but its far too large for retail now, they have to shrink it at least 2 more times, and the cores were very primitive in nature.

 

[Ishmael]

Multi core is going to get very big in the next few years, highly parralel software is going to be an enormous increase in performance.

Intel is planning retail 80 core CPUs by 2010, and has a large project called "Mitosis" that is supposed to be able to optimise plain code for multicore on the fly with a special compiler and a mitosis enabled CPU.

They have already made a proof of concept 80 core cpu, but its far too large for retail now, they have to shrink it at least 2 more times, and the cores were very primitive in nature.

I understand and it's still only evolutionary, not revolutionary. Merely the time line extension of "Moores' Law."

Revolutionary will be 'quantum computers.'

Ishmael

 

[Acanthus]

I understand and it's still only evolutionary, not revolutionary. Merely the time line extension of "Moores' Law."

Revolutionary will be 'quantum computers.'

Ishmael

You mean the one they built in Canada?

 

[linuxgeek]

Multi core is going to get very big in the next few years, highly parralel software is going to be an enormous increase in performance.

Intel is planning retail 80 core CPUs by 2010, and has a large project called "Mitosis" that is supposed to be able to optimise plain code for multicore on the fly with a special compiler and a mitosis enabled CPU.

They have already made a proof of concept 80 core cpu, but its far too large for retail now, they have to shrink it at least 2 more times, and the cores were very primitive in nature.

The biggest problem with multicore is finding programs which actually need to and/or can take advantage of it. More CPU cores in a single machine is basically the same as building a cluster of that many machines. If there isn't anything that needs that much power, its mainly a novelty and waste of resources.

Additionally, at least with a cluster, all the cores aren't fighting to share the same slab of RAM.

 

[Acanthus]

Theres still big things, hypertransport, ddr, holographic storage, CSI, Nmos cache chips, nanotube thermistors, etc.

 

[linuxgeek]

I understand and it's still only evolutionary, not revolutionary. Merely the time line extension of "Moores' Law."

Revolutionary will be 'quantum computers.'

Ishmael

speaking of QCs .. Have you seen any follow up articles if there has been any tangable proof the Canadain company really has a working one?

 

[Acanthus]

The biggest problem with multicore is finding programs which actually need to and/or can take advantage of it. More CPU cores in a single machine is basically the same as building a cluster of that many machines. If there isn't anything that needs that much power, its mainly a novelty and waste of resources.

Additionally, at least with a cluster, all the cores aren't fighting to share the same slab of RAM.

DDR3 by the end of the year

Games promise to benefit tremendously from the additional cores, especially for AI and physics. Video and Audio encoding are highly parrallel. This will also allow "system on a chip" designs, which is the direction AMD is headed. For one low price youll have your CPU, GPU, APU, and everything traditionally tied into the north and south bridge, all on one chip.

Itll drive PC costs through the floor.

 

[Acanthus]

speaking of QCs .. Have you seen any follow up articles if there has been any tangable proof the Canadain company really has a working one?

NASA backs it up.

 

[linuxgeek]

DDR3 by the end of the year

Games promise to benefit tremendously from the additional cores, especially for AI and physics. Video and Audio encoding are highly parrallel. This will also allow "system on a chip" designs, which is the direction AMD is headed. For one low price youll have your CPU, GPU, APU, and everything traditionally tied into the north and south bridge, all on one chip.

Itll drive PC costs through the floor.

I expect some of the technologies being worked on to replace electron pathways with photon pathways within computers will have a more significant effect than more processors in a single box.

 

[Acanthus]

http://www.dailytech.com/article.aspx?newsid=6450

 

[Acanthus]

I expect some of the technologies being worked on to replace electron pathways with photon pathways within computers will have a more significant effect than more processors in a single box.

Thats pretty far off on the horizon still, youre talking 2020.

 

[Ishmael]

speaking of QCs .. Have you seen any follow up articles if there has been any tangable proof the Canadain company really has a working one?

IBM had one working years ago. The issue is, and always has been, the programming and the I/O.

Possibly an advanced predicate based language would work. Afterall, how do you program a machine that actually does support a 'Branch on Maybe' instruction?

Ishmael

 

[OscarTheGrouch]

I remember my second computer. it ran well with a 40mb hard drive and 4meg of ram.

Granted computers can do more now, but I swear programmers have gotten lazier also..

 

[linuxgeek]

http://www.dailytech.com/article.aspx?newsid=6450

The scary part which I don't expect to be talked about much until the media until sweeps is if this Quantum Computer works, encryption as we know it is now useless. The Banks and any government which relys on encryption will be pissing on themselves.

 

[Acanthus]

The scary part which I don't expect to be talked about much until the media until sweeps is if this Quantum Computer works, encryption as we know it is now useless. The Banks and any government which relys on encryption will be pissing on themselves.

Yeah, there's been a few discussions on that over at anandtech.

Basically all encryption could be broken in days or even minutes.

 

[linuxgeek]

IBM had one working years ago. The issue is, and always has been, the programming and the I/O.

Possibly an advanced predicate based language would work. Afterall, how do you program a machine that actually does support a 'Branch on Maybe' instruction?

Ishmael

The last I was keeping track of the progress on them was when I was still teaching. The offical reports to the public only claimed a 7-qubit one so far.

 

[Ishmael]

The last I was keeping track of the progress on them was when I was still teaching. The offical reports to the public only claimed a 7-qubit one so far.

Sounds about right. The IBM proto was a 5 qubit machine.

And the interesting thing here is that the leaps seem to be in the form of primes. It'll blow the hell out of "Moores Law."

Ishmael

 

[linuxgeek]

Yeah, there's been a few discussions on that over at anandtech.

Basically all encryption could be broken in days or even minutes.

I would expect more like seconds. The only reason why encryption like public key-private key is considered unbreakable is because current tech would take years to finally guess the prime number with the size they are using.

In theory, with a QC, even encryptions which use mulitple passes of encryption would be breakable --may be the minutes to days set -- since all answers are given at once.

The funky part in my head with QCs is with what I've read, it appears once they are working, they just work. No power to cut. The way you turn them off is to break them.

 

[Acanthus]

Sounds about right. The IBM proto was a 5 qubit machine.

And the interesting thing here is that the leaps seem to be in the form of primes. It'll blow the hell out of "Moores Law."

Ishmael

Moores law really only applies to transistor count, not performance.

 

[Acanthus]

I would expect more like seconds. The only reason why encryption like public key-private key is considered unbreakable is because current tech would take years to finally guess the prime number with the size they are using.

In theory, with a QC, even encryptions which use mulitple passes of encryption would be breakable --may be the minutes to days set -- since all answers are given at once.

The funky part in my head with QCs is with what I've read, it appears once they are working, they just work. No power to cut. The way you turn them off is to break them.

I really havent followed them all that much, but the cooling does rely entirely on electrical energy (usually laser).

 

[Ishmael]

Moores law really only applies to transistor count, not performance.

Well, qubits are related to atomic shell states. And more precisely to atom count. All of which is related to performance.

Ishmael