Intel Patents Redundant Cores In a Many-Core Processor

Status
Not open for further replies.

kjsfnkwl

Distinguished
Jan 28, 2011
65
0
18,630
The problem here is that enthusiasts will want to unlock those extra cores, which would undermine the entire system.
 

billybobser

Distinguished
Aug 25, 2011
432
0
18,790
I guess this would be a fail-safe against cores going down under warranty.

As I imagine activating the inactive cores will give little to no benefit.

Although I don't really see how this is patentable. Having redundant hardware to click in in case of failure?

I guess you should be able to get round but having all cores active when needed, and sleeping when not (as is possible today), and making it so a processor can carry on if a core dies mid process.
 

mcd023

Distinguished
Nov 9, 2010
370
0
18,780
I think that they're patenting the monitoring of temperature to activate and deactivate cores to absorb and dissipate heat, not just having the redundant one there. So, others might be able to measure load on the cores to do the same thing as opposed to measuring the heat. I think?
 

randomstar

Distinguished
May 2, 2007
259
0
18,810
think about a satelite or spacecraft - you do not want the extra cores active using power- but on a mission critical piece of equipment, calling a repair main a year away to replace a failing part might not be a good back up plan.
extend the lifetime and reliability of the whole endevor without having to have wholy seperate computers..
 
This is a terrible idea.

Why not ship the CPU with all cores active and give it a 'soft fail' feature for failing cores?
By 'soft fail' I mean that a failing core could be dynamically deactivated while allowing all other cores to function normally.

This would allow you to have higher initial performance and give you uninterrupted computing in the case of a core failure.
 

saturnus

Distinguished
Aug 30, 2010
212
0
18,680
The idea makes no sense neither from a business PoV or a customer PoV. Why would you in principle cripple the performance of your product to retain a lower performance threshold longer? You'll have to have really crappy product quality if you calculate in that a significant number of the cores fail within the normal 3-5 year life cycle to warrant this patent. Instead it would make much more sense to have all cores anabled from the start to have maximum performance, and the as cores fail it would read out as a indicator for when to exchange them instead.
 

nottheking

Distinguished
Jan 5, 2006
1,456
0
19,310
I can see some use of this (it sounds as if it could re-balance load before cores fail, simply to prevent individual cores from stay hot too long) but I can't help but think... Wasn't this sort of design perhaps already patented for use in spacecraft? Their designs tend to make use of redundant CPUs, and seem to use a method not entirely unlike what Intel described here.
 

saturnus

Distinguished
Aug 30, 2010
212
0
18,680
[citation][nom]nottheking[/nom]I can see some use of this (it sounds as if it could re-balance load before cores fail, simply to prevent individual cores from stay hot too long) but I can't help but think... Wasn't this sort of design perhaps already patented for use in spacecraft? Their designs tend to make use of redundant CPUs, and seem to use a method not entirely unlike what Intel described here.[/citation]

Standard fail-over systems used on spacecraft, and by every single serious server provider in the world, are completely seperate systems that steps in if one system fails. It's completely unrelated to this patent both in spirit and in practice, as it is much safer to switch to a different system all-together instead of relying on fail-over on the same silicon that is failing in the first place. No serious business would ever rely on that.
 

Honis

Distinguished
Mar 16, 2009
702
0
18,980
I don't think we'll see CPUs based on this at a consumer level for a long time. Redundancy is far more important in a server environment than it is on a home desktop. Having a fallback core on the CPU will (in theory) give the IT guy time to mitigate load off a server that's fallen to the fail safe and in the end have 0 downtime or even a slow down in available services.
 
This is exactly the type of article that makes me wonder about AMD fans. They talk about bulldozer as if the only reason that it is not great is because it is ahead of it's time, and that the architecture is the way of the future. Meanwhile Intel is already working with 64+ core setups and laying the groundwork for many-core products while still selling stuff that works well today!
 

infernolink

Distinguished
Apr 20, 2010
16
0
18,520
[citation][nom]outlw6669[/nom]This is a terrible idea.Why not ship the CPU with all cores active and give it a 'soft fail' feature for failing cores?By 'soft fail' I mean that a failing core could be dynamically deactivated while allowing all other cores to function normally.This would allow you to have higher initial performance and give you uninterrupted computing in the case of a core failure.[/citation]

This would mean that it would only got hotter faster, and possibly heat up all the cores to a point of deactivation. At that point you would technically have no CPU anymore.
 
G

Guest

Guest
I am sure that Sir Clive Sinclair proposed something similar to this for 16bit processors and memeory... many moons ago (20 or so years ago)...
 
[citation][nom]infernolink[/nom]This would mean that it would only got hotter faster, and possibly heat up all the cores to a point of deactivation. At that point you would technically have no CPU anymore.[/citation]
Not really sure how :??:
As long as you have a properly designed cooling system, you should have no issues at all.
When a defective core is disabled (presumably by power gating the affected area of the chip) overall power consumption (and therefore heat production) will be reduced.
This will lead to a cooler system as it ages, not a hotter one...
 

theuniquegamer

Distinguished
Sep 7, 2011
279
0
18,790
Looks like intel has its spare wheels (cores) for its superfast muscle cars (upcoming 32 core 14nm haswell or broadwell cpus). Nice idea to patent before amd or arm or (ipeoples).
 

memadmax

Distinguished
Mar 25, 2011
2,492
0
19,960
The person that talks about this regarding space based processors hit it right on the money.

Intel is talking about a processor that is super mission critical. Think space, nuclear, or some other place that man can't go to repair a computer. It sounds to me that Intel may be gearing up to knock out IBM/Motorola in the industrial compute department.
 

infernolink

Distinguished
Apr 20, 2010
16
0
18,520
[citation][nom]outlw6669[/nom]Not really sure how As long as you have a properly designed cooling system, you should have no issues at all.When a defective core is disabled (presumably by power gating the affected area of the chip) overall power consumption (and therefore heat production) will be reduced.This will lead to a cooler system as it ages, not a hotter one...[/citation]

I see it this way, if you have all your cores running at 100% they all will get hot quick. Since they all will have a similar temperature they may all reach the critical temperature at one point, which would signal them to shut off. Since Intel is working on tri-gate transistors I believe they mean they will stack cores on each other, thus making it harder for even the best coolers to cool. With this method of disabling cores and having inactive ones they can better compensate for that fact.
 

alidan

Splendid
Aug 5, 2009
5,303
0
25,780
[citation][nom]saturnus[/nom]The idea makes no sense neither from a business PoV or a customer PoV. Why would you in principle cripple the performance of your product to retain a lower performance threshold longer? You'll have to have really crappy product quality if you calculate in that a significant number of the cores fail within the normal 3-5 year life cycle to warrant this patent. Instead it would make much more sense to have all cores anabled from the start to have maximum performance, and the as cores fail it would read out as a indicator for when to exchange them instead.[/citation]

correct me if im wrong but in labs, doesn't intel already have a 14nm chip? thats and actual product, not just the on paper planing.

im going to give them the benefit of the doubt that they know what they are doing, though at the same time, i cant see a reason to lock cores, at least in a desktop, were power isn't an issue.
 

kronos_cornelius

Distinguished
Nov 4, 2009
365
1
18,780
Just like branch prediction, this makes a simple problem more complicated. Just make all the cores available with the ability to accept core failures.
This, like DRM, is intended to keep a lid of the availability of high power processing units to control prices rather than a solution to a problem that can lead to the betterment of society.
 
I see no reason not to do this. Think about it. If ALL cores did run all the time, the TDP would shoot through the roof. So if you have lets say 6 cores and 2 spares. The 2 spares are gated so not heating or drawing power. one of the other 6 fails(or overheats), and one of the 2 extras step in(while the defective core goes offline) to carry on working.

I like fail safe. Its not as fail safe as another entire system(those will cover anything failing, not just a cpu core) to swap over too, but seems like a decent idea and even a good temp solution while the secondary systems take over.
 

alidan

Splendid
Aug 5, 2009
5,303
0
25,780
[citation][nom]CaedenV[/nom]This is exactly the type of article that makes me wonder about AMD fans. They talk about bulldozer as if the only reason that it is not great is because it is ahead of it's time, and that the architecture is the way of the future. Meanwhile Intel is already working with 64+ core setups and laying the groundwork for many-core products while still selling stuff that works well today![/citation]

im betting it went something like this, intel was first to market with their threading solution, and it failed pretty bad, and took till the i7 to even be relevant over more cores.

amd saw that more cores = better performance, and went an opposite threading solution to intels, that on paper gives better over all performance, and they are having the same teething pains that intel had with their first ht chips.

now correct me if im wrong, but intels solution uses less die space than amds, but amds should be able to function as separate cores, unlike intels.

what im expecting is that in 1 revision, amd will solve most of the problems, and in 2 they will have single core performance more or less worked out, it may not be 100% the sb or phenom level, but close enough that it doesn't matter, and hopefully by than the applications that are still single thread that are commonly used will move to mutithread.

intels X cores on a chip projects are little more than real world experiments to better understand cpu architectures, and will most likely never make it to market in any current form, and some of them are just for the prestige
 
Status
Not open for further replies.