Intel plans to deliberately limit Sandy Bridge overclocking

[Kej]

LTE? well, the specs are not even finished

SandyB + Wimax = Wroooooooooooooooooom ;-)

Does it matter.....? It Works!

[Kej]

well, in Tokyo, there is no problem, neither in USA ... usually when you lose signal with Wimax, you will lose it with LTE too, because it is usually the steel bars in the concrete that are doing a F-cage and block the signal ...

A lot of people speak about LTE ... Where is it ? The issue of Wimax deployment is even worst for LTE ... There is just no LTE network. there are only few little points world wide ... and Wimax is open on the patent side, not hostage of Quadcomm patents ... just saying ... ;-) (This is my personal opinion)


Francois

If the steel bars will block the signal or not are heavily dependent on the
wavelength and length between the bars. Higher frequencies will go through.
The type of concrete and also how much water the concrete has, have impact
on signal.

IIRC Rule of thumb: Half the wavelength, or more, in space between the steel
bars should let the signal trough.

There is no WIMAX network! You can say things from your view, and me from
mine, but that view does not necessarily exist all over the world.

Here in Sweden we have LTE but no WIMAX. Intel bought the WIMAX
frequencies, but at least I haven't heard a word on any development.
Perhaps they run it for them selfs in Kista

The LTE network is rolling in 4 cities in Sweden now with 200 more to follow
during 2011. FWIW, YMMW etc etc.

[BeepBeep2]

well, in Tokyo, there is no problem, neither in USA ... usually when you lose signal with Wimax, you will lose it with LTE too, because it is usually the steel bars in the concrete that are doing a F-cage and block the signal ...

A lot of people speak about LTE ... Where is it ? The issue of Wimax deployment is even worst for LTE ... There is just no LTE network. there are only few little points world wide ... and Wimax is open on the patent side, not hostage of Quadcomm patents ... just saying ... ;-) (This is my personal opinion)


Francois

Our school can't have school wide student WiFi installed due to the walls and floor construction blocking the signal. Have a Wireless N router in one room, then 5 ft away in the next room and the signal is poor at best.

We are spending $20,000 to install access points and routers throughout the school...

[saaya]

well, in Tokyo, there is no problem, neither in USA ...

thats not what ive heard... but ok...

usually when you lose signal with Wimax, you will lose it with LTE too, because it is usually the steel bars in the concrete that are doing a F-cage and block the signal ...

thats not what ive heard either... service providers and chip manufacturers are mostly in favor of LTE here and say its easier to deploy and upgrade to from 3g and 3.5g than wimax... and according to what they said wimax has more problems with signal strength because of the frequencies it uses...

then again, as fqar as i know there are 3 frequency ranges for wimax which are pretty far apart...

A lot of people speak about LTE ... Where is it ?

oh come on, as if wimax was widely available... :P
nothing in LA
nothing in SF
nothing in SJ
nothing in miami
nothing in orlando
nothing in tampa
nothing in NY

The issue of Wimax deployment is even worst for LTE ... There is just no LTE network. there are only few little points world wide ...

from what ive heard its much easier to switch from 3g and 3.5g to LTE than wimax... no idea why or if thats true... but i dont see why theyd make this up...

and Wimax is open on the patent side, not hostage of Quadcomm patents ... just saying ... ;-) (This is my personal opinion)
Francois

true, yet chip companies here in taiwan still seem to prefer LTE over wimax...

[STEvil]

Our school can't have school wide student WiFi installed due to the walls and floor construction blocking the signal. Have a Wireless N router in one room, then 5 ft away in the next room and the signal is poor at best.

We are spending $20,000 to install access points and routers throughout the school...

On the note of wireless signals, this has been talked about a lot here recently:

http://www.muchmormagazine.com/2010/...ents-think-so/

What the article (and most others) dont mention is the "turn-on" date of the wireless networking, which was some time late in 2009. The wireless systems were installed for 4 years but were not turned on. Oh, and the transmitters/repeaters are about 20x more powerful than an average wireless router as well..

They should just use powerline networking technology given that you need to plug your laptop in to use it in a school setting anyways really.. sure it will last for a class or maybe three, but its not going to last all day..

[saaya]

OT: BB2

if the walls inside the school are blocking the signal, why dont you guys set up wifi outside, then all youd have to do is either rework the facade walls or add a repeater next to a central window in every room? but in my experience wifi goes through windows pretty well... so no matter how bad the walls are, the windows should be enough, especially since windows in schools tend to be rather large?

i did something like that when i still lived in germany, we wanted wifi in the backyard of our apartment building and noticed that by setting it up outside of the building, the wifi signal strength improved so much that a few neighbours even canceled their isp contracts and shared a single connection instead

[Mungri]

If this does happen, then I'll just get an I7 970 after its price comes down. Dont really need anything better than that.

(Just wanting a 32 nm to clock higher than my I7 920).

[Chumbucket843]

wow ... I did not confirm, neither deny ... this is intel PR job ... What I am saying is that when computers become more complex, the overclocking become a little harder, the design much harder ... That's all I said.
Don't make me say something I did not

when you say "computers become more complex" it just seems that some secrets are hidden behind those words. are the limited OC capabilities are from intel's incompetence to provide a decent clock architecture? this is my best explanation if complexity or some other technical reason is the problem. surely your engineers should be able to understand another layer of complexity? i mean a 22nm process has over 1,000 steps and you are able to design chips with billions of transistors. why does this limit exist?
/rant

[JumpingJack]

when you say "computers become more complex" it just seems that some secrets are hidden behind those words. are the limited OC capabilities are from intel's incompetence to provide a decent clock architecture? this is my best explanation if complexity or some other technical reason is the problem. surely your engineers should be able to understand another layer of complexity? i mean a 22nm process has over 1,000 steps and you are able to design chips with billions of transistors. why does this limit exist?
/rant

The oc'ability limit is no doubt due to increasing complexity and the drive for higher integration. Moving the GPU, PCIe, and most of the northbridge components into the CPU socket it becomes more practical to derive all those buses and components from a common clock.

The real OCability limiter here is not the clock generator or the CPU for that matter but the tight tolerance (or intolerance if you wish) of the PCIe bus, since that now derives its clock from the same that drives the CPU and others, it would not take Einstein to predict that OC will be limited.

[Boogerlad]

yeah, it might be at 100mhz bclck to sync with pci-e bus.

[Booj]

Wouldn't there be a pcie multiplier? just drop the pcie multi as bclk increases. Sounds easy enough.. Maybe thats the extra level of complexity being talked about heh... Getting ram, uncore, qpi, cpu and pcie to all play nice

[JumpingJack]

Wouldn't there be a pcie multiplier? just drop the pcie multi as bclk increases. Sounds easy enough.. Maybe thats the extra level of complexity being talked about heh... Getting ram, uncore, qpi, cpu and pcie to all play nice

That is where it is locked I would suspect, I have never seen a PCIe ratio/multiplier option in a BIOS, however i would suppose it possible or existed on boards I never owned. Typically, the PCIe on the chipset has it's own clock, separate from the system clock. Clarksdale/Allendale have the PCIe on the CPU but also has it's own clock -- supplied by the board.

yes, that is the level of complexity, the tight integration of all the different IOs. Sux for the enthusiast, but great for the volume OEMs and system builders -- two less specifications to design two (two different clock generators and the associated traces).

[zanzabar]

pci-e2 has a multi that is auto selected, so if u have no pci-e1 devices and run someware from 100-199 u would be below the range, that may be a bypass but i have no idea on the stability or how it would effect the chipset

[saaya]

That is where it is locked I would suspect, I have never seen a PCIe ratio/multiplier option in a BIOS, however i would suppose it possible or existed on boards I never owned.

yes, that is the level of complexity, the tight integration of all the different IOs.

then how can you defend intel for breaking bclock adjustments?
chum made a perfect point, francois is bragging about how great their designs are, but if they are really so great then how come they couldnt solve this problem by adding a simple divider or multiplier? nobody cares where the clock signal comes from, thats not the point... the point is what you do with it...

Sux for the enthusiast, but great for the volume OEMs and system builders -- two less specifications to design two (two different clock generators and the associated traces).

well good thing for us here at xtremeOEMsandsystembuilders.org then! hooray! wait...

[JumpingJack]

then how can you defend intel for breaking bclock adjustments?
chum made a perfect point, francois is bragging about how great their designs are, but if they are really so great then how come they couldnt solve this problem by adding a simple divider or multiplier? nobody cares where the clock signal comes from, thats not the point... the point is what you do with it...

well good thing for us here at xtremeOEMsandsystembuilders.org then! hooray! wait...

I don't defend it, i think it sux as much as the next guy but I am also not so closed minded I pretend not to understand why companies do the things they do.

Intel does not make their money with xtremesystems.org, they do not even make the most volume of their revenue with DIY boxed processors but they do make money with the guys that would be part of xtremeOEMsandsystembuilders.org that is who they are trying to sell the bulk of their product to in the first place ... they are offering unlocked skus which is fine by me. I won't participate in skt 1155 any way, I will only go for skt 2011.... those won't be BLCLK locked from what is understood on the net.

[zanzabar]

I don't defend it, i think it sux as much as the next guy but I am also not so closed minded I pretend not to understand why companies do the things they do.

Intel does not make their money with xtremesystems.org, they do not even make the most volume of their revenue with DIY boxed processors but they do make money with the guys that would be part of xtremeOEMsandsystembuilders.org that is who they are trying to sell the bulk of their product to in the first place ... they are offering unlocked skus which is fine by me. I won't participate in skt 1155 any way, I will only go for skt 2011.... those won't be BLCLK locked from what is understood on the net.

but u cannot sell chips without people like those of XS liking them, just look at the p4 it lost market share since diy and overclockers did not want it so then tuned in consumer grade OEM buying consumers started to want amd, then the core2 and am2 came out destroying amd in the DIY and intel went back up but amd still kept some people. if this happened again with amd already in the 20% or so and most companies having amd builds then intel would loose a much larger chunk. so they would need an unlocked bclock platform or cheap unlocked cpus or they will suffer in the long term.

[richierich]

but u cannot sell chips without people like those of XS liking them, just look at the p4 it lost market share since diy and overclockers did not want it so then tuned in consumer grade OEM buying consumers started to want amd, then the core2 and am2 came out destroying amd in the DIY and intel went back up but amd still kept some people. if this happened again with amd already in the 20% or so and most companies having amd builds then intel would loose a much larger chunk. so they would need an unlocked bclock platform or cheap unlocked cpus or they will suffer in the long term.

Well let the market speak for itself. Personally I think it's more rewarding tweaking a locked multiplier cpu than an unlocked one (although a lot more patience is needed ). Let them have fun with their K-series CPUs. Had it not been for AMD's Black Edition processors, Intel may have never even released the K-series.

[zanzabar]

Well let the market speak for itself. Personally I think it's more rewarding tweaking a locked multiplier cpu than an unlocked one (although a lot more patience is needed ). Let them have fun with their K-series CPUs. Had it not been for AMD's Black Edition processors, Intel may have never even released the K-series.

that was if the buss is locked, they need unlocked multi parts cheap

[ajaidev]

Slide sorry had to edit...





Another slide i don't know what the numbers mean

[Chumbucket843]

The oc'ability limit is no doubt due to increasing complexity and the drive for higher integration. Moving the GPU, PCIe, and most of the northbridge components into the CPU socket it becomes more practical to derive all those buses and components from a common clock.

The real OCability limiter here is not the clock generator or the CPU for that matter but the tight tolerance (or intolerance if you wish) of the PCIe bus, since that now derives its clock from the same that drives the CPU and others, it would not take Einstein to predict that OC will be limited.

why cant a PLL fix that? nehalem is a highly modular design which is why every core, mem controller, and the uncore have their own PLL. everything is decoupled to provide flexibility.

it doesnt take an einstein to fix this problem. aside from that einstein would more likely avoid problems in the first place rather than fix problems.

[FlanK3r]

interesting Intels message

[JumpingJack]

why cant a PLL fix that? nehalem is a highly modular design which is why every core, mem controller, and the uncore have their own PLL. everything is decoupled to provide flexibility.

it doesnt take an einstein to fix this problem. aside from that einstein would more likely avoid problems in the first place rather than fix problems.

A PLL for the PCIe bus is certainly the right way to do it, I just have never run across a board the provides a multiplier (or adjustable PLL) for the PCIe bus, they usually have their own clock.

As you say, it is not rocket science. Frankly, I think Intel is making a huge mistake ... I am no different in my opinion than anybody else. While I can undestand the reasons behind the design decision, I do not agree with them.

The budget overclocking enthusaist is a very small part of the market, even in the most generous assumptions, maybe 5-6% of all CPUs sold, what I don't think Intel really comprehends is that that 5-6% is the most demanding, most vocal. What they tell their friends, family, and neigbors influences those decisions as well. This move, undoubtedly, will cost them the any mindshare they have in the enthusast community and it will have farther reaching repercussions than just losing a few boxed processor sales.

No amount of 'K' CPUs is going to alter that perception.

EDIT: then there is the perception, which most certainly could be true, that Intel is trying to monetize overclocking. With the integration of the GPU in Clarksdale/Allendale and the release of 1366 sockets, Intel is clearly putting a hard, and definitive boundary between the low end/mainstream CPUs and the high end/performance CPUs derived without an IGP integrated. If I were deciding this, I would put out budget, low costs socket 2011 chips and chipsets and eliminate any enthusiast interest in socket 1155. That's just me though.

[JumpingJack]

but u cannot sell chips without people like those of XS liking them, just look at the p4 it lost market share since diy and overclockers did not want it so then tuned in consumer grade OEM buying consumers started to want amd, then the core2 and am2 came out destroying amd in the DIY and intel went back up but amd still kept some people. if this happened again with amd already in the 20% or so and most companies having amd builds then intel would loose a much larger chunk. so they would need an unlocked bclock platform or cheap unlocked cpus or they will suffer in the long term.

Did they lose P4 marketshare because the DIY market did not want them or did they lose market share because their competitor simply had a better product?

It is short-sighted to make that correlation of cause and effect, there is no evidence that your thesis is actually true.

[JumpingJack]

Slide sorry had to edit...

Are these really Intel slides? Were they translated from different language, because the grammar is worse than mine ...

[villa1n]

On the note of wireless signals, this has been talked about a lot here recently:

http://www.muchmormagazine.com/2010/...ents-think-so/

What the article (and most others) dont mention is the "turn-on" date of the wireless networking, which was some time late in 2009. The wireless systems were installed for 4 years but were not turned on. Oh, and the transmitters/repeaters are about 20x more powerful than an average wireless router as well..

They should just use powerline networking technology given that you need to plug your laptop in to use it in a school setting anyways really.. sure it will last for a class or maybe three, but its not going to last all day..

I have a hard time believing this.. cellphone directly to the head, yes.. that makes sense

But ambient 2.4ghz at what... 50 milli watts? Cathode tv's emitted much more radiation, and directly at you, than a wifi router could, or even your wiring in your house, unless its fully insulated (usually not) is going to create more background radiation that a wifi router.. Do they have a proposed mechanism, or taken any EEG, fmri's?