Intel to break the 4GHz barrier in 2008

[FrankJH]

No it doesn't, only light travels at the speed of light.

The problem is you have to make a transistor which works with light and thats kinda difficult.

Exactly right - or as IBM have done slow down the speed of light until a more conventional transistor works (if I recall it correctly)

Of course IBM dont do desktop processors , I dont think they have for around 15 years or more - but still an interesting development none the less

 

[muon]

No it doesn't, only light travels at the speed of light.

Doesnt it depend on whether you're talking about drift velocity or the transmission of electricity? I believe the transmission of electricity is indeed at the speed of light.

Thats pretty irrelevant anyway since electricty can be easily controlled. The IBM article is about controlling the speed of light so that they can be used with gates in transistsors? Well thats what i thought it meant.

 

[Mitch007]

Intel along with other CPU manufacturers have a product roadmap, interlinking all sorts of technologies. If we had visibility of this PCI-E, single to dual to quad core, northbridge/southbridge chipsets and where this all comes together would fall into place.

Also dual/quad core CPU's are developed to provide a broad range of solutions accross many applications not just gaming. But game developers will be taking advantage of dual/quad core cpu's in their code writing, utilising the features and benefits. I bet soon some games will perform poorly on single core CPU's and run seemlesly on duals.

 

[Mitch007]

'The problem is you have to make a transistor which works with light and thats kinda difficult.'

What about opto isolators, they use light from an LED to swich a transistor... only they dont work that quickly... but they work on the principle. Great devices for isolating noisy supplies.

 

[manveruppd]

Bloomfield will slip into new B sockets—land grid arrays with 1,366 pins instead of the 775 in Intel's current desktop LGA socket.

When will this socket be introduced? Will it be S775's direct successor, or will there be an intermediate platform in between?

Shall we assume that, with the introduction of Socket B, there won't be any more CPUs released for S775, or will Intel keep milking it for a little while longer? I'm asking because it's an issue for anyone jumping onto the Conroe platform within the next year: knowing that there'll only be another year of new chips being released for your motherboard will limit your upgrade path substantially and make buying a Conroe a slightly less appealing prospect - though I'm not sure if it would be enough to warrant investing into the slower AM2/AM2+ platform in the hope that it will stay around longer as a platform and eventually outstrip S775's CPU offerings.

 

[Joe42]

'The problem is you have to make a transistor which works with light and thats kinda difficult.'

What about opto isolators, they use light from an LED to swich a transistor... only they dont work that quickly... but they work on the principle. Great devices for isolating noisy supplies.

They use a special sort of transistor called a photo-transistor which seems to be triggered by light but its apparently quite slow. I would imagine what it actually is is an ldr strapped to a transistor somehow. One way or another the transistor has to convert the light back to electricity before it can work and in a processor this would make it far slower than just using electricity.

So in an opto-isolator its not a transistor working with light, its converted to electricity first i think.

 

[Starfall]

It'll probably be for clock distribution around the die, to do this currently accounts for quite a large amount of power used by the CPU.

The speed of light depends on the medium it travells through. Wave propagation on a transmission line is usually quite a lot slower than the speed of light in a vacuum.

IBM have "done" CPU's and still do, just not directly.

 

[peetee]

that dgmug guy - genious.

 

[FrankJH]

IBM have "done" CPU's and still do, just not directly.

What do you mean not directly? I know they have had a hand in the design and production of the Cell processor for the PS3 (as direct as you can get with two partners of Toshiba and Sony themselves)

You have have misread my post saying that IBM havent been doing desktop CPU's for a while , but I should have been more accurate and stated x86 style desktop.

 

[Dravic]

Well it looks like valve have some interesting stuff coming up for their multithreaded version of the source engine.

Will be interesting to see how much differance it makes.

http://techreport.com/etc/2006q4/source-multicore/index.x?pg=1

 

[peetee]

i hope a few more developers follow valve's lead. They seem to be some of the only people to be trying to come up with new ways to impress us. Be in through gameplay features or coming up with new ideas.