The IDF Thread (Fall 06)

The way the title is written its like you're writing the time period in which it is taking place.

Just so people know, I am taking the Piss.

 

ahhhhhh ok, thanks for the enlightment.

please define IP in this context.

 

intellectual property

 

Excuse me but i have to call BS on that statement. For the last few generations Intel, AMD and IBM have pushed each generation of technology until it was broken... then they pushed it some more.

I'm sure the parallell stuff will come but I dunno about the timeframe, i mean, in terms of hardware seems to me that serial interfaces are making a come back. SATA, PCI-Express and FB-DIMMS for example.

And then the software side, which is where the real multi-core gains will come from, is going to drag it's heels all the way. Hardly any consumer software even has optimisation for 2 processors.

 

As for the communications between cores, and using a network-like structure - isn't that how PCIexpress works already? So, to a certain extent, the technology already exists, albeit not at the speeds necessary for inter-core communication.

 

In regards to the You Tube bandwidth argument, isn't what intel is saying that through more powerful processors for post processing and decoding and more efficient algorithms the quality of the videos will increase. Whilst moving to high def would increase the bandwidth if all things were equal in terms of codecs used etc. But if more efficient algorithms are used bandwidth needed can be cut down albeit at the cost of needing more processing power to decode, or the quality of the video can increase at the same bandwidth levels.

 

Maybe Paul is using the same stuff Steve Jobs is currently using. But I like the idea at the end

genesisofthesith: I see your point but compression can only go so far. The gap in quality is WAY too big at the moment. But then again I could be wrong - I was once before

 

Well if The AA now call it fall in their TV adverts perhaps the UK is heading this way? .........

Oh and people messing around in their bedrooms in HD, whoop whoop I can't wait

 

Hmm... youtube in HD? How? Aside from the fact that they really need to move up to LD first (and then SD would be nice after that), the only people with cams and whatnot that can shoot in HD are people who are doing professional work. IIRC, the 4/4/4 1080p cam that MacBreak uses costs something like $50k, not to mention the fibre channel RAID array it's hooked up to in order to have enough recording bandwidth.
AFAIK, youtube doesn't have a Google Video-esque option to charge for "real" content - the only place you'd find HD material from anyways.

 

Any pictures or videos from this game? Oh Oh Oh, i want to test a quad core For drul purposes only

 

With the virtual PCI-E, look here forvisual aid.

 

I have to ask, why is all of the articles discussions being pushed into one thread?

It makes it a pain in the butt to be jumping around from one topic to the next.

 

Mmm... that always has bugged me a bit with these IDF/event threads in the past as well. Really hard to follow things.

Just curious though... is the quad-core ready game (among others) scalable beyond quad, or are game engines written to take advantage of only up to x processors/cores?

 

bye bye PhysX, you will not be missed.

 

Cool, Kentsfield on a normal 975x motherboard with BIOS updates, would mean i dont have to buy a new motherboard when i buy allendale this week

Sam

 

I think that once programmers get useed to parrallel programming, it will be able to scale. As they would be creating for a wide audiance, it would have to work on a wide range of equipment, so it should be able to utilise all cores regardless of how many there are. That is if the code is written well

 

saddly "should" does not equal "will"
If I get a BIOS upgrade that will allow me to use the 45nm quad core I'll be happy and surprised (this would be a change from the Intel I know (to whom new chip = same socket but new chipset))

Good question, and one I want an answer for too.

 

Regarding the "teraflop on a chip" concept intel showed:

1) Any idea what sort of timeframe intel expects this to come out in?

*Edit* Found this quote on Engadget:

...so, it's 32nm by 2010, and likely 80 core teraflop chips around late 2011.

2) Does anybody have a link to a good timeline indicating the flop-abilities of, say, the pentium I-IV line and core2duo from intel, along with perhaps the Athlon line of chips from AMD? I was hoping to figure out how much of a jump this would be in the historic progression of processor power. My quick google and wikipedia searches didn't give me the information I was looking for....although I did learn that high-end gpu's already claim 1Tflop or better (though that is based on specific tasks and not general purpose).

 

Call bs all you want mate! Was the architecture a bad move? Yes could it have been better? yes was the scalability there to reach 10GHz? Yes! And thats what this is about. However due to how crappy and inefficient netburst was we'd be running chips that gave off massive amounts of heat. You would need something like a prometeia mach ii to cool it, but the bottom line is that it could be done!

What the hell do you expect, its a new technology you cant expect the best software to be released and for people to take to it right away, these things take time. If people took to every new techology without holding out for a while to see how it performs we'd all be running physx cards!

 

Wow so Intel plans to just keep throwing in cores. Man I guess my children we'll just buy the whole wafer with pins on one side.




LOL I remember when I moved to US and said Autumn the people were like what...? Many here don't even know that it is synonymous to the word fall. =)