If makers really start selling CPU's with lots and lots of cores, unless something changes drastically most of their users will find that even at their busiest, most of the cores will sit idle.
Assuming the makers don't count on this by underdoing the cooling and power supply arrangements, this set of users might be able to donate more Flops more cheaply than any previous group of volunteers--possibly enabling Grid Computing to go much further.
It was obvious at least two decades ago that increments of processor complexity at that stage were providing less than linear performance improvement. It is done nothing but get worse since. It was never in doubt that multi-core could provide more computer/dollar or per watt, or per unit design time, or lots of others pers. What was (and still is) in doubt is whether in real normal use enough of these advantages will trickle through to the user for it to make sense.
Alternatively, the merchandising of these things may take leave of reality as firmaly as has the megapixel race in low-end consumer cameras.
Well, if that 80-core chip Prototype was really upto 5600MHz, it can mean only one thing :
They've created a proprietary design study with extremely limited functionality. It wouldn't surprise me if that Core was hard-wired to perform nothing but a simple task and the resulting performance being entirely theoretical.
It'll be another two decades before we'll see anything close to general-purpose 80 cores, let alone at reasonable power consumption.
The Information is just what it is... Propaganda. But it shows what they're trying to do in the far future (which is no surprise though)
Well, if that 80-core chip Prototype was really upto 5600MHz, it can mean only one thing :
They've created a proprietary design study with extremely limited functionality. It wouldn't surprise me if that Core was hard-wired to perform nothing but a simple task and the resulting performance being entirely theoretical.
It'll be another two decades before we'll see anything close to general-purpose 80 cores, let alone at reasonable power consumption.
The Information is just what it is... Propaganda. But it shows what they're trying to do in the far future (which is no surprise though)
Greetings:
I agree with you wholeheartly. What I find amazing is they are tring to realize this performance in silicon. Jerry Bautista, director of the tera-scale research program at intel says 'the company has no plans to bring this exact chip designed with floating point cores to market, chips with 20 to 40 cores could hit the market within a decade.
There are some who can live without wild things and some who cannot. - Aldo Leopold
If I'm reading the article correctly the chip had generalized cores since ASIC versions were mentioned as possible future research efforts. It looks like the way they achieved the density they did was to only use 5k of cache per core instead of the several megs of thier C2D line. It probably was however a pure risc design instead of an x86 compatable unit (to save on the cisc to risc converters).
Intel 80 Core Chip in Research Mode
)
If makers really start selling CPU's with lots and lots of cores, unless something changes drastically most of their users will find that even at their busiest, most of the cores will sit idle.
Assuming the makers don't count on this by underdoing the cooling and power supply arrangements, this set of users might be able to donate more Flops more cheaply than any previous group of volunteers--possibly enabling Grid Computing to go much further.
It was obvious at least two decades ago that increments of processor complexity at that stage were providing less than linear performance improvement. It is done nothing but get worse since. It was never in doubt that multi-core could provide more computer/dollar or per watt, or per unit design time, or lots of others pers. What was (and still is) in doubt is whether in real normal use enough of these advantages will trickle through to the user for it to make sense.
Alternatively, the merchandising of these things may take leave of reality as firmaly as has the megapixel race in low-end consumer cameras.
Well, if that 80-core chip
)
Well, if that 80-core chip Prototype was really upto 5600MHz, it can mean only one thing :
They've created a proprietary design study with extremely limited functionality. It wouldn't surprise me if that Core was hard-wired to perform nothing but a simple task and the resulting performance being entirely theoretical.
It'll be another two decades before we'll see anything close to general-purpose 80 cores, let alone at reasonable power consumption.
The Information is just what it is... Propaganda. But it shows what they're trying to do in the far future (which is no surprise though)
RE: Well, if that 80-core
)
Greetings:
I agree with you wholeheartly. What I find amazing is they are tring to realize this performance in silicon. Jerry Bautista, director of the tera-scale research program at intel says 'the company has no plans to bring this exact chip designed with floating point cores to market, chips with 20 to 40 cores could hit the market within a decade.
There are some who can live without wild things and some who cannot. - Aldo Leopold
If I'm reading the article
)
If I'm reading the article correctly the chip had generalized cores since ASIC versions were mentioned as possible future research efforts. It looks like the way they achieved the density they did was to only use 5k of cache per core instead of the several megs of thier C2D line. It probably was however a pure risc design instead of an x86 compatable unit (to save on the cisc to risc converters).